In re Google Dig. Advert. Antitrust Litig.

Full title: IN RE: GOOGLE DIGITAL ADVERTISING ANTITRUST LITIGATION THIS DOCUMENT…

Court: United States District Court, S.D. New York

Date published: Sep 13, 2022

627 F. Supp. 3d 346 (S.D.N.Y. 2022)

Opinion

21-md-3010 (PKC) 21-cv-6841 (PKC)

2022-09-13

IN RE: GOOGLE DIGITAL ADVERTISING ANTITRUST LITIGATION. This Document Relates to: State of Texas, By Attorney General Ken Paxton, State of Alaska, By Attorney General Treg R. Taylor, State of Arkansas, By Attorney General Leslie Rutledge, State of Florida, By Attorney General Ashley Moody, State of Idaho, By Attorney General Lawrence G. Wasden, State of Indiana, By Attorney General Todd Rokita, Commonwealth of Kentucky, By Attorney General Daniel Cameron, State of Louisiana, By Attorney General Jeff Landry, State of Mississippi, By Attorney General Lynn Fitch, State of Missouri, By Attorney General Eric Schmitt, State of Montana, By Attorney General Austin Knudsen, State of Nevada, By Attorney General Aaron D. Ford, State of North Dakota, By Attorney General Drew H. Wrigley, Commonwealth of Puerto Rico, By Attorney General Domingo Emanuelli-Hernández, State of South Carolina, By Attorney General Alan Wilson, State of South Dakota, By Attorney General Jason R. Ravnsborg and State of Utah, By Attorney General Sean D. Reyes, Plaintiffs, v. Google LLC, Defendant.

CASTEL, Senior District Judge

OPINION AND ORDER

CASTEL, Senior District Judge:

I. OVERVIEW OF THE BUYING AND SELLING OF DISPLAY AND IN-APP ADS 360

II. PRODUCT AND GEOGRAPHIC MARKETS AND MARKET POWER 361

A. Publisher Ad Servers 361

B. Ad Exchanges 363

C. Ad-Buying Tools for Large Advertisers 364

D. Ad-Buying Tools for Small Advertisers 365

E. In-App Mediation Tools 365

F. In-App Networks 365

III. PLEADING STANDARD FOR THE SHERMAN ACT CLAIMS 366

IV. COUNT III OF THE COMPLAINT PLAUSIBLY ALLEGES A SECTION 1 TYING CLAIM 367

V. COUNT IV DOES NOT PLAUSIBLY ALLEGE A SECTION 1 CLAIM BASED ON GOOGLE'S AGREEMENTS WITH FACEBOOK 370

A. The States Have Not Plausibly Alleged an Unlawful Agreement Between Google and Facebook to Restrain Facebook's Use of Header Bidding 370

1. Header Bidding 371

2. The Complaint Does Not Plausibly Allege Collusion Between Google and Facebook to Thwart Header Bidding 372

B. The States Have Failed to Plausibly Allege an Agreement between Google and Facebook to Limit Competitive Bidding for In-App Ad Inventory 373

C. The Alleged Restraint on Bidding for In-App Impressions Is Properly Scrutinized Under the Rule of Reason 374

D. The States Have Failed to Plausibly Allege a Restraint on Bidding for In-App Impressions under the Rule of Reason 375

VI. CERTAIN OF THE STATES' ALLEGATIONS PLAUSIBLY DESCRIBE ANTICOMPETIVE CONDUCT AND STATE CLAIMS FOR MONOPOLIZATION AND ATTEMPTED MONOPOLIZATION UNDER SECTION 2 378

A. Monopolization 378

B. Attempt to Monopolize 380

C. Monopoly Broth 381

D. The Alleged Anticompetitive Conduct Supporting the Monopolization and Attempt to Monopolize Claims 381 1. Google's Use of Encrypted User IDs Is Not Plausibly Alleged to be Anticompetitive Conduct 381

2. The Complaint Plausibly Alleges Google's Use of Dynamic Allocation Was Anticompetitive Conduct in the Ad Exchange Market 384

3. The Complaint Plausibly Alleges that Google's Use of Enhanced Dynamic Allocation Was Anticompetitive Conduct in the Ad Exchange Market 386

4. The Complaint Plausibly Alleges that Project Bernanke Was Anticompetitive in the Market for Ad-Buying Tools for Small Advertisers and the Bell Variation Was Anticompetitive in the Ad-Server and Ad-Exchange Markets 387

5. The Complaint Plausibly Alleges that Dynamic Revenue Sharing Was Anticompetitive Conduct that Harmed Competition in the Ad-Exchange Market 389

6. The Complaint Does Not Plausibly Allege that Reserve Price Optimization Was Anticompetitive Conduct 391

7. The Complaint Does Not Plausibly Allege that the Challenged Aspects of Exchange Bidding Were Anticompetitive in Any Market 393

8. The Complaint Plausibly Alleges that Google's Redaction of Auction Data and Limitations on Publisher Line items Was Anticompetitive Conduct in the Exchange Market and Ad Server Market 395

9. The Complaint Plausibly Alleges that Projects Poirot and Elmo Were Anticompetitive Actions in the Ad-Exchange Market and the Market for Ad-Buying Tools of Large Advertisers 397

10. The Complaint Does Not Plausibly Allege Anticompetitive Conduct Relating to Mobile Web Page Development 398

11. The Claim Directed to Google's Proposed Privacy Sandbox Is Not Ripe for Adjudication 399

12. The Complaint Plausibly Alleges that Google's Unified Pricing Policy Was Anticompetitive Conduct Directed to the Ad-Exchange Market and Ad-Buying Tools for Small and Large Publishers 400

13. The Facts Underlying the Section 1 Tying Claim Are Anticompetitive Conduct in the Publisher Ad Server Market in Support of the Section 2 Claims 402

VII. THE COMPLAINT DOES NOT PLAUSIBLY ALLEGE THAT DYNAMIC ALLOCATION AND DRS HAVE CONTINUING, PRESENT ADVERSE EFFECTS, AND THIS CONDUCT CANNOT BE ENJOINED 402

VIII. THE COURT DECLINES TO ADJUDICATE GOGGLE'S LACHES DEFENSE AT THE PLEADING STAGE 404

CONCLUSION 407

The advertising industry has kept pace with consumers' near-universal use of websites and mobile apps to obtain news and information. Publishers and advertisers can now participate in a milliseconds-long auction to sell an ad directed to a specific web user based on browsing history and characteristics. Pricing varies based on the consumer's perceived value to the particular advertiser: a seller of motorcycles or sunglasses is generally willing to pay more for ads targeted to likely purchasers. This antitrust action focuses on the multiple roles played by Google LLC ("Google") in the purchase and sale of display ads on commercial websites and ad impressions on mobile apps.

The Attorneys General of ten states brought an action in the Eastern District of Texas against Google, alleging that Google's

digital advertising practices violate sections 1 and 2 of the Sherman Act, as well as the laws of their states. The action was transferred to this Court by the Judicial Panel on Multi-District Litigation for coordinated pre-trial proceedings. Since then, a 702-paragraph Third Amended Complaint ("the Complaint") has been filed in this District on behalf of sixteen states and the Commonwealth of Puerto Rico (collectively, the "States").

The States describe the Complaint as cataloguing a "sweeping variety of anticompetitive conduct." (Mem. in Opp. at 1.)¹ They allege that Google has monopolized or attempted to monopolize various markets related to online display ads (Counts I and II) and unlawfully used its market power to tie the sale of Google's "ad server," a tool used by publishers to manage their inventory of display ads, to Google's "ad exchange," a distinct product that conducts auctions for the sale of display ads (Count III).² They also allege that Google entered into an unlawful restraint of trade with non-parties Facebook, Inc. and Facebook Ireland Limited ("Facebook") (Count IV).

1 The operative pleading also alleges violations of various state statutes. At a pretrial conference of September 21, 2021, this Court stayed the filing of any motions directed to the several state law claims of the Attorneys General, thereby permitting the parties and the Court to focus on the federal antitrust claims. See Pre-Trial Order No. 1 ¶ 11 (Aug 13, 2021; Docket # 4).

2 The action does not relate to other forms of advertising on the internet, including targeted text-based ads sold by search engines, video ads that run before or during video content or sharable ads on social media platforms.

The States seek only injunctive relief for the claimed Sherman Act violations and bring this action as parens patriae on behalf of their citizens. The Complaint alleges some anticompetitive conduct undertaken by Google in the past that it has since abandoned. Such conduct may be relevant to Google's motive and intent, but marketplace conduct that is no longer practiced generally may not be the subject of injunctive relief unless it has a continuing, present adverse effect.³

3 See Zenith Radio Corp. v. Hazeltine Research, Inc., 395 U.S. 100, 130, 89 S.Ct. 1562, 23 L.Ed.2d 129 (1969) (plaintiff must "demonstrate a significant threat of injury from an impending violation of the antitrust laws or from a contemporary violation likely to continue or recur."). Google separately asserts that by reason of unreasonable delay and prejudice, certain claims are barred by laches.

Google moves to dismiss the federal antitrust claims for failure to state a claim for relief. Rule 12(b)(6), Fed. R. Civ. P. Google need not feature all of its defenses in its motion, and its motion accepts the Complaint's product and geographic market definitions, and, with limited exceptions, its allegations of monopoly power or market power. The motion focuses instead on Google's assertion that its marketplace conduct has been lawful and innovative and has provided consumers with meaningful choices.

On a motion to dismiss, the Court accepts the non-conclusory allegations of the Complaint as true and determines whether they plausibly state claims for relief. As will be seen, the Supreme Court has emphasized the importance of this gatekeeping role in the antitrust arena. The Court must exercise this role even though a motion to dismiss neither allows for a factual narrative that supplements the four corners of the pleading nor a counter-narrative by the defendant. Consideration of actual evidence must await the completion of discovery and a motion for summary judgment or trial. Experience teaches that

cases often look very different when evidence from both sides is considered.

On this motion, the Court principally concludes that:

• The States have plausibly alleged that Google has used its market power in the ad-exchange market to coerce publishers to license its publisher ad server and thus stated a claim for an unlawful tying arrangement in violation of section 2 (Count III).

• The States have not plausibly alleged Google's Network Bidding Agreement with Facebook amounts to an unlawful restraint of trade in violation of section 1 (Count IV).

• The States have plausibly alleged a monopolization claim under section 2 in the nationwide markets for (1) publisher ad servers, (2) ad exchanges and (3) ad-buying tools for small advertisers (Count I).

• The States have plausibly alleged an attempt-to-monopolize claim under section 2 in the nationwide market for ad buying tools for large advertisers and an alternative claim for attempt to monopolize the markets for ad exchanges and ad-buying tools for small advertisers (Count II).

• Google has challenged the timeliness of many of the State's assertions of anticompetitive conduct under the doctrine of laches. The Court concludes that the consideration of this affirmative defense must await the development of a factual record.

Thus, the motion to dismiss will be granted as to Count IV and otherwise denied.

I. OVERVIEW OF THE BUYING AND SELLING OF DISPLAY AND IN-APP ADS.

An online content publisher, such as a news website, sells advertising space through ad exchanges. These ad exchanges run automated auctions of ad impressions, in which competing advertisers submit bids based on the characteristics of the person who will view the ad. The auction occurs in a fraction of a second, taking place as the user's webpage loads and displays the ad of the successful bidder. The Google ad exchange, called AdX, processes approximately eleven billion display ads each day. (Compl't ¶ 5.) In addition to operating an ad exchange, Google also offers a tool for managing a publisher's sale of online display ads and tools for advertisers to purchase display-ad space. Google also offers products and services to developers of mobile-device apps, which seek to sell ad impressions, and to the in-app networks that purchase those impressions.

The Court begins with an overview of how publishers and app developers sell their advertising inventory. Much of the terminology and jargon will be important to an understanding of the States' antitrust claims. For ease of reference, boldface text will be used when a new term is introduced.

Large publishers sell display ads directly to advertisers, but they also sell them indirectly through ad exchanges, which conduct automated auctions of publisher inventory. Large publishers manage their inventory of display ads — also known as impressions — through a type of software known as an ad server. The ad server interfaces on behalf of the publisher with an ad exchange. Advertisers use an ad-buying tool to bid on display ads. The ad-buying tool used by large, sophisticated advertisers has distinct features from those used by small advertisers. Ad-buying tools interface with ad exchanges on behalf of advertisers.

The Complaint neatly summarizes how these components interact to conduct an auction on an ad exchange:

When a user [i.e. consumer] visits a publisher's website, the publisher's ad server sends a "bid request" to the ad buying tools who have a "seat" to bid in the exchange and purchase on behalf of their advertiser clients. This bid request announces the publisher's available impressions to exchanges, along with information about the impression, including the user's ID, the ad slot's parameters, and any rules about pricing. These bid requests also contain information about the impression at issue and convey a "timeout," which is the amount of time prospective buyers are allotted to respond with their "bid response." Within this timeframe, which is typically a mere fraction of a second, each ad buying tool must unpack the information contained in the bid request, gather and deploy personal information about the user, determine the appropriate price to bid on behalf of the prospective advertiser, and return a bid response to the exchange. When time expires, each exchange closes its auction, excludes any late bids, and passes its highest bid to the ad server. The publisher's ad server then selects which ad to display and effectuates the display of the ad to the user.

(Id. ¶ 74.)

The marketplace for the sale of ads by developers of apps used in mobile devices such as smartphones and tablets is somewhat different. Developers use an in-app mediation tool that (1) manages the developer's inventory of impressions; (2) includes a software development kit installed on a user's device that enables the developer to obtain information about the user; and (3) serves as the vehicle for conducting auctions. Advertisers do not typically interact directly with the in-app mediation tool. Instead, in-app networks buy impressions in a manner akin to a wholesaler and then resell them at a mark-up to advertisers.

In the marketplace for display ads, Google offers ad servers for publishers and ad-buying tools for large and small advertisers. It also operates an ad exchange. In the in-app impression marketplace, Google offers an in-app mediation tool for developers and operates an in-app network for advertisers.

II. PRODUCT AND GEOGRAPHIC MARKETS AND MARKET POWER.

The States allege that Google has market power in six distinct product markets, each of which is alleged to be nationwide in geographic scope.⁴ For the purposes of the motion, Google has not challenged the existence of these product and geographic markets, the States' definitions of the markets or the allegations of monopoly or other market power, except in a footnote that challenges the existence of monopoly power in the ad-exchange market. (Google Mem. at 30 n.9.)

4 The Court recognize that Google's activities — e.g., licensing of software, conduct of auctions, furnishing of technical support — are services rather than products, but will use the terminology "product" market in conformity with the parties' usage and the practice utilized in most case law.

A. Publisher Ad Servers.

To manage their inventory of display ads, publishers license a software product called an ad server. The ad server is used for both direct and indirect sales of display ads. "Publishers typically use a single ad server to manage all of their web display inventory; using multiple ad servers would substantially frustrate a publisher's ability to effectively optimize management of their inventory and maximize revenue." (Compl't ¶ 49.) An ad server allocates and

routes available display ad space between direct sales per pre-arranged agreements with advertisers and indirect sales conducted through exchanges. The ad server directly connects to the ad exchange. (Id. ¶¶ 52-53.)

Ad servers also assign a unique ID to each user, which identifies users by various characteristics and is intended to match ads to their target viewers. As described by the States:

[W]hen a user visits a webpage, the ad server — on behalf of and with the permission of the publisher — identifies the user through identification technology facilitated by the user's web browser (e.g., Chrome or Safari) and/or mobile device (e.g., Android or iOS). To keep track of individual users, the ad server assigns each user a unique user ID (e.g., 5g77yuu3bjNH). By essentially 'tagging' users with a unique user ID, an ad server helps publishers, ad exchanges, and advertisers identify and track various characteristics and behaviors of each particular user who accesses the publisher's content. For example, an advertiser can correlate a user's pseudonymous ID (e.g., 5g77yuu3bjNH) with the user's identity (e.g., John Connor) and use that identity 'link' to look up additional information about the user (e.g., John Connor lives in Los Angeles, drives Harley-Davidson motorcycles, and wears Oakley sunglasses). This, in turn, allows a prospective ad purchaser (an advertiser or network) to place a value on the ad space each individual user will see.

(Id. ¶ 51.)

The States allege that ad servers for large publishers are a relevant product market and that Google has willfully acquired monopoly power in that market. (Id. ¶¶ 93, 528.) Large publishers have unique characteristics and customers, and ad servers are used by these publishers to manage a large volume of ad sales made through direct and the indirect sales channels. (Id. ¶ 94.)⁵ The States allege that for large publishers, there are no reasonable substitutes for ad servers, and that there are high barriers to entry for competing ad servers due to the high "switching costs" (risk and intensive use of internal resources) that publishers would encounter. (Id. ¶¶ 98, 126.)

5 According to the Complaint, "most" small publishers do not need an ad server. (Id. ¶ 97.) The Complaint does not demarcate small publishers from large ones. Small publishers often sell their ad inventory to a "web display ad network" or "ad network" which, in turn, sells that inventory to advertisers. (Id. ¶ 65.) Networks are middlemen holding inventory risk. (Id. ¶ 66.) Sales are not conducted through real-time auctions but are direct sales by the network to small advertisers. Google describes its ad network, known as the Google Display Network or GDN, as "the largest ad network in the world." (Id. ¶ 69.) Google's margin on network sales is typically 32-40% of each transaction. (Id.)

Google entered the ad server market in 2008 through its acquisition of DoubleClick. (Id. ¶ 245.) According to the States, "[w]hen Google urged the FTC to clear its acquisition of DoubleClick, it argued that several competing ad servers constrained its ability to increase price or decrease quality; these included WPP's 24/7 Real Media ad server, Microsoft's Atlas/aQuantive ad server, and ValueClick's ad server. All of those competitors have since exited the market." (Id. ¶ 122.)

At the time of the acquisition, DoubleClick's share of the ad-server market was between 48-57%. (Id. ¶ 245.) By 2010, Google's share of the ad-server market had grown to 78%, by 2012 to 85%, and by 2015 to 90%. (Id. ¶ 114.) The States allege that Google internal documents show that by

Q2 of 2018, Google's market share of large publishers had reached 99% in the United States. (Id.) Google's ad server for publishers has been known at various times as DoubleClick for Publishers or DFP, and Google Ad Manager or GAM. (Id. at ¶¶ 97, 100.)

B. Ad Exchanges.

Ad exchanges are "real-time auction marketplaces that match multiple buyers and multiple sellers on an impression-by-impression basis." (Compl't ¶ 58.) Ad exchanges are typically used by large publishers and have minimum-impression requirements. (Id. ¶ 59.) Ad exchanges do not hold an inventory of display ads but act as a go-between, and charge publishers a "take-rate" or exchange fee as a commission on the clearing price of the transaction. (Id. ¶ 60.) An ad exchange auctions a publisher's inventory, as routed through an ad server, and advertisers submit bids through an ad-buying tool. (Id. ¶ 58.)

The States allege that ad exchanges comprise a relevant product market and that Google has maintained or acquired monopoly power in the market for ad exchanges. (Id. ¶¶ 128, 528.) The ad-exchange market is the subject of the States' monopolization and attempted monopolization claims. (Id. ¶ 528, 532.) Ad exchanges have unique customers, features, pricing, and entry and usage requirements, and there are no reasonable product substitutes. (Id. ¶¶ 129, 131.) Exchanges facilitate real-time auctions and, as noted, do not bear inventory risk. (Id. ¶ 129.)

Google's ad exchange is known as AdX. (Id. ¶ 5.) "By 2015, Google's internal documents demonstrate that 80 percent of the publishers using Google's ad server also contracted with Google's exchange. Since 90 percent of publishers were using Google's ad server, this means that the large majority of available publisher customers were using Google's exchange ...." (Id. at ¶ 150.) In the four quarters preceding October 2019, AdX allegedly "transacted over 60 percent of all display inventory sold through exchanges in the United States." (Id. ¶ 151.)

The States acknowledge that three major exchanges compete with AdX: Rubicon, Xandr, and Index Exchange. (Id. ¶ 153.) While $7.6 billion in gross revenues was transacted on AdX in 2018, the next-largest exchange (Xandr) transacted $2 billion in gross revenues and all competitor exchanges transacted $6 billion combined. (Id.)⁶ The Complaint also alleges that AdX transacts impressions targeted to high-value users that advertisers cannot purchase in rival exchanges. (Id. ¶ 154.) The Complaint alleges that Google has monopoly power in the ad-exchange market that is shown not by market share alone but also by its ability to charge supracompetitive prices, with an average take rate of 20% of a transaction's value. (Id. ¶ 156.) According to the States, this is double or quadruple the rate charged by AdX's nearest competitors. (Id. ¶ 61.)

6 According to the Complaint, "[t]he exchange market is also characterized by market exit and lack of recent entry. Microsoft (AdECN) exited the exchange market in 2011, Yahoo! (RMX) in 2015, and Facebook (FBX) in 2016." (Id. ¶ 159.) A new entrant would have to achieve sufficient scale among both publishers and advertisers to be viable. (Id. ¶ 160.)

The States allege that Google's DFP ad server preferentially routes publisher inventory to AdX and that "Google operates the largest ad exchange in the market and maintains its monopoly position in ad serving, creating inherent conflicts of interest between publishers' best interests and its own. Google imposes one fee for its ad server to manage publishers' inventory

and then takes another (substantially higher) fee when that inventory trades through AdX." (Id. ¶ 64.)

In a footnote to its memorandum in support of the motion to dismiss, Google urges that the States have failed to allege that Google possessed monopoly power in the ad-exchange market, implying the existence of a bright-line rule that "market shares below 65 percent cannot support a Section 2 claim." (Google Mem. at 30 n.6.) In this Circuit, there is no bright-line rule. See PepsiCo, Inc. v. Coca-Cola Co., 315 F.3d 101, 109 (2d Cir. 2002) ("Absent additional evidence, such as an ability to control prices or exclude competition, a 64 percent market share is insufficient to infer monopoly power.") (emphasis added); Tops Markets, Inc. v. Quality Markets, Inc., 142 F.3d 90, 99 (2d Cir. 1998) (depending on other market factors, "a share between 50% and 70% can occasionally show monopoly power....") (quotation marks omitted); Broadway Delivery Corp. v. United Parcel Serv. of Am., Inc., 651 F.2d 122, 130 (2d Cir. 1981) (error to instruct a jury that monopoly power could not be found if the market share was less than 50%).

At the pleading stage, Google's market share and other relevant market characteristics outlined above permit a plausible inference of monopoly power in the ad-exchange market.

C. Ad-Buying Tools for Large Advertisers.

Large advertisers require ad-buying tools to implement their buying programs. The tools allow large advertisers to achieve advertising "campaign objectives, including purchasing the best quality inventory on exchanges for the lowest prices." (Compl't ¶ 196.) "Ad buying tools let advertisers set various parameters integral to their automated purchasing decisions, including crucial details about the types of users they want to target and the maximum bids they are willing to submit for various types of display ad inventory." (Id. ¶ 72.) Ad-buying tools connect to an ad exchange, which, in turn, is connected to publishers' ad servers, such as Google's DFP. Large advertisers may use an ad-buying tool across multiple exchanges and networks. (Id. ¶ 71.) An ad-buying tool for large advertisers is sometimes referred to as demand side platform or DSP. Google's DSP is called DV360, and is described as the "largest" ad-buying tool for large advertisers. (Id. ¶ 76.)

The Complaint alleges that DSPs used by large advertisers are a relevant product market for antitrust purposes and that Google has attempted to monopolize that market. (Id. ¶¶ 196, 532, 533.) It asserts that there are no suitable substitutes. (Id. ¶¶ 197-209.) The States cite GEICO, McDonalds and Ford as examples of large advertisers. (Id. ¶¶ 196, 198.) The monthly spend rates for ad-buying tools are very high, with Google's DV360 requiring at least a $10 million spend per year, Media Math requiring $2.4 million and The Trade Desk at least $1 million. (Id. ¶ 199.) Amazon's DSP requires a monthly commitment of $35,000. (Id. ¶ 73.)

The Complaint does not allege Google's market share in ad-buying tools for large advertisers, other than to say that DV360 "is the largest ad buying tool for large advertisers." (Id. ¶ 76.) The monopolization claim in Count I is not directed to Google's conduct in the DSP market, though the attempted monopolization claim in Count II asserts that "Google has monopoly power, or in the alternative, a dangerous probability of acquiring monopoly power, in the market ... for ad buying tools for large ... advertisers." (Id. ¶ 532.) Certain other allegations suggest that DV360 has tailored its activities based on the risk that

its customers would shift business to rival DSPs. (Id. ¶ 397.)

D. Ad-Buying Tools for Small Advertisers.

The States allege that the ad-buying tools used by small advertisers is a relevant product market in which Google has acquired or maintains monopoly power. (Id. ¶¶ 163, 528.). It is the subject of the States' claims of monopolization and attempted monopolization. (Id. ¶¶ 528, 532.) The tools are simple and easy to use, and appeal to individuals and small business that do not have the resources to master more complex buying tools. (Id. ¶ 169.) The Complaint is replete with allegations as to why there are no reasonable product substitutes for small advertisers. (Id. ¶¶ 170-179.)

A single advertiser (or its agency) typically uses only a single ad-buying tool. (Id. ¶ 71.) While competing ad-buying tools for small business have minimum spend requirements, Google Ads, its small-advertiser tool, does not. (Id. ¶ 166.) "In 2010, 600,000 small and medium size businesses in the United States used Google Ads," far more than any competing tool. (Id. ¶ 188.) Google Ads buys approximately 50% of all display ads sold on AdX, which is the largest ad exchange, and buys about 30% of all display ads on all exchanges. (Id. ¶ 190.) "In 2012, Google internally compared Google Ads to eight competitors; out of those eight competing buying tools, not even one still operates as a buying tool for small advertisers." (Id. ¶ 191.)

E. In-App Mediation Tools.

Ad impressions are also displayed on mobile apps. The party developing an app and selling ads on that app is referred to as a "developer." (Id. ¶ 79.) The required format of ads may vary across apps, and successful integration requires Software Development Kits or SDKs. (Id. ¶¶ 81, 235, 230.) "Developers typically use just one mediation tool for an app; using multiple mediation services would be exceedingly complex and frustrate the developer's ability to maximize ad revenue." (Id. ¶ 80.)

The States allege that in-app mediation tools are a relevant product market for antitrust purposes and that Google's mediation tools "dominate the market," but they do not allege that Google has monopoly power or the dangerous probability of acquiring monopoly power. (Id. ¶¶ 227-229, 528, 532.)

Google has two products in the market: Google Ad Manager for apps or GAM for apps, which it promotes as a product for large developers, and AdMob mediation as a product for other developers, and internally, Google treats them as two products in the same market. (Id. ¶ 227.) The States allege that by 2019, 50% of the apps in the United States selling in-app ads used one of Google's two products. (Id. ¶ 228.) When apps that do not use a mediation tool are taken out of the mix, Google's percentage of developer-users rises to 60%. (Id.) None of the three other competing in-app mediation tools (MoPub, a Twitter tool, ironSource, and AppLovin) exceeds 30% of the market. (Id.) The States allege that Google's power in the market is entrenched because of the high switching costs that a developer would incur in changing to a different mediation tool. (Id. ¶¶ 230-231.)

F. In-App Networks.

The States allege that in-app networks that purchase in-app display impressions from developers and sell them to advertisers are a separate product market; there is no allegation of Google's market share, market power or monopoly power in the in-app networks. (Id. ¶¶ 232, 528, 532.)

There are no reasonable substitutes for in-app networks. (Id. ¶¶ 233-236.)

Google's network is known as AdMob network and it competes with a number of products, including Facebook Audience Network or "FAN," Unity, ironSource, Vungle and others. (Id. ¶ 232.)

III. PLEADING STANDARD FOR THE SHERMAN ACT CLAIMS.

Google has moved to dismiss the four federal claims brought under the Sherman Act, 15 U.S.C. § 1, et seq. Count I alleges that Google has willfully acquired or maintained monopoly power through anticompetitive conduct in the markets for (1) publisher ad servers, (2) ad exchanges, and (3) ad-buying tools for small advertisers, in violation of section 2 of the Sherman Act, 15 U.S.C. § 2. Count II alleges that Google, through anticompetitive conduct, has monopoly power or a dangerous probability of acquiring monopoly power in the markets for (1) ad exchanges and (2) the markets for ad buying tools for large and small advertisers, also in violation of section 2. Count III alleges that Google has unlawfully tied its AdX exchange to its DFP publisher ad server, in that Google has monopoly power or, at a minimum, sufficient market power in the ad-exchange market to coerce publishers to license its ad server in violation of sections 1 and 2 of the Act. Finally, in Count IV, the States allege that the Network Bidding Agreement between Facebook and Google is a per se restraint of trade in violation of section 1. Alternatively, they argue that under a rule-of-reason analysis, the same agreement caused significant anticompetitive effects that outweigh any procompetitive benefits, also in violation of section 1.

The Supreme Court has addressed the "question of what a plaintiff must plead in order to state a claim under § 1 of the Sherman Act." Bell Atlantic Corp. v. Twombly, 550 U.S. 544, 554-54, 127 S.Ct. 1955, 167 L.Ed.2d 929 (2007). The Court concluded that while the complaint "does not need detailed factual allegations ... labels and conclusions, and a formulaic recitation of the elements of a cause of action will not do." Id. at 555, 127 S.Ct. 1955.⁷ "Factual allegations must be enough to raise a right to relief above the speculative level ...." Id. For example, an allegation of a conspiracy in restraint of trade requires "allegations plausibly suggesting (not merely consistent with) agreement ...." Id. at 557, 127 S.Ct. 1955.

7 The States employ much of the antitrust lexicon and other colorful prose in the body of their Complaint, e.g., "anticompetitive acts" (Compl't ¶ 7), "substantial anticompetitive harm" (Id. ¶ 18), "brazenly unlawful" (Id. ¶18), "extraction of monopoly profits" (Id. ¶ 18), "coercive tactics" (Id. ¶ 19), "auction-manipulation programs" (Id. ¶ 21), "reduced output" "exit of rival firms" (Id. ¶ 502), "exclusionary conduct" (Id. ¶ 505), "foreclosure of competition" (Id. ¶ 519), "allocate markets" (Id. ¶ 544). These legal labels are useful in framing a legal theory but, in assessing whether the facts alleged plausibly state a claim, they are not afforded the presumption of truth.

Justice Souter, writing for the majority, noted the particular importance of enforcing pleading standards in the antitrust context:

[I]t is one thing to be cautious before dismissing an antitrust complaint in advance of discovery, but quite another to forget that proceeding to antitrust discovery can be expensive. As we indicated over 20 years ago ... "a district court must retain the power to insist upon some specificity in pleading before allowing a potentially massive factual controversy to proceed."

* * * *

It is no answer to say that a claim just shy of a plausible entitlement to relief

can, if groundless, be weeded out early in the discovery process....

Id. at 558-59, 127 S.Ct. 1955 (citation omitted; quoting Associated Gen. Contractors of Cal., Inc. v. Carpenters, 459 U.S. 519, 528, n.17, 103 S.Ct. 897, 74 L.Ed.2d 723 (1983)).

Thus, to survive a motion to dismiss under Rule 12(b)(6), "a complaint must contain sufficient factual matter, accepted as true, to 'state a claim to relief that is plausible on its face.'" Ashcroft v. Iqbal, 556 U.S. 662, 678, 129 S.Ct. 1937, 173 L.Ed.2d 868 (2009) (quoting Twombly, 550 U.S. at 570, 127 S.Ct. 1955). Legal conclusions are not entitled to the presumption of truth, and a court assessing the sufficiency of a complaint disregards them. Id. Instead, the Court must examine only the well-pleaded factual allegations, if any, "and then determine whether they plausibly give rise to an entitlement to relief." Id. at 679, 129 S.Ct. 1937. The Complaint must include non-conclusory factual allegations that "'nudge[]'" its claims "'across the line from conceivable to plausible.'" Id. at 680, 129 S.Ct. 1937 (quoting Twombly, 550 U.S. at 570, 127 S.Ct. 1955). A court must draw all reasonable inferences in favor of the plaintiff as non-movant. Peretti v. Authentic Brands Grp. LLC, 33 F.4th 131, 137 (2d Cir. 2022). "Dismissal is appropriate when 'it is clear from the face of the complaint, and matters of which the court may take judicial notice, that the plaintiff's claims are barred as a matter of law.'" Parkcentral Global Hub Ltd. v. Porsche Auto. Holdings SE, 763 F.3d 198, 208-09 (2d Cir. 2014) (quoting Conopco, Inc. v. Roll Int'l, 231 F.3d 82, 86 (2d Cir. 2000)).

Because the Court's ruling on the tying claim and the claim relating to the Facebook agreement may have bearing on the monopolization and attempted monopolization claims, the Court will address Counts III and IV before addressing Counts I and II.

IV. COUNT III OF THE COMPLAINT PLAUSIBLY ALLEGES A SECTION 1 TYING CLAIM.

The States allege that Google leveraged the monopoly power of its AdX ad exchange to coerce publishers to license its DFP publisher ad server, in violation of sections 1 and 2 of the Sherman Act. "A tying arrangement is 'an agreement by a party to sell one product but only on the condition that the buyer also purchases a different (or tied) product, or at least agrees that he will not purchase that product from any other supplier." Eastman Kodak Co. v. Image Tech. Servs., Inc., 504 U.S. 451, 461, 112 S.Ct. 2072, 119 L.Ed.2d 265 (1992) (quoting Northern Pacific Railway Co. v. United States, 356 U.S. 1, 5-6, 78 S.Ct. 514, 2 L.Ed.2d 545 (1958)). "To state a valid tying claim under the Sherman Act, a plaintiff must allege facts plausibly showing that: (i) the sale of one product (the tying product) is conditioned on the purchase of a separate product (the tied product); (ii) the seller uses actual coercion to force buyers to purchase the tied product; (iii) the seller has sufficient economic power in the tying product market to coerce purchasers into buying the tied product; (iv) the tie-in has anticompetitive effects in the tied market; and (v) a not insubstantial amount of interstate commerce is involved in the tied market." Kaufman v. Time Warner, 836 F.3d 137, 141 (2d Cir. 2016); accord E & L Consulting, Ltd. v. Doman Indus. Ltd., 472 F.3d 23, 31 (2d Cir. 2006).

"[T]he essential characteristic of an invalid tying arrangement lies in the seller's exploitation of its control over the tying product to force the buyer into the purchase of the tied product that the buyer either did not want at all, or might have

preferred to purchase elsewhere on different terms." Jefferson Parish Hosp. Dist. No. 2 v. Hyde, 466 U.S. 2, 12, 104 S.Ct. 1551, 80 L.Ed.2d 2 (1984), abrogated on other grounds by Illinois Tool Works Inc. v. Indep. Ink, Inc., 547 U.S. 28, 126 S.Ct. 1281, 164 L.Ed.2d 26 (2006). If monopoly power (or sufficient market power) is alleged and proven, the tying arrangement may be unlawful per se without the need to prove anticompetitive effects or other market conditions. Jefferson Parish, 466 U.S. at 15, 17, 104 S.Ct. 1551.

To allege an unlawful tying arrangement, a plaintiff must plausibly allege two separate and distinct products, and an agreement by defendant to sell one product only on condition that the buyer also purchase the other product. Northern Pacific, 356 U.S. at 5-6, 78 S.Ct. 514. In determining whether two products are separate, "the question ... turns not on the functional relation between them, but rather on the character of the demand for the two items." Jefferson Parish, 466 U.S. at 19, 104 S.Ct. 1551. In addressing this question, courts have considered whether the products are sold independently. See, e.g., Eastman Kodak, 504 U.S. at 462, 112 S.Ct. 2072 ("[F]or service and parts to be considered two distinct products, there must be sufficient consumer demand so that it is efficient for a firm to provide service separately from parts.").

The Court concludes that the States have plausibly alleged that publisher ad servers and ad exchanges are separate and distinct products. First, the States have alleged that ad servers and ad exchanges are separate markets. (Compl't ¶¶ 93-112, 128-160, 528.) On this motion to dismiss, Google has not challenged the sufficiency of these allegations. Second, there is an adequate and plausible allegation that the ad servers and ad exchanges were sold separately before the alleged unlawful tying. (Id. ¶¶ 245-46.) Google entered the ad server market in 2008 with the acquisition of Double Click. (Id. ¶ 245.) At the time of the acquisition, DoubleClick's market share was between 48% and 57%, with the balance of the market in the hands of "well-funded" competitors, including 24/7 Real Media (owned by a publicly traded company), aQuantive (owned by Microsoft), and ValueClick (a publicly traded company). (Id.)

The Court concludes that the States plausibly allege actual coercion by Google. "[A]ctual coercion by the seller that in fact forces the buyer to purchase the tied product" is a required element of a tying violation. Unijax, Inc. v. Champion Int'l, Inc., 683 F.2d 678, 684 (2d Cir. 1982). "Actual coercion supporting a finding of a tying violation is present only if the manufacturer goes beyond persuasion and conditions [the buyer's] purchase of one product on the purchase of another product." Id. at 685. "[W]here the buyer is free to take either product by itself there is no tying problem even though the seller may also offer the two items as a unit at a single price." Northern Pacific, 356 U.S. at 6 n.4, 78 S.Ct. 514.

Beginning in 2010, Google restricted the ability of publishers using a non-Google ad server to trade through AdX, only allowing publishers that license Google's ad server to receive live, competitive bids from AdX. (Compl't ¶ 246.) This became coercive because publishers depend on AdX to access hundreds of thousands of small advertisers that purchase ad space through Google's ad-buying tool for small advertisers and transact exclusively on AdX. (Id.) "[O]ne large publisher contemplating a switch from Google's ad server in 2017 abandoned those plans after determining that the resulting loss of AdX bids would cost the publisher several million dollars per year." (Id.) The States allege that publishers' revenue

would decline by between 20% and 40% if they used an ad server other than Google's. (Id.) Google's restriction on the receipt of live, competitive bids only to those publishers using its DFP ad server was successful, and Google's share of the ad-server market grew to 78%. (Id. ¶ 249.)

In early 2018, Google began renegotiating agreements with publishers, "requiring publishers to sign a combined contract that included both Google's DFP ad server and Google's AdX exchange." (Id. ¶ 251.) By 2019, Google's share of the ad server market had grown to 90%. (Id. ¶ 249.) The States allege that all agreements with publishers have been negotiated or renegotiated to require publishers seeking access to Google's AdX exchange also license Google's ad server for publishers. (Id. ¶ 251.)

The Court concludes that the States also have plausibly alleged that Google had monopoly power in both the markets for ad servers and for ad exchanges, and that its actions have had anticompetitive effects in both markets. (Id. ¶¶ 505-17, 528.) The States have also plausibly alleged a substantial effect on interstate commerce. (Id. ¶ 526.)

On this motion, Google argues that there is no plausible claim of actual coercion. It asserts that live competitive bidding on its AdX exchange was an important technological innovation that drove publishers to use its DFP ad server. Google endeavors to recast the States' allegations: it "boils down to an assertion that Google should be required to give competing ad servers access to AdX live bids." (Google Mem. at 29.) But ad servers and ad exchanges are plausibly alleged to be separate products in separate markets. Google enjoyed monopoly power in the exchange market with AdX. For the purpose of the tying claim, it does not matter how that monopoly power was acquired, but rather how it was used. At the motion to dismiss stage, the States have plausibly alleged that in 2018, Google began "requiring publishers to sign a combined contract that included both Google's DFP ad server and Google's AdX exchange." (Id. ¶ 251; emphasis added).)⁸ Of course, at the summary judgment stage or at trial Google is free to meet and dispute the allegations of actual coercion with its own evidence.

8 Google urges that this allegation should be read as merely requiring a publisher to enter into a contract that permitted the publisher to decide whether it wishes to include either or both DFP and AdX. (8/31 Tr. at 14-15.) But on a motion to dismiss, the Court accepts as true the States' non-conclusory factual allegations and draws all reasonable inferences in their favor. The Court accepts the allegation as it is most naturally read: publishers were required to sign a contract for both products.

Section 1 prohibits "[e]very contract, combination in the form of trust or otherwise, or conspiracy, in restraint of trade or commerce among the several States ...." 15 U.S.C. § 1.⁹ An unlawful tying arrangement is most naturally examined under section 1. Eastman Kodak, 504 U.S. at 462, 112 S.Ct. 2072. But a tying arrangement may also satisfy the elements of a monopolization claim under section 2, provided the party is alleged to have "(1) the possession of monopoly power in the relevant market and (2) the willful acquisition or maintenance of that power as distinguished from growth or development as a consequence of a superior product, business acumen, or historic accident." United States v. Grinnell Corp., 384 U.S. 563, 570-71, 86 S.Ct. 1698, 16 L.Ed.2d 778 (1966).

9 It has long been recognized that section 1 "reaches only unreasonable restraints" of trade. E & L Consulting, 472 F.3d at 29 (quotation marks omitted).

The States plausibly allege that Google used its monopoly power in AdX to actually

coerce publishers into licensing a separate and distinct product, Google's DFP ad server, and that Google's actions had anticompetitive effects in both markets, affecting a substantial amount of interstate commerce. This is more than sufficient to state a claim for relief under section 1 of the Sherman Act. The Court will later consider the tying arrangement in the context of the section 2 monopolization claim asserted in Count I.

V. COUNT IV DOES NOT PLAUSIBLY ALLEGE A SECTION 1 CLAIM BASED ON GOOGLE'S AGREEMENTS WITH FACEBOOK.

A. The States Have Not Plausibly Alleged an Unlawful Agreement Between Google and Facebook to Restrain Facebook's Use of Header Bidding.

In late September 2018, Google and Facebook entered into a Network Bidding Agreement (the "NBA").¹⁰ The States allege that the NBA was one part of an "unlawful agreement" by which Facebook substantially curtailed its use of a practice called header bidding in return for Google giving Facebook a leg up in publishers' web display and developers' in-app ad auctions, allocating a portion of the wins to Facebook, and helping Facebook's FAN ad network beat the competition." (Compl't ¶ 413.)

10 In the Complaint and briefing on the motion, the NBA is often referred to as the "Jedi Blue" agreement, a purported "codename" used by Google. (Compl't ¶¶ 26, 425.) The Star Wars reference refers to Google and "Blue" to Facebook, the color of Facebook's logo. (Id.)

The terms of the NBA do not expressly or by reasonable implication refer to or restrict Facebook's use of header bidding.¹¹ By its express terms, it "does not restrict Facebook from developing or enhancing a product or service that competes with ... [Google's] Program without Google's Information." (NBA ¶ 2.4(e).) It is "not an exclusive agreement." (NBA ¶ 19.10.) It contains commercial terms described below that were advantageous to Google and to Facebook. The States assert that the "special advantages" given to Facebook in the NBA were "part of [Google's] effort to kill header bidding."

11 The NBA is cited and quoted in the text of the Complaint and thus may be considered on Google's motion to dismiss. DiFolco v. MSNBC Cable L.L.C., 622 F.3d 104, 111 (2d Cir. 2010) (on a motion to dismiss, a court may consider documents incorporated by reference in the complaint and documents that the complaint heavily relies upon, rendering it integral to the complaint).

"In order to establish a conspiracy in violation of [section] 1, whether horizontal, vertical, or both, proof of joint or concerted action is required; proof of unilateral action does not suffice." Anderson News, L.L.C. v. American Media, Inc., 680 F.3d 162, 183 (2d Cir. 2012).¹² There is no requirement that the terms of an unlawful agreement be expressed in any particular manner or in written form. Conspiratorial agreements "nearly always must be proven through 'inferences that may fairly be drawn from the behavior of the alleged conspirators.'" Id. (quoting Michelman v. Clark-Schwebel Fiber Glass Corp., 534 F.2d 1036, 1043 (2d Cir. 1976)). The question

12 The Court's conclusion that the States have not plausibly alleged an unlawful agreement between Google and Facebook to substantially curtail Facebook's use of header bidding dispenses with the need to decide whether such an agreement should be an analyzed as a horizontal or vertical agreement. Compare Complaint ¶¶ 423-424, alleging that Facebook was a potential competitor in certain relevant markets, with the NBA, the entire thrust of which is directed to Facebook as a user of Google's services.

is whether the States have alleged facts that permit the plausible inference that any curtailment of header bidding by Facebook was the product of an agreement with Google.

The Court concludes that the States' allegations do not plausibly allege joint or concerted action between Google and Facebook to restrict Facebook's use of header bidding. The States' allegations are not plausible because they fail to adequately account for Facebook's motivation to use its economic clout as an advertiser to drive the hardest bargain it could with Google, and that Google was motivated by the legitimate, pro-competitive desire to obtain as much business as possible from Facebook. Twombly, 550 U.S. at 566, 127 S.Ct. 1955 ("there is no reason to infer that the companies had agreed among themselves to do what was only natural anyway...."). An inference of conspiracy is not supported by acts that "made perfect business sense." Mayor & City Council of Baltimore, Md. v. Citigroup, Inc., 709 F.3d 129, 138 (2d Cir. 2013). There is nothing inexplicable or suspicious in the parties entering into the NBA for reasons that go no further than the four corners of that agreement. To better understand the Court's conclusion, further background is provided.

1. Header Bidding.

An innovation known as "header bidding" was introduced in 2014, which enabled publishers, including those that used Google's DFP ad server, to solicit live, competitive bids from multiple exchanges. (Compl't ¶¶ 353-54.) It was called "header bidding" because a publisher could insert a piece of code into the "header" section of a publisher's webpage, which then collected information about the site's visitor and used that information to bid on multiple ad exchanges. (Id.) According to the States, header bidding was "wildly popular" among publishers, and by 2016, approximately 70% of major publishers were using header bidding, resulting in an "ad revenue jump overnight." (Id. ¶¶ 355, 358.)

According to the States, header bidding was a competitive threat to Google's ad server and exchange markets. (Id. ¶ 366.) In June 2017, Google began to allow publishers to route their inventory to more than one exchange at a time and to receive live, competitive bids from other exchanges, as well as AdX, through an initiative called Exchange Bidding. (Id. ¶¶ 366, 377.) Internally, Google referred to Exchange Bidding as "Jedi." (Id. ¶ 366.) Header bidding permitted each exchange to access a cookie on the user's page, while Exchange Bidding encrypted the user's ID. (Id. ¶ 367.) Google also charged publishers a 5% fee on any impression sold through a non-Google exchange. (Id. ¶ 369.) Also, a publisher using Google's Exchange Bidding was required to route its inventory through AdX but could also select other ad exchanges as well. (Id. ¶ 370.) Google maintained the ability to displace the winning bid from another exchange by paying a penny more. (Id. ¶ 377.) Google abandoned this practice in 2019. (Id. ¶ 379).

Google is alleged to have adopted other strategies in 2019 to disadvantage publishers and advantage Google's exchange (e.g., "Minimum Bid to Win" and a "new secret bid optimization strategy" "deception of publishers" "artificial capping" "accelerated Mobile Pages"). (Id. ¶¶ 381-412.) These other strategies, not specifically related to Facebook, are alleged to be anticompetitive behavior that supports its section 2 claim, and also background to Google's unlawful agreement with Facebook allegedly in violation of section 1. In the States' view, these anticompetitive behaviors show a desire to thwart header bidding, which is

the motive and purpose of the alleged unlawful agreement with Facebook.

The States have marshalled evidence that Google feared the impact of header bidding on its business. (Id. ¶ 362 ("RISK: If header bidding consolidates all non-Google demand, we could lose our must-call status and be disintermediated."); ¶ 415 ("Need to fight off the existential threat posed by Header Bidding and FAN [Facebook Audience Network]. This is my personal #1 priority. If we do nothing else, this need[s] to [be] an all hand[s] on deck approach.").) In 2016, an executive-level employee at Google made a presentation in which he stated, "to stop these guys from doing HB [Header Bidding] we probably need to consider something more aggressive." (Id. ¶ 418.)

Facebook publicly flaunted its use or intended use of header bidding. According to the States, "[i]n March 2017, Facebook publicly announced that it would submit bids from FAN to open web publishers using header bidding, via partnerships with technology providers such as Amazon Publisher Services, Amazon's header bidding code library that facilitated implementation of header bidding by open web publishers." (Id. ¶ 414.) Facebook understood that "they [Google] want this deal [the NBA] to kill header bidding." (Id. ¶ 420.) Facebook realized that "build/buy ad tech" was a conceivable option but would require "huge [engineering] and services investment, and patience for sales cycle." (Id. ¶ 422.) Nowhere in these internal memos is there a hint that Facebook was offering a commitment to substantially curtail use of header bidding or that Google was insisting on such a commitment.

2. The Complaint Does Not Plausibly Allege Collusion Between Google and Facebook to Thwart Header Bidding.

As noted, the 48-page NBA does not touch upon or purport to restrict Facebook's use of header bidding. It establishes a program where, depending on the volume of impressions Facebook purchased from publishers, it would receive price concessions off the 10% fee Google was charging other networks. (Id. ¶ 428.) The program also gave Facebook a timing advantage by replacing the 160-millisecond timeout imposed on other marketplaces with a 300-millisecond timeout, thereby giving Facebook the competitive advantage of a longer timeout. (Id. ¶ 429.) Facebook, unlike other exchanges and networks that participated in Exchange Bidding, was allowed to directly contract with publishers. (Id. ¶ 430.) Google also agreed to inform Facebook which impressions were likely targeted to bots, rather than humans, information not provided to other exchanges or networks. (Id. ¶ 431.) Google also committed to improving Facebook's match rates, i.e., the percentage of users that Facebook would be able to identify. (Id. ¶¶ 432-33.) The NBA also restricted Google's use of Facebook's bid data. (Id. ¶ 434 ("Facebook was explicit in demanding that Google be prohibited from using Facebook's bid data for the purpose of advantaging itself.").)

The States assert that the practical impact of the NBA was that "Google ensured that Facebook would not — and, economically, could not — return to support header bidding by imposing significant minimum spend requirements running to hundreds of millions of dollars a year." (Id. ¶ 426.) The Annual Minimum Spend Commitment — the gross amount spent by Facebook on advertising, as distinguished from Google's fee for such advertising — would be a minimum of $20 million during the first phase of the program and could rise to as high as $500 million. (NBA Ex. B.) Google's fees were 10% of the first $500 million and 5% of any spend above $500 million. It is a truism that an advertising

dollar spent under the NBA is a dollar not available to be spent through header bidding, but the Complaint offers no context on the size of Facebook's annual spending on display ads or in-app impressions. The Complaint also does not purport to describe the NBA's effect on Facebook's use of header bidding.

There is no allegation that Facebook was participating in Google ad auctions in the time period immediately preceding the NBA. Facebook itself had been in the exchange market but exited in 2016, two years before the NBA. (Compl't ¶ 159.) Google's actions are consistent with a firm seeking to secure the business of a very large potential customer by offering it favorable terms. In the absence of an allegation of predatory pricing or other exclusionary or improper conduct, these actions are entirely consistent with competition. See Matsushita Elec. Indus. Co. v. Zenith Radio Corp., 475 U.S. 574, 594, 106 S.Ct. 1348, 89 L.Ed.2d 538 (1986) ("[C]utting prices in order to increase business often is the very essence of competition."). Offering favorable terms to a large potential customer, such as Facebook, undoubtedly had the effect of diverting business from a competing service, but this does not convert the agreement into an unreasonable restraint of trade. See Monahan's Marine, Inc. v. Boston Whaler, Inc., 866 F.2d 525, 529 (1st Cir. 1989) (ruling consistent with other circuits that "have found nothing anticompetitive in the simple fact that a seller selectively cuts its prices, or offers other favorable terms, to some of its dealers, even though such discrimination harms the non-favored dealers.") (Breyer, J.); Northeastern Tel. Co. v. Am. Tel. & Tel. Co., 651 F.2d 76, 93 (2d Cir. 1981) ("[A] monopolist's right to compete is not limited to actions undertaken with an altruistic purpose. Even monopolists must be allowed to do as well as they can with their business.").

The Court concludes that the States have failed to plausibly allege an agreement between Google and Facebook to restrain Facebook's use of header bidding.

B. The States Have Failed to Plausibly Allege an Agreement between Google and Facebook to Limit Competitive Bidding for In-App Ad Inventory.

The States have alleged that Google and Facebook agreed in the NBA to "fix[] a minimum share of impressions that Facebook will win in developer auctions" and create a "hard limit on contractually acceptable auction outcomes." (Compl't ¶ 437.) They also assert that Google has effectively excluded rival bidders because the NBA contains terms more favorable to Facebook than those offered to others. (Id. ¶ 441.) The States' allegations are premised on the express terms of the NBA and Google's motion to dismiss is premised on its interpretations of these same terms. Google asserts that the NBA sets no floor on wins, no limits on outcomes and excludes Facebook from no rival in-app network.

Before turning to the provisions of the NBA, some background is necessary on how developers sell in-app impressions to in-app ad networks. Google offers in-app mediation tools to app developers seeking to sell ad impressions. (Compl't ¶ 84.) Unlike display ads, there are no exchanges and no ad-buying tools used in the sale of in-app impressions. The in-app mediation tools, including Google's "AdMob mediation" and "GAM for apps," manage developers' inventory and conduct impression auctions. (Id. ¶ 217.) Advertisers do not typically bid in these auctions: rather, in-app networks do so, and then resell the impressions to advertisers using various pricing devices to achieve a markup. (Id.

¶ 86.) For example, an in-app network may buy impressions at auction on a per-impression basis but then sell them to advertisers on a per-click or per-action basis, or even through the sale of large blocks of impressions. (Id.) "[A]n in-app network must provide a specialized SDK [software development kit] so that the developer's app can call for and display in-app ads in an appropriate manner." (Id. ¶ 87.)

C. The Alleged Restraint on Bidding for In-App Impressions Is Properly Scrutinized Under the Rule of Reason.

The States and Google dispute whether Google's understandings with Facebook are a horizontal restraint between competitors that are a per se violation of section 1, or, instead, should be examined under the rule of reason because they are at different levels of the distribution chain, and therefore a vertical restraint.¹³ "[C]ertain kinds of agreements will so often prove so harmful to competition and so rarely prove justified that the antitrust laws do not require proof that an agreement of that kind is, in fact, anticompetitive in the particular circumstances." NYNEX Corp. v. Discon, Inc., 525 U.S. 128, 133, 119 S.Ct. 493, 142 L.Ed.2d 510 (1998). "A vertical restraint is not illegal per se unless it includes some agreement on price or price levels." Id. at 136, 119 S.Ct. 493 (quotation marks omitted).

13 The States alternatively plead that the arrangements between Google and Facebook fail under a rule of reason analysis.

The Complaint describes Google operating at two different levels of distribution in the sale of developers' in-app impressions. First, Google offered and operated the in-app mediation tool for developers, which acted as an auction mechanism. (Compl't ¶ 15.) In this capacity, Google, acting as auctioneer, was not in competition with Facebook, and stood to profit from Facebook's successful bidding in the auctions. In effect, Google took an auctioneer's fee. Second, Google acted in a different capacity as the operator of an in-app network, and, in that context, it competed with Facebook for impressions offered at auction.¹⁴ (Id. at ¶ 439.)

14 The States acknowledge that the NBA only applies to in-app impressions offered by third parties and not to any that Google or Facebook offer for sale. (States' Mem. at 27 (citing NBA ¶ 1.55).)

Taking the NBA as a whole and in the context of the entirety of the Complaint's allegations, Google, the auctioneer, provided pricing and other incentives to Facebook, the in-app network, to participate in the auctions that Google offered on behalf of developers. Facebook is not alleged to be a participant in the market for in-app mediation tools that Google uses to conduct auctions of developers' inventory of impressions. Prior to entering the NBA, Facebook did not participate in auctions conducted through Google's mediation tool. The NBA's favorable terms to Facebook are most naturally understood as an inducement to do so, with Google standing to gain an auctioneer's fees of 5 or 10%, depending on volume. (NBA Ex. B ¶ 5; Compl't ¶ 427 (provisions of the NBA were intended "to induce Facebook to shift ... to routing bids through Google's ... in-app mediation tools.").)

Read as such, it is principally a vertical agreement, with potential horizontal consequences. See Cenedella v. Metropolitan Museum of Art, 348 F. Supp. 3d 346, 360 (S.D.N.Y. 2018) ("[a]s pleaded, the alleged conspiracy appears to be based primarily upon vertical agreements between the [museums], who occupy one rung of the market structure, and the galleries, auction

houses, and purchasers, who occupy another.") (Koeltl, J.).

In Copy-Data Systems, Inc. v. Toshiba America, Inc., 663 F.2d 405, 406 (2d Cir. 1981), the Second Circuit was presented with a dual-distribution network in which Toshiba initially sold copying equipment to wholesaler distributers, such as plaintiff, who in turn sold to retailers. Thereafter, Toshiba decided to enter the wholesale distribution market, acting as a wholesale distributor competing with plaintiff. Id. at 407. In that capacity, Toshiba, the district court found, had coerced plaintiff to exit certain geographic markets. Id. at 408. The district court concluded that because of the horizontal competition, the actions of the defendant should be assessed as a per se violation. Id. The Circuit reversed, concluding that a dual distributorship was not a per se violation "merely because it contained horizontal elements." Id. at 409.

In a novel marketplace with rapidly changing technology that can alter the nature of relationships between market participants, it is especially important to follow the admonition of Copy-Data: "Expansion of the per se rule should be approached with great caution. 'It is only after considerable experience with certain business relationships that courts classify them as per se violations of the Sherman Act.'" Id. at 411 (internal citation omitted; quoting United States v. Topco Associates, Inc., 405 U.S. 596, 607-08, 92 S.Ct. 1126, 31 L.Ed.2d 515 (1972)).¹⁵ The Court concludes that this hybrid relationship between Google and Facebook is predominately vertical and properly scrutinized under the rule of reason.

15 See also Texaco Inc. v. Dagher, 547 U.S. 1, 5, 126 S.Ct. 1276, 164 L.Ed.2d 1 (2006) ("[T]his Court presumptively applies rule of reason analysis...."); F.T.C. v. Indiana Fed'n of Dentists, 476 U.S. 447, 458-59, 106 S.Ct. 2009, 90 L.Ed.2d 445 (1986) (the Supreme Court has been slow "in general, to extend per se analysis to restraints imposed in the context of business relationships where the economic impact of certain practices is not immediately obvious ...."); E & L Consulting, 472 F.3d at 29 (per se rule applies to a "small class of actions.").

D. The States Have Failed to Plausibly Allege a Restraint on Bidding for In-App Impressions under the Rule of Reason.

Under the rule of reason, the plaintiff must show that "defendants' challenged behavior 'had an actual adverse effect on competition as a whole in the relevant market.'" Geneva Pharms. Tech. Corp. v. Barr Labs. Inc., 386 F.3d 485, 507 (2d Cir. 2004) (quoting Capital Imaging Assocs., P.C. v. Mohawk Valley Med. Assocs., Inc., 996 F.2d 537, 543 (2d Cir. 1993)). To state a claim under the rule of reason, a complaint must contain facts that plausibly describe harm to competition. See E & L Consulting, 472 F.3d at 29-31; Elecs. Commc'ns Corp. v. Toshiba Am. Consumer Prod., Inc., 129 F.3d 240, 245 (2d Cir. 1997). Harm to competition may be inferred through facts such as "reduced output, increased prices and decreased quality" in the relevant market. Virgin Atl. Airways Ltd. v. British Airways PLC, 257 F.3d 256, 264 (2d Cir. 2001). "Without any allegation as to how market-wide competition will be affected, the complaint fails to allege a claim on which relief can be granted." Elecs. Commc'ns, 129 F.3d at 245.

A plaintiff alleging harm to competition may, in lieu of showing actual harm, allege that defendants possess "market power" in the relevant market. Tops Mkts., 142 F.3d at 97. Market power is defined as "the ability to raise price significantly above the competitive level without losing all of one's business." CDC Techs., Inc. v. IDEXX Labs., Inc., 186 F.3d 74, 81

(2d Cir. 1999). Market share may be used as a proxy to demonstrate market power. K.M.B. Warehouse Distributors, Inc. v. Walker Mfg. Co., 61 F.3d 123, 129 (2d Cir. 1995). However, market share alone is insufficient to demonstrate market power; plaintiff must show market share "plus some other ground for believing that the challenged behavior could harm competition in the market, such as the inherent anticompetitive nature of the defendant's behavior or the structure of the ... market." Tops Mkts., 142 F.3d at 97.

The States focus particular attention on a term of the NBA that provides that "Facebook will ensure that the Win Rate is at least 10%." (NBA. Ex. A. ¶ 5.) The win rate is defined as the percentage of auctions that Facebook wins out of the total number of bids (called "bid responses") that Facebooks submits in the auction process. (Id. Ex. A ¶ 1(c).) It also provides that Google and Facebook would work together so that Facebook could identify the end user for at least 80% of auctions conducted by Google, and that Facebook would use commercially reasonable efforts to bid on at least 90% of those auctions; based on these other provisions, the States calculate that the effective win rate is actually 7.2%. (Compl't ¶ 438.) While not agreeing with the States' legal conclusions, Google has not challenged the accuracy of the 7.2% figure. According to the States, "more than 60 percent" of all in-app impressions are sold through the auction on Google's in-app mediation tool. (Id. ¶ 228.)

But the win-rate provision does not on its face predetermine the outcome of any auction. It ensures that Facebook submits competitive bids, not throwaway ones. The provision does not constrain Google's actions as a participant in the auctions and does not limit the number of auctions in which Facebook may participate.¹⁶

16 In its brief but not in the Complaint, the States argue that Google is penalized if Facebook does not achieve the target win rate. (States' Mem. at 31.) The cited provision applies only if the shortfall is "due to Google's categorization" of certain data, and not because of the bidding practices of Google or Facebook. (NBA ¶ 6.10.)

The States do not adequately explain why inducing Facebook to actively participate in the Google-run auctions — and endeavor to win a designated percentage of auctions — does not promote rather than harm competition in the in-app network market. The win-rate provision cannot be reasonably read to require collusive bidding or any other form of distorted bidding by Google or Facebook. Conspirators do not always reduce their understandings to writing and if they do, it does not necessarily follow that the writing should be taken at face value. But the Complaint does not plausibly allege the existence of side deals, winks or nods that augment the terms of the NBA insofar as they relate to the States' claim concerning bidding for in-app impressions. That collusive bidding agreements could exist is not the same as offering facts that support a plausible inference that they do.

The NBA does not dictate which impressions Facebook may bid on or at what price. Rather than insulate Google's in-app network from competition, it promotes competition with Google's in-app network by bringing in a new competing bidder. "By threatening to disrupt and then cutting a deal with Google, Facebook was able to achieve what others could not: an opportunity to compete against Google for publishers' and developers' inventory on equal terms." (Id. ¶ 436.)

The States seek to stretch the NBA beyond a fair reading, arguing that if Facebook is contractually obligated to strive

to win 7.2% of the auctions, then this means that Google may compete for only 92.8% of the auction wins. The States make no plausible allegation that Google had the capacity to absorb such a high percentage of impressions offered at auction, given that a successful bidder owns the impressions and must resell them to advertisers, assuming the risk of loss on the resale.

The States neatly set out "The Relevant Markets and Google's Market Power" in Section VI of the Complaint and the "Anticompetitive Effects" in Section VIII. Notably, the Complaint alleges that Google has "market power" in the in-app mediation tool market but does not plausibly allege any anticompetitive effects in that market. It does not plausibly allege that Google has market power in the in-app networks market, but does allege anticompetitive effects in that market.

With respect to the market for in-app mediation tools, the States allege that 60% of all apps with an inventory of impressions for sale use the Google in-app mediation tool, i.e., offer their inventory of in-app impressions through the Google auction. (Id. ¶ 228.) The Complaint alleges in a conclusory fashion that Google has market power in the in-app mediation tool market, but it has failed to plausibly allege that Google has the ability to control prices or exclude competition.

With respect to the in-app networks market, the Complaint does not allege the market share of any of the in-app networks, including those of Google or Facebook. According to the States, there are 25 in-app networks that participate in Google-run auctions; Google's in-app network is the largest participant on the buy side of the auctions and Facebook is the second-largest bidder. (Id. ¶¶ 89, 439.)¹⁷

17 See K.M.B. Warehouse, 61 F.3d at 128 ("[A]n examination of the intrabrand market belies even [plaintiff's] assertion that intrabrand competition has been harmed. There are over twenty Walker distributors of various size in the Tri-state area.").

As noted, the States do not allege harm to competition in the in-app mediation tool market. (See id. ¶ 502, et seq.) They do purport to allege harm to competition in the in-app network market. (Id. ¶¶ 437, 518, 519.) They assert that the overall effect is to "depress prices paid to developers." (Id. ¶ 437.) The States acknowledge that when advertisers "pay less to purchase in-app impressions, [it] permit[s] them to re-invest those cost savings into providing consumers with higher-quality and lower-priced goods and services." (Id. ¶ 519.) But immediately following that statement, they add that "Google's foreclosure of competition in the in-app network market has permitted its in-app networks to pay less for developers' impressions and resell those impressions to advertisers at higher prices." (Id.) Nothing in the NBA or the States' pleading plausibly explains why lower acquisition costs for impressions would necessarily lead to higher resale "prices," as distinguished from higher margins.

As the Court noted at the outset of this discussion, the States must plausibly allege that a defendant's challenged behavior had an adverse effect on competition as a whole in a relevant market. Geneva Pharmaceuticals, 386 F.3d at 507. Because the States have failed to do so, their section 1 claim premised on Google's understandings with Facebook relating to the purchase and sale of in-app impressions fails to state a claim for relief.¹⁸ VI. CERTAIN OF THE STATES' ALLEGATIONS PLAUSIBLY DESCRIBE ANTICOMPETIVE CONDUCT AND STATE CLAIMS FOR MONOPOLIZATION AND ATTEMPTED MONOPOLIZATION UNDER SECTION 2.

18 While every pleading stands on its own footing, pleadings alleging rule of reason violations are frequently dismissed for failure to adequately allege market harm. See, e.g. Cinema Vill. Cinemart, Inc. v. Regal Ent. Grp., 15 cv 05488 (RJS), 2016 WL 5719790, at *4 (S.D.N.Y. Sept. 29, 2016) (Sullivan, J.) (dismissing complaint in part for failure to adequately plead actual market harm sufficient to show an unreasonable restraint of trade), aff'd, 708 Fed. App'x 29 (2d Cir. 2017); Wellnx Life Sciences Inc. v. Iovate Health Sciences Research Inc., 516 F. Supp. 2d 270, 293-95 (S.D.N.Y. 2007) (dismissing section 1 claim because complaint failed to plausibly allege adverse effects on price, output or quality of services); IDT Corp. v. Bldg. Owners & Managers Ass'n Int'l, 03-4113 (JAG), 2005 WL 3447615, at *12 (D.N.J. Dec. 15, 2005) (dismissal for failure to plausibly allege anticompetitive effects).

The States allege that, through a series of anticompetitive actions, Google willfully acquired or maintained monopoly power in the nationwide markets for (1) publisher ad servers, (2) ad exchanges, and (3) ad-buying tools for small advertisers. Google vigorously contests the existence of anticompetitive conduct but does not dispute the existence of the three defined markets or its possession of monopoly power in the markets for ad servers or ad-buying tools for small advertisers. As noted, Google has challenged the existence of monopoly power in the ad-exchange market, but this Court has already ruled that the States have plausibly alleged monopoly power in that market.

With regard to Count I, the Court ultimately concludes that the States have plausibly alleged that Google has monopoly power in and willfully engaged in anticompetitive conduct, i.e., conduct that harms competition in the markets for ad exchanges, ad-buying tools for small advertisers and publisher ad servers. Thus, the States plausibly allege facts that state a claim for monopolization in these markets.

The attempt-to-monopolize claim (Count II) is directed to three markets: the markets for ad exchanges, ad-buying tools for small advertisers and ad-buying tools for large advertisers.¹⁹ On the attempt-to-monopolize claim, Google advances no argument that it lacked the requisite intent to monopolize or that, if there was anticompetitive conduct, there was no dangerous probability of success in achieving monopoly power. The Court concludes that the States have plausibly alleged anticompetitive conduct, a specific intent to monopolize and a dangerous probability of achieving monopoly power in the markets for ad exchanges, ad-buying tools for small advertisers and ad-buying tools for large advertisers.

19 The claim is pled in the alternative to its monopolization claim for ad exchanges and ad-buying tools for small advertisers. (Compl't ¶ 532.)

The Court begins by laying out the elements of a monopolization claim and then the somewhat overlapping elements of an attempt-to-monopolize claim.

A. Monopolization.

"The offense of monopoly under [Section 2] of the Sherman Act has two elements: (1) the possession of monopoly power in the relevant market and (2) the willful acquisition or maintenance of that power as distinguished from growth or development as a consequence of a superior product, business acumen, or historic accident." United Food & Com. Workers Loc. 1776 & Participating Emps. Health & Welfare Fund v. Takeda Pharm. Co. Ltd., 11 F.4th 118, 137 (2d Cir. 2021) (quoting Grinnell, 384 U.S. at 570-71, 86 S.Ct. 1698).

Monopoly power includes "'the power to control prices or exclude competition.'"

PepsiCo, 315 F.3d at 107 (quoting United States v. E.I. du Pont de Nemours & Co., 351 U.S. 377, 391, 76 S.Ct. 994, 100 L.Ed. 1264 (1956)); accord Takeda, 11 F.4th at 137. "While market share is not the functional equivalent of monopoly power, it nevertheless is highly relevant to the determination of monopoly power. A court may infer monopoly power from a high market share." Tops Markets, 142 F.3d at 98 (citation omitted). Other characteristics that should be considered include "strength of the competition, the probable development of the industry, the barriers to entry, the nature of the anticompetitive conduct and the elasticity of consumer demand." Id. (quoting International Distrib. Ctrs., Inc. v. Walsh Trucking, 812 F.2d 786, 792 (2d Cir. 1987)).

But monopoly power does not, standing alone, violate the Sherman Act:

The mere possession of monopoly power, and the concomitant charging of monopoly prices, is not only not unlawful; it is an important element of the free-market system. The opportunity to charge monopoly prices — at least for a short period — is what attracts 'business acumen' in the first place; it induces risk taking that produces innovation and economic growth. To safeguard the incentive to innovate, the possession of monopoly power will not be found unlawful unless it is accompanied by an element of anticompetitive conduct.

Verizon Commc'ns Inc. v. Law Offices of Curtis V. Trinko, LLP, 540 U.S. 398, 407, 124 S.Ct. 872, 157 L.Ed.2d 823 (2004) (emphasis in original); see also U.S. Football League v. Nat'l Football League, 842 F.2d 1335, 1360 (2d Cir. 1988) ("'[A] firm with lawful monopoly power has no general duty to help its competitors, whether by holding a price umbrella over their heads or by otherwise pulling its competitive punches.'") (quoting Olympia Equip. Leasing Co. v. W. Union Tel. Co., 797 F.2d 370, 375 (7th Cir. 1986) (Posner, J.)). "The law directs itself not against conduct which is competitive, even severely so, but against conduct which unfairly tends to destroy competition itself. It does so not out of solicitude for private concerns but out of concern for the public interest." Spectrum Sports, Inc. v. McQuillan, 506 U.S. 447, 458, 113 S.Ct. 884, 122 L.Ed.2d 247 (1993). "It is sometimes difficult to distinguish robust competition from conduct with long-term anticompetitive effects; moreover, single-firm activity is unlike concerted activity covered by § 1, which 'inherently is fraught with anticompetitive risk.'" Id. at 458-59, 113 S.Ct. 884 (quoting Copperweld Corp. v. Indep. Tube Corp., 467 U.S. 752, 767-69, 104 S.Ct. 2731, 81 L.Ed.2d 628 (1984)).

"The pertinent inquiry in a monopolization claim, then, is whether the defendant has engaged in improper conduct that has or is likely to have the effect of controlling prices or excluding competition, thus creating or maintaining market power." PepsiCo, 315 F.3d at 108.

The Second Circuit has described anticompetitive conduct as "conduct without a legitimate business purpose that make sense only because it eliminates competition." In re Adderall XR Antitrust Litig., 754 F.3d 128, 133 (2d Cir. 2014) (quotation marks omitted). "'The test is not total foreclosure, but rather whether the challenged practices bar a substantial number of rivals or severely restrict the market's ambit.'" New York ex rel. Schneiderman v. Actavis PLC, 787 F.3d 638, 656 (2d Cir. 2015) (quoting United States v. Dentsply Int'l, Inc., 399 F.3d 181, 191 (3d Cir. 2005)).

Anticompetitive conduct must represent something more than business activity that occurs in the normal competitive process or fosters commercial success. See

In re Adderall, 754 F.3d at 133-36.²⁰ Harm to competition is different than harm to a single competitor or group of competitors, which does not necessarily constitute harm to competition. Brunswick Corp. v. Pueblo Bowl-O-Mat, Inc., 429 U.S. 477, 488, 97 S.Ct. 690, 50 L.Ed.2d 701 (1977) ("The antitrust laws ... were enacted for the protection of competition not competitors.") (quotation marks omitted).

20 Examples of anticompetitive conduct include engaging in price manipulation to harm rivals after obtaining a dominant market position, as well as conduct that harms competitors with no legitimate economic justification to the monopolist. See In re Crude Oil Commodity Futures Litig., 913 F. Supp. 2d 41, 56 (S.D.N.Y. 2012) (Pauley, J.). Using monopoly power to boost sales through consumer coercion — as opposed to persuasion — also constitutes improper conduct. Actavis, 787 F.3d at 654-55.

Generally, misleading or deceptive statements to market participants standing alone do not amount to anticompetitive conduct for the purposes of section 2. See, e.g., Interstate Properties v. Pyramid Co. of Utica, 586 F. Supp. 1160, 1163 (S.D.N.Y. 1984) ("Not every business tort can be molded into an antitrust violation."). "Deceptive conduct — like any other kind — must have an anticompetitive effect in order to form the basis of a monopolization claim." Rambus Inc. v. F.T.C., 522 F.3d 456, 464 (D.C. Cir. 2008).²¹ "Even if deception raises the price secured by a seller, but does so without harming competition, it is beyond the antitrust laws' reach." Id.

21 Google has urged that any allegation of deceptive conduct must satisfy the particularity required by Rule 9(b) and demonstrate actual harm to competition, pointing to Judge Sand's decision in Michael Anthony Jewelers, Inc. v. Peacock Jewelry, Inc., 795 F. Supp. 639 (S.D.N.Y. 1992). (See 8/31/22 Tr. at 65; Google Reply at 26.) The section 2 claim in Michael Anthony Jewelers "relie[d] heavily" on the anticompetitive effects of allegedly false and fraudulent statements of originality filed with the Copyright Office. Id. at 643-44, 649. Judge Sand concluded that the Complaint satisfied Rule 9(b) by attaching the allegedly fraudulent copyright filings and alleging scienter. Id. at 649-50. Here, Rule 9(b) has no role because the States do not assert that Google's purported section 2 violations sound in fraud and the purportedly deceptive conduct described in the Complaint is offered as context. See, e.g., Hinds Cnty., Miss. v. Wachovia Bank N.A., 700 F. Supp. 2d 378, 391-92 (S.D.N.Y. 2010) (concluding that Rule 9(b) played no role in antitrust-conspiracy claim centered on bid-rigging and price-fixing, as opposed to acts of fraud) (Marrero, J.).

B. Attempt to Monopolize.

For a claim of attempted monopolization, generally "a plaintiff must prove (1) that the defendant has engaged in predatory or anticompetitive conduct with (2) a specific intent to monopolize and (3) a dangerous probability of achieving monopoly power." Spectrum Sports, Inc., 506 U.S. at 456, 113 S.Ct. 884. A claim for attempt to monopolize is not viable "absent proof of a dangerous probability that [defendants] would monopolize a particular market and specific intent to monopolize." Id. at 459, 113 S.Ct. 884. Without plausible allegations of a defined market to which the intent and actions are directed, there is no actionable claim. Id. The specific intent to monopolize is "something more than an intent to compete vigorously." Id. It may be proven through the nature of the anticompetitive actions that have been taken. Id. Intent also "can be derived from [defendants'] words" or a "stated goal." Tops Markets, 142 F.3d at 101.

"A dangerous probability of success may exist where the defendant possesses a significant market share when it undertakes the challenged anticompetitive conduct." H.L. Hayden Co. of New York v. Siemens Med. Sys., Inc., 879 F.2d 1005, 1017 (2d Cir. 1989). A 20% market share

by each of two market participants has been held to be insufficient to support a dangerous probability of achieving monopoly power. Id. at 1018.

While Google challenges the various categories of anticompetitive conduct, on this motion it mounts no separate attack on the "specific intent" or "dangerous probability" prongs of the attempt-to-monopolize claim.

C. Monopoly Broth.

Google urges that, to the extent the Complaint asserts a section 2 claim under a "monopoly broth" theory, it should be dismissed. The Complaint does not expressly invoke "monopoly broth" as a theory of liability, and instead asserts that there was a synergistic effect to certain of Google's anticompetitive conduct.

"[I]n analyzing antitrust claims alleging a variety of anticompetitive conduct, courts must tow the line between two competing considerations. First, the Court must avoid 'tightly compartmentalizing the various factual components' of a plaintiff's claims and 'wiping the slate clean after scrutiny of each.' At the same time, it is unlikely that multiple independently lawful acts can come together to create an unlawful monopoly 'broth' from which antitrust injury can arise." Valassis Commc'ns, Inc. v. News Corp., 2019 WL 802093, at *9 (S.D.N.Y. Feb. 21, 2019) (quoting City of Groton v. Connecticut Light & Power Co., 662 F.2d 921, 928-29 (2d Cir. 1981)).

The Complaint twice asserts that Google's purportedly unlawful schemes were "more powerful" because of their "combined effect". (Compl't ¶¶ 294, 412.) But there is nothing remarkable in such an allegation. See Valassis, 2019 WL 802093, at *9-10. In assessing whether conduct is anticompetitive, it is often necessary to examine other market conduct, including practices that are perfectly lawful. Simply put, there is no monopoly broth claim that could be the proper subject of a motion to dismiss.

D. The Alleged Anticompetitive Conduct Supporting the Monopolization and Attempt to Monopolize Claims.

The Court will review the conduct of Google that the States contend is anticompetitive and determine whether those allegations plausibly allege anticompetitive conduct. Because anticompetitive conduct does not exist in the abstract but is aimed at one or more markets, the Court must also ascertain the markets in which harm to competition has been plausibly alleged.

The Court looks at the specific anticompetitive conduct both in isolation and also as part of a broader alleged pattern of conduct to determine whether the States have plausibly alleged anticompetitive conduct in a particular market.

1. Google's Use of Encrypted User IDs Is Not Plausibly Alleged to be Anticompetitive Conduct.

As described in the Complaint, Google's DFP ad server manages publisher inventory, and, "[o]n behalf of publishers," identifies site visitors and assigns them unique user IDs. (Compl't ¶ 255.) This user ID helps advertisers reach their target audiences: for example, it identifies motorcycle enthusiasts for an advertiser that sells motorcycle accessories. (Id. ¶ 255.)

Before Google's 2008 acquisition of DoubleClick, DoubleClick's user IDs were not encrypted. (Id. ¶ 256.) But beginning in 2009, when Google launched AdX, its DFP ad server encrypted user IDs and thereby restricted publishers from sharing user IDs with non-Google exchanges. (Id. ¶ 257.) Advertisers using non-Google ad exchanges no longer knew the identity of a user associated with a publisher's impressions and could not avoid the possibility of purchasing ads for the same user who had

accessed two different publisher websites, i.e., it impeded "frequency capping." (Id. ¶¶ 51, 257.) Advertisers therefore made lower bids for fewer publisher impressions. (Id. ¶ 259.)

At the same time, Google's ad server shared Google-encrypted user IDs with Google's own ad exchange and ad-buying tools. (Id. ¶ 261.) This had the effect of advantaging Google, because publishers and advertisers using Google's ad exchange had the ability to know that two different, encrypted user IDs applied to the same user. (Id. ¶ 261.)

The Complaint asserts that Google's encryption of user IDs and refusal to de-encrypt for use on non-Google ad exchanges "can only be explained by the promise of monopoly profits." (Id. ¶ 265.) The Complaint acknowledges that Google publicly claimed that the move to encrypted user ids promoted user privacy interests but asserts that this was "pure pretext." (Id. ¶ 263.) The States do not plausibly allege why the stated reason was pretext. They do not explain why the abandonment of the "cookie" system used by other exchanges (and header bidding) in favor of encryption to all but those in privity of contract with Google was not a desirable feature to users. The benefits to users may have been exaggerated or misstated by Google, but the States have not explained why this was not an innovation that consumers rightly or wrongly preferred.

The States assert that Google's encryption of user IDs has foreclosed competition in the exchange market and the buying-tool markets for small and large advertisers. (Id. ¶ 266.) It asserts that advertisers responded by directing more of their business to Google's buying tools and ad exchange. (Id. ¶ 266.)

The Supreme Court has warned against a reading of section 2 that would require a monopolist to provide information or assistance to its competitors. "[A]s a general matter, the Sherman Act 'does not restrict the long recognized right of [a] trader or manufacturer engaged in an entirely private business, freely to exercise his own independent discretion as to parties with whom he will deal.'" Trinko, 540 U.S. at 408, 124 S.Ct. 872 (quoting United States v. Colgate & Co., 250 U.S. 300, 307, 39 S.Ct. 465, 63 L.Ed. 992 (1919)). The Supreme Court has explained:

Firms may acquire monopoly power by establishing an infrastructure that renders them uniquely suited to serve their customers. Compelling such firms to share the source of their advantage is in some tension with the underlying purpose of antitrust law, since it may lessen the incentive for the monopolist, the rival, or both to invest in those economically beneficial facilities. Enforced sharing also requires antitrust courts to act as central planners, identifying the proper price, quantity, and other terms of dealing — a role for which they are ill suited. Moreover, compelling negotiation between competitors may facilitate the supreme evil of antitrust: collusion.

Id. at 407-08, 124 S.Ct. 872; see also Pac. Bell Tel. Co. v. linkLine Commc'ns, Inc., 555 U.S. 438, 448, 129 S.Ct. 1109, 172 L.Ed.2d 836 (2009) ("As a general rule, businesses are free to choose the parties with whom they will deal, as well as the prices, terms, and conditions of that dealing."); Novell, Inc. v. Microsoft Corp., 731 F.3d 1064, 1074 (10th Cir. 2013) ("Even a monopolist generally has no duty to share (or continue to share) its intellectual or physical property with a rival.") (Gorsuch, J.).

Following Trinko, the Second Circuit has scrutinized refusal-to-deal claims by looking to whether the purported monopolist

unilaterally terminated a voluntary and profitable course of dealing, as well as whether it chose "'to forsake short-term profits to achieve an anticompetitive end.'" In re Adderall, 754 F.3d at 135 (quoting Trinko, 540 U.S. at 409, 124 S.Ct. 872); accord In re Elevator Antitrust Litig., 502 F.3d 47, 53 (2d Cir. 2007); Charych v. Siriusware, Inc., 790 Fed. App'x 299, 302 (2d Cir. 2019) ("[D]efendants were under no obligation to develop an interface that was compatible with plaintiffs' product.") (summary order). "And although Trinko does not purport to set out a 'test,' it usefully highlights the distinctions that made Aspen Skiing the rare case in which a refusal to deal amounted to a prohibited act of unilateral monopolization." In re Adderall, 754 F.3d at 135.²²

22 See Aspen Skiing Co. v. Aspen Highlands Skiing Corp., 472 U.S. 585, 105 S.Ct. 2847, 86 L.Ed.2d 467 (1985). "Aspen Skiing is at or near the outer boundary of § 2 liability." Trinko, 540 U.S. at 409, 124 S.Ct. 872.

The Complaint does not assert that Google has sacrificed short-term profits by refusing to share Google-encrypted user IDs with those not doing business with Google. Indeed, the Complaint asserts that Google's restriction of IDs was immediately profitable because, "[a]s Google clearly hoped, advertisers began to redirect spend away from non-Google buying tools and exchanges and toward Google's buying tools and exchange ...." (Compl't ¶ 266.) There is no suggestion in the Complaint that Google's decision to not share unencrypted user IDs required the sacrifice of profits to achieve an anticompetitive goal.

The Complaint does not allege facts that would tend to show that Google's decision to encrypt IDs (or, as the Complaint puts it, stop sharing unencrypted IDs) occurred in the context of a profitable arrangement. Then-Judge Gorsuch observed that, while looking to the termination of an existing agreement has the benefit of establishing a limiting principle for refusal-to-deal claims, placing too much weight on this factor alone may inhibit procompetitive arrangements and compel a firm to maintain unprofitable ones:

[R]equiring a preexisting course of dealing as a precondition to antitrust liability risks the possibility that monopolists might be dissuaded from cooperating with rivals even in procompetitive joint venture arrangements — for fear that, once in them, they can never get out. Inversely, this condition risks deterring the termination of joint ventures when they no longer make economic sense.

Novell, 731 F.3d at 1074. "Put simply, the monopolist's conduct must be irrational but for its anticompetitive effect." Id. at 1075; see also In re Adderall, 754 F.3d at 135 (termination of agreements did not support refusal-to-deal claim because "the agreements here were explicitly unprofitable — they introduced price competition into a market where none would otherwise have existed."). The Complaint does not describe why it would have been in Google's economic self-interest to share encrypted user IDs that Google itself generated. It also does not provide historical context for DoubleClick's sharing of user IDs that would tend to show that Google's decision to change course was "irrational but for its anticompetitive effect," Novell, 731 F.3d at 1075, as opposed to an immediately profitable business strategy that also presented the perception or reality of enhanced consumer privacy.

Drawing every reasonable factual inference in favor of the States, the Complaint does not plausibly allege that Google's refusal to share unencrypted user IDs amounted to anticompetitive conduct. 2. The Complaint Plausibly Alleges Google's Use of Dynamic Allocation Was Anticompetitive Conduct in the Ad Exchange Market.

The Complaint describes a practice called Dynamic Allocation, through which Google used its market power in the ad server market to give AdX the right to win an impression if it offered a bid that was higher than the historical average bid on rival exchanges. The States assert that without Dynamic Allocation, these ad impressions would have transacted on rival exchanges, with terms that were more favorable to Google's publisher clients. As will be explained, the Court concludes that the Complaint plausibly alleges that Google's implementation of Dynamic Allocation was anticompetitive conduct in the market for ad exchanges.

The adoption of Dynamic Allocation followed certain changes in how publisher ad servers placed bids on ad exchanges. Prior to 2009, a publisher using Google's DFP ad server would rank in order which ad exchanges would be permitted to bid on an available impression. (Compl't ¶ 268.) Google's DFP ad server would then offer impressions on exchanges in the order selected by the publisher, meaning that if a higher-ranked exchange did not sell the impression, the ad server would move sequentially to the next exchange on the publisher's list. (Id.) This process was called "waterfalling." (Id. ¶ 268.) According to the Complaint, waterfalling gave publishers the freedom to prioritize ad exchanges that had historically yielded better results. (Id. ¶ 269.) As described in the Complaint, waterfalling worked to the relative disadvantage of AdX, as publishers gave AdX a lower ranking because other exchanges obtained higher prices. (Id. ¶ 273.)

The Complaint asserts that beginning in 2009, the industry evolved away from waterfalling, and exchanges began to compete by submitting real-time bids for publisher inventory. (Id. ¶ 270.) The Complaint asserts that as real-time Exchange Bidding gained popularity, Google adopted Dynamic Allocation as a new "decisioning logic" that gave its own AdX exchange a right of first refusal on publisher bids. (Id. ¶ 271.) With Dynamic Allocation, DFP would "permit AdX to peek at the average historical bids from rival exchanges," and then transact the publisher's impression if AdX "return[ed] a live bid for just a penny more" than the highest of those historical bids. (Id. ¶ 271.) The Complaint asserts that AdX was the only exchange with a right of first refusal on publishers' DFP inventory. (Id. ¶ 271.)

The Complaint asserts that Google implemented Dynamic Allocation in order to give AdX an unfair competitive advantage over other exchanges. (Id. ¶¶ 272-73, 275.) Dynamic Allocation calculated the winning AdX bid based on the "historical average prices" on a rival exchange, and did not distinguish a publisher's more valuable impressions from standard, less-valuable impressions. (Id. ¶¶ 274-75.) This had the effect of reducing publishers' yield by shielding AdX from real-time competition. (Id. ¶ 274). The Complaint does not assert that publishers were required to use Dynamic Allocation or automatically enrolled in it: Rather, it states that Google "induced" publishers to use DFP by claiming that it would maximize publishers' inventory yield. (Id. ¶ 278.) The Complaint asserts that internally, Google knew that the optimal arrangement for publishers was to allow real-time bidding across multiple exchanges. (Id. ¶ 279.) The States assert that Dynamic Allocation was exclusionary and anticompetitive in the markets for exchanges and ad-buying tools. (Id. ¶ 281.)

Google urges that the Dynamic Allocation was not anticompetitive because it did

not prevent publishers from transacting with other exchanges, and it allowed AdX to win an impression only if it bid more than rival exchanges. (Google Mem. at 21.) It also argues that Dynamic Allocation should be understood as a "product design[]" that won Google business at the expense of competitors, as opposed to an improper anticompetitive measure. (Reply at 23.)

At the pleading stage, the Court concludes that the Complaint plausibly alleges that Dynamic Allocation was an anticompetitive strategy that could support a section 2 claim in the market for ad exchanges. Google, according to the Complaint, used information obtained through its DFP ad server to determine the average historical value of bids placed on rival exchanges, then used that information to calculate a bid that the publisher would be required to accept and the transaction would occur through AdX, with AdX earning the fee. A rival exchange might have returned a higher bid than the winning bid on AdX, but, as described in the Complaint, rivals of AdX never had the opportunity to receive bids. In essence, Google used information accessible to it through its ad server to wall off exchange competition and guarantee that transactions were made on AdX. For the ad exchange market, Dynamic Allocation therefore had the effect of controlling prices and excluding competition in the ad-exchange market. See United Food & Com. Workers, 11 F.4th at 137. The Complaint therefore plausibly alleges that Dynamic Allocation was anticompetitive in the ad exchange market.

But the Complaint does not plausibly allege that Dynamic Allocation was anticompetitive conduct in the market for publisher ad servers. While the Complaint alleges that Google misled publishers about aspects of Dynamic Allocation, and that publishers received lower bids through Dynamic Allocation than they would have if bids were accepted by multiple exchanges, the Complaint does not plausibly allege that publishers were required to opt into Dynamic Allocation in order to use DFP. (See Compl't ¶ 278 (stating that Google "advertised" and "induced" participation in Dynamic Allocation).) Individual publishers may have been injured because Google falsely or inaccurately described the returns available under DFP, but the Complaint does not describe how Dynamic Allocation resulted in a harm to competition on the ad-server market.

The Complaint also fails to plausibly allege that Dynamic Allocation was anticompetitive conduct in the market for ad-buying tools. The Complaint does not assert that Dynamic Allocation channeled ad purchases specifically through DV360 or Google Ads. It asserts that by routing publisher impressions to AdX, Google foreclosed competition in the market for ad-buying tools, with AdX permitting Google's ad-buying tools to win more than 80 percent of all AdX auctions. (Id. ¶ 276.) The Complaint does not expressly attribute this win rate to Dynamic Allocation. It also does not distinguish the win rates for DV360 and Google Ads, which are alleged to compete in different product markets, and instead aggregates their overall win rate. Google's ad-buying tools may have predominated over their rivals on AdX, but the Complaint does not explain what role, if any, Dynamic Allocation played in that market.

The Court concludes that the States have plausible alleged that Google's use of Dynamic Allocation harmed competition in the ad-exchange market. The States have not plausibly alleged that it was anticompetitive conduct for publisher ad servers and for large- or small-advertiser ad-buying tools. 3. The Complaint Plausibly Alleges that Google's Use of Enhanced Dynamic Allocation Was Anticompetitive Conduct in the Ad Exchange Market.

The Complaint describes a "decisioning logic" called Enhanced Dynamic Allocation ("EDA") that Google implemented through the DFP ad server. The States assert that EDA exploited Google's monopoly in the ad-server market to harm competition in the markets for ad exchanges, publisher ad servers and advertisers' buying tools. (Compl't ¶¶ 282-96.)

According to the Complaint, EDA channeled the most high-value inventory of Google's publisher clients exclusively to AdX, which had the effect of "starving" rival exchanges of scale and liquidity. (Id. ¶¶ 284-85.) The DFP ad server would then allow AdX to transact the impression if an advertiser submitted a bid that was higher than both 1.) a floor price unilaterally set by Google and 2.) the average historical bids of rival exchanges. (Id. ¶ 289.) The Complaint asserts Google automatically enrolled its publisher clients into EDA, and "coaxed" them to continue using EDA by falsely telling them that it maximized their yields. (Id. ¶ 291.) Internally, however, Google stated that EDA allowed AdX to "cherry pick" for itself publishers' high-value impressions. (Id. ¶ 292.)

The Complaint plausibly alleges that EDA caused injury to competition in the ad-exchange market. It asserts that Google used its presence in the ad-server market to channel publishers' most valuable impressions exclusively to AdX. As described in the Complaint, Google automatically enrolled publisher clients into EDA and attempted to mislead publishers to continue using EDA as a way for publishers to maximize yield, when, in truth, Google internally viewed EDA as a way for AdX to "cherry-pick" high-revenue impressions. (Compl't ¶ 292.) As a result, rival ad exchanges were excluded from publishers' high-value inventory. The Complaint plausibly alleges that AdX did not transact this inventory due to a superior product or innovation, but because its role in the market for publisher ad servers allowed it to "g[i]ve itself access to particularly valuable inventory," which it facilitated by automatically opting publisher clients into EDA. (Id. ¶¶ 285, 290-91.)

However, the Complaint does not plausibly allege that EDA caused competitive harm in the markets for publisher ad servers. The Complaint asserts that publishers were automatically enrolled in EDA, and that Google then "coaxed" publishers to continue using EDA by falsely telling them that it would maximize their yields, when, in truth, Google was seeking to "cherry pick" the most valuable ad inventory for its own ad exchange. These allegations may suggest that Google was untruthful with its publisher clients, but they do not describe anticompetitive conduct in the market for publisher ad servers. The Complaint asserts that publishers ultimately had "no choice" but to participate in EDA, but it is not apparent from the Complaint why this would be the case, or why the implementation of EDA would coerce or inhibit a publisher from using a non-Google ad server.

The Complaint also does not plausibly allege that EDA caused competitive harm in the market for ad-buying tools. The Complaint asserts that "[a]dvertisers wishing to purchase the inventory from Google's exchange without losing sight of the user or the value of impressions had to use Google's buying tools." (Id. ¶ 287.) This assertion repackages the States' claims about the sharing of unencrypted user IDs. The Complaint plausibly alleges that implementation of EDA was anticompetitive conduct directed to the ad exchange market but not the markets for ad servers or ad-buying tools.

4. The Complaint Plausibly Alleges that Project Bernanke Was Anticompetitive in the Market for Ad-Buying Tools for Small Advertisers and the Bell Variation Was Anticompetitive in the Ad-Server and Ad-Exchange Markets.

The Complaint describes a Google program called Project Bernanke that it asserts misled display-ad auction participants about how Google organized auctions and distributed the proceeds of winning bids. As described in the Complaint, Project Bernanke would underpay a publisher after a transaction cleared on AdX, with Google secretly retaining a portion of the winning bidder's payment. Google then added this secretly retained payment into a pool that was used to increase the bids of Google's advertiser clients using AdX. As described in the Complaint, the end result was to boost advertisers' bids on AdX, ensuring that the transaction cleared on AdX and not a rival exchange.

Google's purported auction manipulations relate to the use of "second-price auctions" that determined the prices paid by auction winners. Auctions are sometimes structured as "first-price" or "second-price" auctions. (Compl't ¶ 299.) A first-price auction is straightforward: the advertiser-bidder pays the amount of its own winning bid, which is the highest bid among all bids. (Id. ¶ 299.) In a second-price auction, the winning advertiser-bidder is the bidder who bid the highest price, but, instead of paying the amount of its bid, it pays the amount of the second-highest bid. (Id. ¶ 299.) For instance, if the two highest bids are $15 and $12, the advertiser with the $15 bid will win, but will pay only $12. (Id. ¶ 301.) Of course, bidders remain incentivized to submit the highest bid, but they do not know the precise amount that they will pay.

In a second-price auction, publishers also are permitted to set a price floor, reflecting the minimum amount that they will accept for a transaction. (Id. ¶ 302.) If only the highest bid exceeds the price floor, the price floor acts as the second-highest bid, and the winner will pay an amount equal to the price floor. (Id. ¶ 302.) A publisher using DFP can set different price floors on different ad exchanges. (Id. ¶ 302.)

According to the Complaint, from 2010 to 2019, Google publicly stated that AdX was organized as a second-price auction. (Id. ¶ 300.) The Complaint asserts, however, that Google had a secret program called "Project Bernanke,"²³ which was structured as a third-price auction rather than a second-price auction. (Id. ¶¶ 299, 303.) According to the Complaint, publishers were unwittingly paid in amounts that reflected third-place bids rather than second-place ones, and, consequently, suffered revenue declines of as much as 40 percent. (Id. ¶¶ 303-05.) At the same time, Google Ads, its ad-buying tool for small advertisers, continued to charge the winning advertiser as if it had won a second-price auction. (Id. ¶ 306.) Google retained the difference between the second- and third-place bids. (Id. ¶ 306.) In other words, under Project Bernanke, the advertiser with the winning bid paid the price of the second-highest bid, but the publisher would receive a payout equal to the third-highest

23 The program was named for former Federal Reserve Chairman Benjamin Bernanke, and an internal Google document described the project as "Quantitative Easing on the AdExchange." (Compl't ¶¶ 298, 317.)

bid, with Google retaining the difference. (Id. ¶ 306.)

Google then placed the price differential into a pool, which it used to increase the bids of client advertisers using Google Ads in order to help those advertisers win impressions on AdX that might have gone to advertisers that used non-Google ad-buying tools. (Id. ¶¶ 306, 308.) The Complaint asserts that internal Google documents stated that prior to Bernanke's implementation, non-Google ad-buying tools "were winning too often" on AdX. (Id. ¶ 308.) The extra money pooled through Bernanke increased the number of impressions transacted in AdX, and also caused a 20% increase in the win rate of Google ad-buying tools. (Id. ¶ 317.)

Google later implemented two other iterations of Bernanke. (Id. ¶¶ 309-11.) In a version called Bell, Google pre-determined whether a publisher provided AdX with an opportunity to bid on inventory prior to other exchanges, such as permitting Google to sell impressions using Dynamic Allocation or EDA. (Id. ¶ 311.) If the publisher did not give preferential access, the Bell version of Project Bernanke structured the auction from a second- to third-price auction, thereby decreasing the publisher's AdX revenue. (Id. ¶ 311.) Bell then used the differential to inflate the bids returned to publishers who gave preferential access to AdX. (Id. ¶ 311.) In other words, Bell penalized publishers who did not grant AdX preferential access by paying them based upon the third-place bid rather than a second-place bid, while using the difference to increase the bids made to publishers who allowed preferential access. (Id. ¶ 311.)

A version of Project Bernanke called Global Bernanke used pool money to inflate the bids of "Google Ads advertisers who would likely have otherwise lost for being too close to a publisher-set price floor on AdX." (Id. ¶ 310.) "Global Bernanke" was different because it did not just apply to publisher-set price floors, but to floors that Google itself had set. (Id. ¶ 310.)

The Complaint asserts that Project Bernanke increased annual AdX revenue by $230 million, with Bell alone generating an additional $140 million. (Id. ¶ 317.)

The Court reviews the markets in which Project Bernanke, including Global Bernanke and the Bell variation, are alleged to have harmed competition.

The Complaint's allegations about the Bell iteration of Project Bernanke plausibly alleges harm to competition in the ad-server market. As described in the Complaint, if a publisher did not grant AdX the chance to bid on inventory prior to other exchanges, Google dropped the publisher's auction from a second-price auction to a third-price auction. (Id. ¶ 311.) Bell then used the payment difference between the second- and third-place bids to inflate the bids to publishers who allowed preferential access. (Id. ¶ 311.) Thus, a publisher who granted special priority to AdX would receive higher revenues from AdX and a publisher who refused to do so received lower revenues. Accepting the Complaint's allegations as true, Google coercively used its power in the ad-server market to reward publishers that granted it a special priority and punish publishers that did not. The Court concludes that Complaint plausibly alleges that the Bell initiative was anticompetitive conduct causing harm to competition in the ad-server market.

The Court further concludes that the Complaint does not plausibly allege that other aspects of Bernanke were anticompetitive as to the market for publisher ad servers. The States assert that publishers were misled into believing that Google

ran a second-price auction, when, from the perspective of the publishers, they were paid in amounts that would have reflected the results of a third-price auction. These allegations may describe misleading statements and the underpayment of publishers, but they do not explain how such conduct advanced or maintained Google's ad-server monopoly. The Court concludes that only Bell and not other aspects of Project Bernanke harmed competition in the market for publisher ad servers.

The Complaint plausibly alleges that Project Bernanke, including Global Bernanke, was anticompetitive as to the market for ad-buying tools used by small advertisers. As alleged in the Complaint, Google implemented Bernanke after it observed that non-Google tools were outbidding Google Ads on AdX. (Id. ¶ 308.) Through Bernanke, Google pooled funds to increase advertiser bids made on AdX, thus clearing the transaction on AdX through Google Ads. Google Ads won these auctions because Google could access bid information through the publisher ad server, and then inflate an advertiser's bid by drawing from funds that Google had pooled. These allegations describe an anticompetitive effort by Google to manipulate ad auctions in order to give an unfair advantage to Google Ads and thereby injure competition among ad-buying tools for small advertisers. Rival ad-buying tools had no effective way of competing with Bernanke, and apparently did not even know that Bernanke was occurring.

The Complaint therefore plausibly alleges that Project Bernanke was anticompetitive conduct harming competition in the market for ad-buying tools used by small advertisers.

Regarding the ad-exchange market, the Complaint asserts that by pooling money to inflate advertisers' bids on AdX, Google permitted "AdX to cream-skim, i.e., transact publishers' most valuable impressions while leaving mainly low-value impressions for rival exchanges." (Id. ¶ 308.) The Bell variation also allegedly inflated bids returned to publishers who gave AdX preferential access. (Id. ¶ 311.) Rival exchanges were allegedly left with fewer, lower-value impressions. (Id. ¶ 316.) As described in the Complaint, this was not a consequence of a superior product, but of auction manipulations that channeled high-value impressions to AdX and thereby deprived other exchanges of the ability to successfully offer these impressions to their users.

The Complaint plausibly alleges that Project Bernanke, including Global Bernanke, and the Bell variation were anticompetitive measures that harmed competition in the ad-exchange market by using AdX to selectively transact publishers' higher-value impressions and leaving lower-value impressions for competing exchanges.

5. The Complaint Plausibly Alleges that Dynamic Revenue Sharing Was Anticompetitive Conduct that Harmed Competition in the Ad-Exchange Market.

In 2014, Google launched a program called Dynamic Revenue Sharing ("DRS"). (Compl't ¶ 318.) As described in the Complaint, DRS manipulated the fee that AdX charged publishers in ways that were tailored to the bid floors set by publishers. (Id. ¶¶ 318-19.) The Complaint explains that in "a true second-price auction," AdX could transact an impression only if a bid cleared the publisher's pre-set floor after accounting for Google's exchange fee. (Id. ¶ 319.) For instance, if a publisher set a $10 bid floor, a bid would clear only if the amount of the bid, minus Google's exchange fee, exceeded $10. (Id.) A $12 bid with a 20% exchange fee would net $9.60 to a publisher, thus failing to clear the $10 floor. (Id.) The Complaint alleges that Dynamic Revenue Sharing manipulated Google's exchange fee after soliciting auction bids and "peeking" at bids on rival exchanges. (Id. ¶ 320.) AdX would then make a downward adjustment to its own exchange fee in order to clear the publisher's bid floor. (Id. ¶ 320.) For instance, in the above-discussed example, instead of charging a 20% fee on a $12 bid, Google would charge a lower fee in order to clear the publisher's $10 floor. At the same time, DRS would secretly increase AdX's publisher fees on impressions where an advertiser bid significantly above the publisher's floor. (Id. ¶ 322.) The Complaint alleges that because of Google's monopoly power in the market for publisher ad servers, it had a unique ability to know the bids placed on rival exchanges. (Id. ¶ 321.)

The Complaint alleges that DRS harmed competition in the ad-exchange market by manipulating floors after bids were received and after "peeking" at the bids of rival exchanges. (Id. ¶ 330.) In the ad-exchange market, only AdX had the ability to determine its fee with knowledge of the bids placed on rival exchanges, allowing it to win impressions that it may have lost to other exchanges. (Id.) The Complaint asserts that DRS earned AdX an additional $250 million per year in transactions. (Id.)

The Complaint does not describe anticompetitive conduct in the market for publisher ad servers, and instead describes how publishers were misled about the implementation of Dynamic Revenue Sharing. As described in the Complaint, DRS caused certain publishers to run lower-quality advertisements than they might have run if the bid had cleared on a rival exchange, asserting that "[o]n a quality-adjusted basis, DRS harmed publishers." (Compl't ¶ 325.) As an example of lower-quality ads, the Complaint cites hypothetically to an entity advertising fake N95 masks. (Id.) The interrelationship between an ad for a fake N95 mask and lower bids for impressions is neither intuitive nor obvious.²⁴ It asserts that Google did not disclose to publishers that it adjusted the exchange fees only after "peeking" at bids placed on other exchanges, and that Google was aware that DRS was not increasing publisher ad yields. (Id. ¶¶ 328-29.) These allegations describe how Google may have misled publishers about how DRS was implemented, but it does not describe conduct that harmed competition in the ad-server market. While the conduct may support a claim for violation of a deceptive-practice statute or may amount to a breach of contract, the Complaint does not plausibly describe how such conduct harmed competition.

24 N95 masks may be a lower-price, higher-volume product than a luxury good directed to a narrow segment of the market. Rational considerations may cause the seller of masks to place a higher bid than the seller of luxury goods. One product might be of greater prestige than the other, but all are susceptible to false claims that require some degree of policing by publishers.

The Complaint asserts that Dynamic Revenue Sharing harmed advertisers because they were misled into believing that AdX functioned as a second-price auction, when in truth Google manipulated publishers' ad floors after the fact. (Id. ¶ 323.) In essence, it asserts that advertisers overpaid as a result of Google's manipulation of publishers' price floors. (Id. ¶ 323.) Such allegations may describe an injury to advertisers who overpaid for impressions and were misled by Google about the nature of its auctions, but it does not describe an injury to competition in the market for ad-buying tools for either large or small advertisers. The Complaint plausibly alleges that by adjusting its fees only after receiving bids and reviewing the bids placed on rival exchanges, Google harmed competition in the market for ad exchanges, including competition on exchanges' take rates. (Id. ¶ 330.) Based on information uniquely available to it through its ad-server monopoly, Google had the ability to alter bids in order to transact impressions it would have lost to rival exchanges. (Id. ¶ 330.) Accepting the Complaint's allegations as true, they allege anticompetitive conduct that permitted AdX to win bids based on price manipulations by Google, as opposed to a superior product or some other legitimate business factor. This had the effect of advancing or maintaining Google's monopoly in the ad-exchange market. The Complaint plausibly alleges that that Google's Dynamic Revenue Sharing harmed competition in the ad-exchange market.

6. The Complaint Does Not Plausibly Allege that Reserve Price Optimization Was Anticompetitive Conduct.

According to the Complaint, in 2015, Google implemented a program called Reserve Price Optimization ("RPO") that overrode and increased the bidding floors set by publishers and thereby deceptively increased the amount that advertisers paid for impressions on AdX. (Compl't ¶ 331.) As described in the Complaint, Google publicly touted the benefits of selling ads through a sealed, second-price auction, explaining that it incentivizes buyers to bid what they are willing to pay without encouraging them to "game the system." (Id. ¶¶ 332-34.) Under RPO, however, Google overrode the price floors that publishers set for AdX, and adjusted those floors based on the historical bids of advertisers. (Id. ¶¶ 335-37.)

The Complaint uses an example where a publisher sets a $10 price floor for bids made on AdX, in which three advertisers submitted bids of $15, $12 and $11. (Compl't ¶ 336.) In the next auction for the same targeted impression, Reserve Price Optimization — aware of those bids from the prior auction — would override the publisher's $10 floor, and set customized price floors of $14.90, $11.90 and $10.90, respectively, for each of those three prior bidders. (Id. ¶ 336.) With these new price floors, a winning bidder who bid $15 would not pay the second price for the impression, and instead would pay the $14.90 price floor — what the Complaint characterizes as "the artificial and manipulated bid of the RPO floor." (Id. ¶ 336.) The Complaint alleges that by falsely telling advertisers that AdX ran a second-price auction, Google induced advertisers to place true-value bids, only to use those bids against them. (Id. ¶ 338.)

The Complaint asserts that Google automatically opted publishers into Reserve Price Optimization in 2015 without their knowledge or consent. (Id. ¶ 340.) Around the same time, Google publicly denied that it would implement dynamic floors on AdX. (Id. ¶ 341.) In May 2016, Google publicly stated that it would launch "optimized pricing" but did not disclose that it had already launched RPO. (Id. ¶ 343.) Google did not publicly state that it would run a first-price auction until 2019. (Id. ¶ 346.) The Complaint asserts that RPO affected "billions of impressions." (Id. ¶ 349.)

The Complaint's allegation of competitive harm is opaque and theoretical. For example, the Complaint alleges that "[o]n the surface, RPO appeared to increase yield because AdX initially returned higher bids. However, because RPO relied on inside information, combining bid data from AdX with publishers' ad server user IDs, it

exacerbated problems of adverse selection in publishers' inventory auctions.²⁵ The states assert that markets rife with problems of adverse selection are inefficient, dissuade participants from entering, and result in lower output. As a result, RPO ultimately forecloses competition from exchanges and advertisers and reduces inventory yield. Publishers could not discover this harmful effect because Google failed to disclose RPO's reliance on inside information." (Id. ¶ 344.)

25 The Court understands "adverse selection" in this context to mean the asymmetry of information between publishers and AdX.

The Court examines each potentially affected market.

The Complaint does not plausibly allege that Reserve Price Optimization was an anticompetitive scheme directed to the market for publisher ad servers. It asserts that RPO harmed publishers by overriding the floor settings they set for AdX, misleading them into concluding that they could control their own price floors and deceiving them about the nature of the second-price auction run on AdX. (Id. ¶¶ 339, 342, 346.) These allegations describe a practice that misled Google's publisher clients, but they do not describe anticompetitive conduct in the market for ad servers. The antitrust laws generally do not provide that a market participant's failure to be truthful about its own product deprives its head-to-head competitor of competition, and the Complaint does not attempt to make out such a claim. See Nat'l Ass'n of Pharm. Mfrs., Inc. v. Ayerst Labs., 850 F.2d 904, 916 (2d Cir. 1988) (discussing pleading requirements to make out a monopolization claim based on misleading advertising). Indeed, the Complaint expressly alleges that RPO resulted in higher payments to publishers: it uses a hypothetical example where RPO causes a publisher to receive $14.90 as a result of RPO's manipulation of price floors, as opposed to $12 in a sealed second-price auction. (Compl't ¶ 336.) The Complaint does not describe competitive harm from Reserve Price Optimization in the market for publisher ad servers.

The Complaint does not plausibly allege that RPO amounted to anticompetitive conduct in the markets for ad-buying tools for large or small advertisers. Accepted as true, the Complaint's allegations describe how Google exploited advertisers' understanding that Google was running a sealed, second-price auction by using advertisers' bids against them in order to manipulate publishers' price floors. (Id. ¶¶ 334, 335.) The Complaint asserts that RPO harmed advertisers by "forcing them to pay more than Google advertised" and falsely representing that AdX ran a second-price auction. (Id. ¶ 338, 345-46.) But again, this describes how advertisers may have been misled about how Google organized and operated ad auctions, rather than conduct in the markets for ad-buying tools for large and small advertisers that harms competition in those markets.

The Complaint does not plausibly allege that Reserve Price Optimization was anticompetitive conduct in the ad-exchange market. The Complaint describes how Google used advertisers' past bids on AdX to unilaterally override publishers' price floors on the exchange. (Id. ¶¶ 336-37.) The Complaint mentions that RPO "[e]ventually" used competing exchanges' bids as part of RPO, but it does not explain to what effect. (Id. ¶ 337.) In a conclusory fashion, the Complaint asserts that "RPO was and is successful in excluding competition in the exchange market" and affected "billions of impressions" transacted on AdX.

(Id. ¶ 349.)²⁶ It also asserts that RPO brought $250 million in annual revenue to Google. (Id. ¶ 349.)

26 The Complaint tends to lump in a single section its allegations of competitive harm ("Anticompetitive Effects") in a particular market from all conduct alleged to be anticompetitive conduct. See, e.g. Section VIII.B (Compl't ¶¶ 502-542). While this may be somewhat understandable because of the perceived synergy of discrete actions, at times it makes it more difficult for a Court to analyze whether particular conduct of Google is plausibly alleged to be anticompetitive.

These allegations do not plausibly allege harm to competition in the ad-exchange market. The Complaint describes RPO as a project that secretly manipulated the price floors set by publishers, resulting in unwitting overpayments by advertisers. Unlike some other theories of liability, the Complaint does not describe an effort that undermined exchange-market competition by secretly "peeking" at rival exchanges in order to inflate bids placed on AdX for the purpose of directing transactions away from Google competitors. Accepting the truth of the Complaint's allegations, RPO describes a secretive process that misled publishers and advertisers and increased publishers' price floors, but it does not identify the anticompetitive conduct set forth in the States' other auction-manipulation allegations. The Complaint does not plausibly allege that RPO was anticompetitive conduct directed to the market for ad exchanges.

7. The Complaint Does Not Plausibly Allege that the Challenged Aspects of Exchange Bidding Were Anticompetitive in Any Market.

The States contend that Google has pursued a multi-front effort to weaken the competition that arose following the industrywide move in the direction of header bidding. To briefly review, header bidding enabled publishers, including those who used Google's DFP ad server, to solicit live, competitive bids from multiple exchanges. (Id. ¶ 353.) Google perceived it as a serious threat to the company's dominant role in digital advertising. In addition to claims centered on the Facebook NBA, the Complaint asserts that Google's implementation of its competing bid-routing initiative called Exchange Bidding was anticompetitive conduct, and that Google took a series of targeted measures designed to "quash" or "kill" header bidding.

The Complaint asserts that Google was motivated to "quash" or "kill" header bidding in order to maintain its dominant positions in the markets for ad-buying tools and publisher ad servers, and to avoid price competition on the exchange market. (Id. ¶¶ 352, 361-65.) Hence, in April 2016, Google launched its Exchange Bidding (or, internally, "Jedi") program in order to maintain Google's ad-exchange monopoly and to constrain exchange competition.²⁷ (Id. ¶ 366-367.) Google touted Exchange Bidding as an innovation that benefited publishers because now publishers using Google's ad server for publishers — DFP — could offer impressions to non-Google ad exchanges and these exchanges could participate in Exchange Bidding. A publisher using Google's DFP ad server who elected to participate in Exchange Bidding was required to route all of its impressions through AdX. Exchange Bidding permitted publishers utilizing the DFP ad server to route inventory to multiple ad exchanges at one time, in order to mimic the live, multi-exchange

27 The Star Wars reference apparently arose from an internal discussion about the "need to create a jedi mind trick" in order "to get publishers to come up with the idea to remove exchanges from PreBid on their own." (Compl't ¶¶ 385-86.)

competition provided by header bidding. (Id. ¶ 366.)

According to the Complaint (and as explained previously) DFP created a unique ID for users the publisher's website, and that ID was seen only by participants in Google's AdX. This meant that advertisers using a non-Google ad exchange could still bid on an impression originating with a publisher using DFP, but the bidder on a non-Google ad exchange did not know whether it was bidding on one single user who visited two different publisher's websites or two unique users. Not having that information and risking the possibility of a double buy of the same user caused bidders on non-Google ad exchanges, according to the Complaint, to bid less for an impression. (Id. ¶¶ 366-68.) The Complaint also asserts that the DFP ad server charged a non-Google exchange a 5% fee for the impressions that it cleared, while also requiring any publisher who signed up for Exchange Bidding to route their impressions through AdX. (Id. ¶¶ 369-70.) All the while, Google continued to access information about bids placed on rival exchanges, allowing it to continue certain alleged auction-manipulation practices. (Id. ¶ 371.) The Complaint asserts that this allowed AdX to win an auction even if a higher bid was offered on a rival exchange. (Id. ¶ 371.)

The Complaint asserts that, through Exchange Bidding, Google sought to exclude competition in header bidding by giving information advantages to those exchanges that participated in Exchange Bidding, allowing them to trade ahead of bids submitted by header-bidding exchanges. (Id. ¶¶ 376-78.) Google permitted these participating exchanges to take a "last look" at the highest bids placed through header bidding, and to clear the impression by bidding a penny more than the bids placed with header bidding. (Id. ¶¶ 377-78.) Google accessed this information through its DFP ad server. (Id. ¶ 377.)

In 2019, Google changed course in how it operated Exchange Bidding. Google publicly announced that Exchange Bidding participants would no longer be able to trade ahead of header-bidding operations and that all "buyers will compete in the same unified auction ...." (Id. ¶ 379.) But, according to the Complaint, Google continued to give information advantages to exchanges that participated in Exchange Bidding, including "sensitive pricing information derived from publishers' sensitive clearing auction records...." (Id. ¶ 380.) The Complaint asserts that this information allowed Exchange Bidding participants to continue trading ahead of other bidders. (Id.) Google also used data to predict with precision the bid required for an exchange to win an impression. (Id. ¶ 381.)

The Complaint alleges that Exchange Bidding was anticompetitive because it advantaged Google's AdX exchange. (Id. ¶¶ 529(d), 534(d).) But the Complaint describes Exchange Bidding as a voluntary venture, one that arose as a response to the popularity and innovation of header bidding. Exchange Bidding allowed non-Google exchanges — and indirectly those advertisers submitting bids through those exchanges — to participate in auctions, a move that tended to increase competition for publisher ad inventory. If a non-Google exchange, which presumably would be sophisticated in the nature and operation of an ad exchanges, chose to participate in Exchange Bidding, this benefitted publishers, the non-Google ad exchange and the users of the non-Google exchange. If Google inadequately or deceptively described its pricing or any part of its Exchange Bidding process, that may be actionable under a State deceptive practice law; without a plausible explanation of how

it harmed competition, it is not actionable under section 2.

"Even a monopolist ... must generally be responsive to the demands of customers, for if it persistently markets unappealing goods it will invite a loss of sales and an increase of competition." Berkey Photo, Inc. v. Eastman Kodak Co., 603 F.2d 263, 287 (2d Cir. 1979). As described in the Complaint, Exchange Bidding had traits that may be both more and less appealing to customers than header bidding. Among Exchange Bidding's drawbacks, Google charged non-Google exchanges a 5% fee for impressions cleared through Exchange Bidding, and, because of Google's encryption of user IDs, non-Google exchanges could not identify users with the ease facilitated by header bidding. Google's more appealing offerings may have included the sharing of "sensitive pricing information" that allowed exchanges that participated in Exchange Bidding to trade ahead of bids made through header bidding. These allegations describe both drawbacks and benefits of importance to a publisher or exchange on deciding whether to participate in voluntary Exchange Bidding.

The Complaint does not assert that any participant was improperly compelled to participate in Exchange Bidding. (See, e.g., Compl't ¶ 370) ("A publisher signing up for Exchange Bidding can select which non-Google exchanges to route their impressions to, but the publisher must route their impressions to AdX.").) It also does not allege any details about how Exchange Bidding was marketed to publishers and exchanges, or the terms of any agreements that govern their participation in Exchange Bidding.

The Complaint repeatedly characterizes Exchange Bidding as an attempt "to kill header bidding." (Id. ¶¶ 25, 26, 352, 374, 413, 420, 505, 509, 512.) But it does not describe the anticompetitive effect, if any, that Exchange Bidding has had on the popularity or profitability of header bidding. The Complaint's allegations regarding the impact on header bidding is conclusory: "Exchange Bidding is exclusionary and successfully forecloses competition from header bidding and in the exchange market (Section VIII.B)." (Id. ¶ 372) The cross-referenced section, Section VIII.B, is similarly conclusory and repeats in similar verbiage the same claim: "This scheme was successful and substantially suppressed the adoption and growth of header bidding while at the same time causing its AdX Exchange to continue gaining market share." (Id. ¶ 516.)

As to the assertion that Google charges publishers a 5% fee when a non-Google exchange clears an impression, "[p]rices not based on superior efficiency do not injure competitors, but rather invite competitive entry." U.S. Football League, 842 F.2d at 1361 (internal citation omitted). "Indeed, although a monopolist may be expected to charge a somewhat higher price than would prevail in a competitive market, there is probably no better way for it to guarantee that its dominance will be challenged than by greedily extracting the highest price it can." Id. In the Complaint's telling, Exchange Bidding is an inferior alternative to header bidding, but participants in the online ad market apparently remain free to participate in either, or both.

The Complaint does not plausibly allege that Exchange Bidding harmed competition in any market.

8. The Complaint Plausibly Alleges that Google's Redaction of Auction Data and Limitations on Publisher Line Items Was Anticompetitive Conduct in the Exchange Market and Ad Server Market.

The Complaint asserts that Google undertook additional, coordinated steps for

the purpose of impeding or preventing publishers from participating in header bidding. The States assert that Google used its monopoly power in the publisher ad-server market to withhold bid information from publishers and render header bidding less effective than bids submitted through Exchange Bidding.

In 2018, the DFP ad server began to redact auction records that showed the relative success of header bidding compared to the performance of Exchange Bidding. (Compl't ¶¶ 387-88.) The Complaint asserts that publishers relied on data fields called KeyPart and TimeUsec2 in order to compare the relative performance of exchanges in header bidding and Exchange Bidding, and to adjust their use according. (Id. ¶ 387.) By redacting the two data fields, Google allegedly prevented publishers from measuring the performance of different exchanges and foreclosed the competition brought by header bidding. (Id.) The Complaint also asserts that Google "splits" data in a way that makes it impossible for publishers to track auction results and limits information about the bids submitted for an impression. (Id. ¶ 388.) This allegedly left publishers unable to track the source of winning impressions, and even to see whether the highest bidder won at auction. (Id.) These actions appear not to have any legitimate business purpose or benefit to Google other than harming competition from header bidding.

The Complaint also asserts that the DFP ad server limits publishers' ability to receive bids through header bidding. (Id. ¶¶ 389-94.) Through line items, publishers are able to set a to-the-penny price for bids that they will accept through header bidding. (Id. ¶ 390.) For instance, a publisher can list a line item with the price of $4.29. (Id.) If, instead, the publisher has entered a line item of $4.20, and received a bid of $4.29, DFP would round down the $4.29 bid to the nearest line item, and the publisher would be paid $4.20. (Id.) Accordingly, publishers must create separate line items ($4.20, $4.21, $4.22 and so on) to precisely capture a competitive bid. (Id.) The Complaint asserts that Google purposely limits the number of line items available to publishers in order to foreclose competition from header bidding. (Id. ¶ 391.) By contrast, a rival ad server named OpenX allowed publishers to capture bids on header bidding through a single line item. (Id. ¶ 393.) The Complaint asserts that OpenX was unable to compete with Google's monopoly power and exited the ad server market in 2019. (Id. ¶ 393.) Google has internally described its caps on the number of line items as a "tool" to fight header bidding and "push[]" publishers toward Exchange Bidding. (Id. ¶ 391.) The Complaint asserts that the limit on publishers' line items has driven down publishers' yields and made bids from header-bidding exchanges less competitive than those made through AdX. (Id. ¶ 394.)

Accepting these allegations as true, the Complaint plausibly alleges that Google used its monopoly power in the ad-server market for the purpose of impairing publisher participation in header bidding. Google's redaction of auction data prevented publishers from assessing the relative performance of header bidding versus Exchange Bidding. (Id. ¶ 387.) Google began to restrict this data in 2018, only after facing competition from header bidding. Accepting these allegations as true, the Complaint does not describe activity in the nature of a product innovation or a mere refusal to share information with a competitor, but a measure that thwarts the ability of publisher clients to assess the relative performance of auction results on Exchange Bidding and header bidding. The Complaint plausibly alleges that this was anticompetitive conduct that harmed

competition in the exchange market. It also describes anticompetitive conduct in the ad-server market because Google's data redaction plausibly resulted in depressed prices to publisher clients who were restrained from tailoring bids specifically to header bidding.

Similar reasoning applies to Google's limitation on the number of line items available to publishers for use on header bidding. Google urges that the limitation cannot plausibly be considered anticompetitive conduct because the Complaint describes it as an "existing" approach, as opposed to a coercive policy change. (Google Mem. at 23-24.) Google also urges that it has no obligation to design its products in ways that would improve the competitiveness of header bidding. (Id.) But the Complaint alleges that Google internally viewed these caps as a "tool" against header bidding and a way to direct more transactions toward Exchange Bidding. The Complaint plausibly alleges that the limitation on line items was an anticompetitive measure that constrained publishers' participation in header bidding, while also constraining competition on the exchange market.

9. The Complaint Plausibly Alleges that Projects Poirot and Elmo Were Anticompetitive Actions in the Ad-Exchange Market and the Market for Ad-Buying Tools of Large Advertisers.

The Complaint alleges that Google used its ad-buying tool for large advertisers, DV360 — also known as a "DSP" or demand side platform — to identify which rival exchanges were likely participating in header bidding, and then punish those exchanges by directing ad buys away from those exchanges to Google's own AdX exchange. Google dubbed these initiatives "Poirot" and "Elmo." DV360 is one of several viable competitors in the market for ad-buying tools for large advertisers, though the Complaint describes DV360 as the "largest" DSP. (Compl't ¶¶ 71, 73.) DSPs offer more complex options for sophisticated clients that are supervised by specialized ad-buying teams assembled by advertisers. (Id. ¶ 73.) The Complaint alleges that Poirot and Elmo were anticompetitive actions supporting an attempt-to-monopolize claim in the market for ad-buying tools for large advertisers.

The Complaint asserts that Google has attempted to use DV360 to undermine header bidding and retain its inventory advantage in the ad-exchange market. The Complaint asserts that header bidding disrupted AdX's "captive supply" of publisher inventory, and that because of header bidding's popularity, DV360 was "forced" to participate in rival exchanges, or else it would lose ad spend and market share to rival ad-buying tools. (Id. ¶¶ 396-97.) Google allegedly perceived the need for DV360 to transact on multiple exchanges as a threat to Google's control over online ad inventory and advertiser spending. (Id. ¶ 398.)

As described in the Complaint, DV360 implemented a program called "Poirot," which used an algorithm to detect whether other exchanges were deviating from running a true second-price auction and were instead running a "dirty" auction. (Id. ¶ 400.) DV360 adjusted its bids on those exchanges, with the effect of directing transactions to AdX — which, itself, was not running a true second-price auction. (Id. ¶ 400.) According to the Complaint, DV360 suffered a -1.9% revenue drop as a result of the Poirot algorithm. (Id. ¶ 401.) Google then expanded Poirot to "optimize" its bidding in first-price auctions, such as those used in header-bidding exchanges. (Id. ¶ 402.) Internal Google documents stated that, following this expansion of Poirot, large ad buyers utilizing DV360 were

spending 7% more on AdX and had reduced their spending on most other exchanges. (Id.)

Google launched a separate project called "Elmo," under which DV360 could discern whether a bid request had been made across multiple exchanges, thereby indicating that the bid had been placed through header bidding. (Id. ¶ 403.) Under Elmo, DV360 decreased ad spending on exchanges that it suspected of engaging in header bidding. (Id.) By March 2018, Elmo decreased DV360 ad spending through header bidding by 25% while adding approximately $220 million in spending to AdX. (Id. ¶ 404.)

The Complaint asserts that Poirot and Elmo successfully reduced spending on rival exchanges and "starve[d]" them of the primary source of demand. (Id. ¶ 405.) It asserts that a Google employee concluded that the combined effect of Poirot and Elmo caused an average 21% revenue decrease on affected exchanges and a $300 million increase on AdX. (Id.)

The Complaint plausibly alleges that Poirot and Elmo harmed competition in the ad-exchange market. It describes how Google used DV360 to obtain information about rival exchanges and direct spending away from rival exchanges and toward AdX. On the closer question of the harm to competition in the market for ad tools for large advertisers, the Complaint plausibly alleges that Google "locked advertisers into using DV360" and directed ad spend to AdX instead of rival exchanges. (Compl't ¶¶ 395-400.) That DV360 saw an initial negative revenue impact of 1.9% under Poirot supports the conclusion that Google's conduct was anticompetitive activity undertaken for the purpose of harming rivals. (Compl't ¶ 401.) As with many of the Court's rulings on the surviving claims, the record may look very different at the summary judgment or trial stage.

10. The Complaint Does Not Plausibly Allege Anticompetitive Conduct Relating to Mobile Web Page Development.

Google created Accelerated Mobile Pages ("AMP") as a way to develop mobile webpages. (Compl't ¶ 407.) The Complaint asserts that Google manipulated search-engine traffic on mobile webpages in order to "strongarm" publishers not to use header bidding. (Id. ¶¶ 406-12.) The Complaint asserts that AMP was designed to be fully compatible with Google's DFP ad server, but that it frustrated publishers' use of header bidding by making JavaScript incompatible with header bidding. (Id. ¶ 407.) After publishers adopted a JavaScript workaround using remote.html, Google further restricted AMP code to prohibit publishers from routing bids and sharing user data with "more than a few exchanges at a time." (Id.) However, AMP remained compatible with Google's DFP ad server, which allowed Google to favor AdX and exclude rival exchanges. (Id.) According to the Complaint, publishers who did not use AMP ranked lower in Google's search engine results, causing them to lose site traffic and ad revenue. (Id. ¶¶ 408-09.)

The Complaint is both sweeping and vague. The Complaint makes the flat-out statement that "[p]ublishers that did not adopt AMP would see the traffic to their site drop precipitously from Google suppressing their ranking in search and re-directing traffic to AMP-compatible publishers." (Id. ¶ 408.) The statement in context does not enable the reader to definitively discern its meaning. If true, it would mean that all publishers — or at least all publishers using DFP who did not adopt AMP — would have all or some Google search rankings suppressed. According to the Complaint, Google has described AMP as an open-source collaboration, but it is actually a Google-controlled initiative, and Google controls AMP's governing board. (Id. ¶ 410.) The Complaint asserts that Google has falsely portrayed AMP as an effort to improve page loading, but that it internally views AMP as a measure to combat header bidding. (Id. ¶ 411.)

The Complaint does not plausible allege AMP to be an anticompetitive strategy. The assertion that AMP limited bids to "a few exchanges at a time" suggests that AMP is — at least to some extent — compatible with header bidding. The Complaint does not address whether providing access to all sites using header bidding would affect the loading time of an AMP-enabled app.

The Complaint does not allege that Google had monopoly power over the platforms or software used to develop mobile web pages, and Google's activity in that product market is far afield from the claims raised in the Complaint. To the extent that Google's search engine purportedly lowered the search rankings for pages that were not developed with AMP, that allegation is remote from the product markets and claims in this case. The Complaint makes no allegations about the mechanics of Google's search engine and how it ranks results for non-AMP enabled sites. It does not provide details that plausibly explain why or how search results would direct traffic away from sites that facilitated header bidding, as opposed to some other reason.

11. The Claim Directed to Google's Proposed Privacy Sandbox Is Not Ripe for Adjudication.

The Complaint describes a purported plan to implement a Privacy Sandbox that would block rivals from using cookies to track users on Google's popular web browser, Chrome. (Id. ¶¶ 473-81, 534(f).) Because the plan is contingent and hypothetical as of this Opinion and Order, it would be advisory for this Court to opine on whether the plan would be anticompetitive.

"To be justiciable, a cause of action must be ripe — it must present a real, substantial controversy, not a mere hypothetical question." Kurtz v. Verizon New York, Inc., 758 F.3d 506, 511 (2d Cir. 2014) (quotation marks omitted). "A claim is not ripe if it depends upon contingent future events that may or may not occur as anticipated, or indeed may not occur at all. The doctrine's major purpose is to prevent the courts, through avoidance of premature adjudication, from entangling themselves in abstract disagreements." Id.

The Complaint asserts that Google intends to modify its popular Chrome browser in order to create a "walled garden" that allows only Google to track and identify users, leading other participants in the online advertising marketplace to rely more heavily on Google's products. (Compl't ¶¶ 470-501.) Briefly summarized, under the so-called Privacy Sandbox initiative, the Chrome browser would block the use of cookies that track and identify web users, under the false pretext that this is a privacy-enhancement measure. (Id.) At the same time, Google itself would continue to track users with a high degree of sophistication, based on user activities in the Chrome browser and various Google properties, such as Gmail, YouTube and the Google search engine. (Id. ¶¶ 474-78.) By blocking the cookies used by competitors and continuing to gather user data for its own purposes, the Privacy Sandbox allegedly would preclude rival ad-buying tools and exchanges from identifying users on Chrome while amassing "much richer" data for Google. (Id. ¶¶ 474-76.) These

changes would purportedly be enacted "[b]y the end of 2022 ...." (Id. ¶ 474.) The Complaint asserts that Google previously considered, then abandoned, a publisher-centered effort called Project NERA, and that Google has a long history of violating user privacy and lobbying against privacy-centered regulations and legislation. (Id. ¶¶ 471-72, 480-81.)

As described in the Complaint, Project Sandbox is a collection of proposed actions that Google may or may not implement. Google previously abandoned Project NERA, allegedly because it gained the attention of lawmakers and the public, and the particulars of Project Sandbox could be heavily modified, or perhaps abandoned altogether. It would be premature to adjudicate whether Google's contingent, hypothetical initiative would be anticompetitive conduct if adopted. See, e.g., Motor Vehicle Mfrs. Ass'n of U.S., Inc. v. New York State Dept. of Env't Conservation, 79 F.3d 1298, 1306 (2d Cir. 1996).

The Court declines to reach the question of whether the Complaint plausibly alleges that Project Sandbox, if implemented, would be anticompetitive.

12. The Complaint Plausibly Alleges that Google's Unified Pricing Policy Was Anticompetitive Conduct Directed to the Ad-Exchange Market and Ad-Buying Tools for Small and Large Publishers.

The Complaint asserts that beginning in 2019, Google began to implement a requirement that publishers using its DFP ad server set uniform price floors across multiple ad exchanges. (Compl't ¶¶ 451-69.) Historically, a publisher would set hundreds of different price floors, with variations tailored to specific ad exchanges and purchasers. (Id. ¶ 452.) These variable price floors helped publishers increase revenue and improve the quality of ads run on their sites. (Id. ¶ 453.) The Complaint states that publishers' ability to set variable price floors on AdX and for Google's ad-buying tools helped diversify returns and counter the adverse-selection problems caused by, for example, Google's use of encrypted IDs. (Id. ¶¶ 453, 464.)

Google perceived publishers' higher price floors for AdX and Google's buying tools as an impediment to growing those products' market share. (Id. ¶ 454.) Results of a Google survey indicated that publishers were setting higher price floors on Google products in order to improve ad quality and increase yield. (Id. ¶ 455.)

According to the Complaint, Google began to adopt and enforced unified pricing floors in order to foreclose competition and channel transactions to AdX. (Id. ¶ 456.) An internal memo stated that a unified pricing rule would cause more DV360 transactions on AdX at a higher margin. (Id. ¶ 456.) The Complaint asserts that instead of trying to attract publishers by offering an improved product, Google began to "punish" publishers that set higher floors on Google, and eventually eliminated publishers' ability to set variable floors. (Id. ¶ 458.) The Complaint also asserts that Facebook had conveyed opposition to variable price floors, and that Google set unified price floors in part to further its purportedly anticompetitive arrangement with Facebook. (Id. ¶ 467.)

Google's unified price floors blocked publishers from setting price floors that varied between exchanges and ad purchasers. (Id. ¶ 460.) The Complaint states that instead of leveling the playing field, unified rates give a price advantage to Google due to the fees that it charges for transactions made on a non-Google exchange. (Id. ¶ 461.) Because the pricing rules are enforced through the DFP ad server, they also affect competition even where AdX and Google's buying tools are not participants.

(Id. ¶ 462.) The inability to set a variable pricing floor for Google products allegedly enhanced the anticompetitive effects of Google's other auction-related activities. (Id. ¶ 464.)

The Complaint also asserts that unified floors disrupted the investments and expectations of publishers. One large publisher allegedly made "significant" investments to develop machine-learning algorithms that fine-tuned optimal price floors, and set a higher price floor on AdX, resulting in 11% revenue growth. (Id. ¶ 465.) The Complaint asserts that, externally, Google falsely told publishers that price floors were being implemented for their benefit. (Id. ¶ 466.)

The Complaint asserts that uniform pricing rules are exclusionary and have successfully foreclosed competition in the exchange market and the market for ad-buying tools. (Id. ¶ 468.) AdX's share of impressions allegedly grew "drastically" after the rules were implemented, winning nearly twice as many impressions "but paying roughly half as much." (Id.) One publisher allegedly observed that Google's ad-buying tools began to win three to four times as many impressions as before the rule change. (Id.) According to the Complaint, unified pricing also "coerce[d]" publishers to transact with Google's ad-buying tools on AdX. (Id. ¶ 469.) Previously, publishers could set a higher price floor for DV360 transactions on AdX relative to the floors on other exchanges. (Id.) Unified pricing ended this practice and "forced" publishers to transact with DV360 on AdX. (Id.)

The Complaint plausibly alleges that Google used its monopoly power in the market for publisher ad servers to coerce publishers to transact on AdX, advancing Google's monopolistic goals in the ad-exchange market. The Complaint asserts that unified pricing required publishers to "set the same price floor for different exchanges and the same price floor for different buyers." (Id. ¶ 460.) Through its monopoly power in the ad-server market, Google had the ability to effectively require publishers to set uniform prices that advantaged Google's own products and harmed its competitors.

Google argues that "equal treatment cannot harm competition" and "firms have no obligation to aid their competition in perpetuity." (Google Mem. at 25.) But Google's reliance on principles of equal treatment and the absence of an obligation to assist competitors misses the mark. Google's version of a uniform price floor does not permit publishers to adjust those floors to take account of the higher fees charged by Google for transactions on non-Google ad exchanges. Google may have the right to set its own prices (provided they are not predatory) but its right to restrict publishers' pricing decisions does not appear to have a legitimate business purpose other than to restrict competition in the ad exchange market. The Complaint plausibly explains why unified price rules did not provide "equal treatment" that leveled the competitive playing field but instead restricted competition in the ad-exchange market. Similarly, Google's assertion that it has "no obligation to aid [its] competition in perpetuity" suggests that publishers' use of variable floors required Google to extend affirmative assistance to its competitors. The Complaint does not describe variable pricing as "aid" extended by Google, and instead plausibly explains why Google's adoption of uniform price floors was a targeted anticompetitive measure that used Google's monopoly power to constrain competition between ad exchanges.

For largely the same reasons, the Complaint plausibly describes anticompetitive market in the ad-buying tools used for small advertisers and large advertisers.

Prior to the establishment of uniform price floors, publishers would set higher price floors for Google's buying tools in an effort to improve ad quality. (Compl't ¶ 453.) Uniform price floors prevented publishers from setting different floors for different ad buyers, which suppressed competition between ad-buying tools and coerced publishers into transacting with Google ad-buying tools. (Id. ¶ 460, 464.) One large publisher has concluded that uniform price floors resulted in Google's ad-buying tools winning three to four times as many ad impressions. (Id. ¶ 468.)

The Complaint plausibly alleges that the use of uniform price floors was anticompetitive conduct in the market for ad exchanges and the markets for ad-buying tools used by small advertisers and large advertisers.

13. The Facts Underlying the Section 1 Tying Claim Are Anticompetitive Conduct in the Publisher Ad Server Market in Support of the Section 2 Claims.

For the reasons discussed in connection with the States' tying claim brought under section 1, the Complaint also plausibly alleges under section 2 that Google unlawfully tied publishers' access to AdX to the requirement that they enter into contracts to license the DFP ad server. (Compl't ¶ 537.) "Tying may violate Section 2 of the Sherman Act as well as Section 1 of the Sherman Act ... if the defendant has either the requisite power in the market for the tying product or a dangerous probability of acquiring market power in the market for the tied product." Ortho Diagnostic Sys., Inc. v. Abbott Laboratories, Inc., 920 F. Supp. 455, 472 (S.D.N.Y. 1996) (Kaplan, J.). Otherwise, the elements of a tying claim are the same. See, e.g., E & L Consulting, 472 F.3d at 31 (reciting tying elements for tying claim brought under sections 1 and 2); Coniglio v. Highwood Servs., Inc., 495 F.2d 1286, 1293 (2d Cir. 1974) (dismissing section 2 tying claim for the same reasons that section 1 tying claim was dismissed).

As previously discussed, the Complaint plausibly alleges that Google used its monopoly power in the ad-exchange market to maintain and expand its monopoly power in the ad-server market by coercing publishers into using the DFP ad server in order to access AdX. The Court therefore concludes that the facts underlying the tying claim actionable under Count III amounts to anticompetitive conduct for the purposes of the section 2 monopolization claim in the ad server market.

VII. THE COMPLAINT DOES NOT PLAUSIBLY ALLEGE THAT DYNAMIC ALLOCATION AND DRS HAVE CONTINUING, PRESENT ADVERSE EFFECTS, AND THIS CONDUCT CANNOT BE ENJOINED.

The States seek only injunctive and other equitable relief for their Sherman Act claims. (Compl't ¶ 683.) The Complaint's prayer for relief includes separate provisions for "injunctive relief" and "structural relief" that would "restore competitive conditions in the relevant markets affected by Google's unlawful conduct ...." (Id. ¶¶ 683(c), (d).) It separately seeks to "[e]njoin and restrain" Google "from continuing to engage in any anticompetitive conduct" and from adopting future practices that are similar to those described in the Complaint. (Id. ¶ 683(e).)

The Complaint identifies Dynamic Allocation and DRS (VI.D.2 & 5, above) as schemes that concluded in 2019. In the heading that precedes the Complaint's allegations about Dynamic Allocation, the Complaint states that the purported scheme was in place from 2010 to 2019. (Id. at p. 94.) The heading that precedes

the allegations about DRS states that the scheme was in place from 2014 to 2019. (Compl't at 111.) The Complaint does not allege that Dynamic Allocation or DRS are programs that are likely to recur. Google urges that the States cannot seek injunctive relief directed to conduct that has ended.

The States urge that Google's argument is misplaced because Rule 12(b)(6) provides a vehicle to seek the dismissal of a claim, and not the dismissal of remedies. See, e.g., Burkina Wear, Inc. v. Campagnolo, S.R.L., 2008 WL 1007634, at *3 (S.D.N.Y. Apr. 9, 2008) ("[T]he availability of the specific relief requested pursuant to any given count of the Complaint is not relevant to the question of whether [plaintiff] has stated a claim.") (Sweet, J.). But when a plaintiff seeks injunctive relief, "'[p]ast exposure to illegal conduct does not in itself show a present case or controversy regarding injunctive relief ... if unaccompanied by any continuing, present adverse effects.'" City of Los Angeles v. Lyons, 461 U.S. 95, 102, 103 S.Ct. 1660, 75 L.Ed.2d 675 (1983) (quoting O'Shea v. Littleton, 414 U.S. 488, 495-96, 94 S.Ct. 669, 38 L.Ed.2d 674 (1974)). If a complaint does not allege that the defendant's past conduct is responsible for continuing, adverse effects, the claim for injunctive relief is non-justiciable, and the plaintiff lacks Article III standing. See, e.g., Pungitore v. Barbera, 506 Fed. App'x 40, 41-42 (2d Cir. 2012) ("past wrongs may serve as evidence bearing on whether there is a real and immediate threat of repeated injury" but they do not "show a present case or controversy regarding injunctive relief ....") (quotation marks omitted) (summary order).

The Complaint's allegations about Dynamic Allocation are historical in nature. They describe the practice's introduction in 2010, at a time when the digital advertising market moved away from the practice of waterfalling ad exchanges, and asserts that Dynamic Allocation "propelled Google's AdX exchange to the top of the market by 2013." (Compl't ¶¶ 267-71, 281.) The Complaint does not assert that Google continues to implement Dynamic Allocation and does not explain the circumstances of how the scheme was wound down, noting only in a heading that it concluded in 2019. The Complaint does not identify continuing adverse effects that are attributable to Dynamic Allocation. Google's use of Dynamic Allocation may be evidence that bears on whether there is a threat of repeat injury, but the Complaint does not plausibly allege ongoing or imminent harms caused by Dynamic Allocation that could allow for injunctive relief. See Pungitore, 506 Fed. App'x at 41-42.

DRS is described as a scheme that ran from 2014 to 2019. Google allegedly began its implementation in 2014; by 2015, it allegedly opted all publishers into DRS without disclosing its terms, and in 2016, told publishers that it was enacting a form of revenue-sharing optimization, without disclosing the scope and prior implementation of DRS. (Compl't ¶¶ 318, 326-27.) As with Dynamic Allocation, the Complaint does not describe the circumstances of Google's decision to wind down DRS, and notes in a heading that it concluded in 2019. The Complaint does not identify continuing adverse effects that are attributable to DRS. DRS is one of several Google initiatives that involved Google's alleged manipulation of publisher-set price floors, and therefore may be evidence bearing on whether there is a threat of repeat injury, but the Complaint does not plausibly allege ongoing or imminent harms caused by DRS that could warrant injunctive relief directed to that now-ended conduct. See Pungitore, 506 Fed. App'x at 41-42. VIII. THE COURT DECLINES TO ADJUDICATE GOOGLE'S LACHES DEFENSE AT THE PLEADING STAGE.

Google urges that laches bars the States' claims directed to conduct that occurred before December 16, 2016, which is four years before the filing of the States' original complaint. "Laches is a defense developed by courts of equity to protect defendants against unreasonable, prejudicial delay in commencing suit." SCA Hygiene Prod. Aktiebolag v. First Quality Baby Prod., LLC, 580 U.S. 328, 137 S. Ct. 954, 960, 197 L.Ed.2d 292 (2017) (quotation marks omitted). "The elements of a traditional laches defense are: '(1) lack of diligence by the party against whom the defense is asserted, and (2) prejudice to the party asserting the defense.'" Fed. Ins. Co. v. United States, 882 F.3d 348, 365 (2d Cir. 2018) (quoting Costello v. United States, 365 U.S. 265, 282, 81 S.Ct. 534, 5 L.Ed.2d 551 (1961)).

When a plaintiff brings a claim under a statute that does not include a limitations period, the laches doctrine serves a "gap-filling" function. SCA Hygiene, 137 S. Ct. at 961. Here, the States seek injunctive relief pursuant to section 16 of the Clayton Act, 15 U.S.C. § 26, and assert claims "in their respective sovereign capacities" and parens patriae on behalf of their citizens. (Compl't ¶ 31.) Because section 16 does not include its own limitations period, the laches doctrine governs the claims' timeliness. See, e.g., Telectronics Proprietary, Ltd. v. Medtronic, Inc., 687 F. Supp. 832, 841 (S.D.N.Y. 1988) ("The doctrine of laches does not normally apply to antitrust claims, although it can be applied to equitable claims made under section 16 of the Clayton Act.") (Leisure, J.).

In applying laches to a statutory claim, courts typically look to the closest comparable limitation period. See Conopco, Inc. v. Campbell Soup Co., 95 F.3d 187, 191 (2d Cir. 1996) ("Although laches is an equitable defense, employed instead of a statutory time-bar, analogous statutes of limitation remain an important determinant in the application of a laches defense."). For a section 16 claim, courts use the Clayton Act's four-year limitations period for money-damages claims as a "guideline." State of New York v. Facebook, Inc., 549 F. Supp. 3d 6, 34 (D.D.C. 2021) (collecting cases); 15 U.S.C. § 15b. "The starting presumption, then, is that regardless of whether a Section 16 plaintiff seeks damages or an injunction, it must file its lawsuit within four years from the accrual of the claim." State of New York, 549 F. Supp. 3d at 35 (quotation marks omitted).

"When a suit is brought within the time fixed by the analogous statute, the burden is on the defendant to show ... circumstances exist which require the application of the doctrine of laches. On the other hand, when the suit is brought after the statutory time has elapsed, the burden is on the complainant to aver and prove the circumstances making it inequitable to apply laches to his case." Conopco, 95 F.3d at 191 (quoting Leonick v. Jones & Laughlin Steel Corp., 258 F.2d 48, 50 (2d Cir. 1958)); accord Reconstruction Fin. Corp. v. Harrisons & Crosfield, 204 F.2d 366, 370 (2d Cir. 1953). "[O]nce the analogous statute has run, a presumption of laches will apply and plaintiff must show why the laches defense ought not be applied in the case." Conopco, 95 F.3d at 191.

"The determination of whether laches bars a plaintiff from equitable relief is entirely within the discretion of the trial court." Tri-Star Pictures, Inc. v. Leisure Time Prods., B.V., 17 F.3d 38, 44 (2d Cir. 1994). "The equitable nature of laches necessarily requires that the resolution be

based on the circumstances peculiar to each case." Id. Judge Stanton has observed that laches can rarely be successfully invoked on a Rule 12(b)(6) motion because the application "necessarily involve[s] a fact-intensive analysis and balancing of equities that would require the Court to consider matters outside of the pleadings that are in dispute." Broadcast Music, Inc. v. Liberman Broad., Inc., 2016 WL 3919654, at *2 (S.D.N.Y. July 14, 2016) (quotation marks omitted).

The States urge that laches ought not apply because, as sovereigns acting to vindicate the rights of the public, principles of equity afford them powers not granted to private litigants. See generally Guaranty Trust Co. of N.Y. v. United States, 304 U.S. 126, 132-33, 58 S.Ct. 785, 82 L.Ed. 1224 (1938) (discussing sovereign's historical exemption from laches). The States bring their claim pursuant to section 16 of the Clayton Act, which specifically established a private right of action to seek injunctive relief. 15 U.S.C. § 26 ("Any person, firm, corporation, or association shall be entitled to sue for and have injunctive relief ... against threatened loss or damage by a violation of the antitrust laws...."). Section 16 has long been construed to permit states to bring a claim parens patriae, as well as to bring claims advancing their own sovereign interests. State of Georgia v. Pennsylvania R. Co., 324 U.S. 439, 447, 65 S.Ct. 716, 89 L.Ed. 1051 (1945).

The recent district court decision in State of New York explained that the Sherman Act originally permitted only the United States to seek injunctive and equitable relief, and that they Clayton Act was adopted to permit enforcement by broader categories of plaintiffs, including states. 549 F. Supp. 3d at 39. "As such, the Congressional judgment was that states, like private parties, are entitled to relief under Section 16 'under the same conditions and principles as injunctive relief against threatened conduct that will cause loss or damage is granted by courts of equity....'" Id. (quoting 15 U.S.C. § 26). State of New York also explained that the Supreme Court has indicated, though not expressly held, that states are treated as private parties when they invoke section 16. Id. at 38; see California v. American Stores Co., 495 U.S. 271, 279, 287, 296, 110 S.Ct. 1853, 109 L.Ed.2d 240 (1990) (referring to state-brought proceeding as a "private action" and discussing the effect of section 16 on "private litigants"). State of New York concluded that states, like other plaintiffs authorized to bring a claim under section 16, are bound by the doctrine of laches and may not unreasonably delay the bringing of claims. 549 F. Supp. 3d at 39-40; see also State of N.Y. v. Kraft Gen. Foods, Inc., 862 F. Supp. 1030, 1033 (S.D.N.Y. 1993) ("Although the State of New York is a governmental actor, it is considered a private party when seeking an injunction pursuant to the Clayton Act.") (Wood, J.), aff'd, 14 F.3d 590 (2d Cir. 1993); Gov't of Puerto Rico v. Carpenter Co., 442 F. Supp. 3d 464, 473-75 (D.P.R. 2020) (applying laches bar to section 16 claims brought by the Commonwealth of Puerto Rico).

The thorough and detailed reasoning of State of New York is persuasive as to the application of laches to section 16 claims brought by a state. Section 16 of the Clayton Act was enacted to expand antitrust enforcement to include persons and entities other than the United States, and a plaintiff proceeding under section 16 does not act in the capacity of a sovereign but as a private enforcer. The Court therefore concludes that the doctrine of laches may be applied to the States' claims, and that the laches analysis is guided by the

four-year limitations period of 15 U.S.C. § 15b.

At this early stage of the proceedings, the circumstances specific to this case weigh against applying laches to bar the States' tying claim and its section 2 claim. Discovery is needed to shed further light on whether the States unreasonably delayed bringing these claims.

The Complaint describes conduct on the part of Google that lacked transparency, occurred out of the public eye, and had effects that were not immediately obvious or well understood. This weighs against a conclusion that, as a matter of law, the States unreasonably delayed bringing this action, and complicates the question of whether the States' claims accrued prior to December 16, 2016, or sometime thereafter. The issue is more properly revisited on a more developed record following discovery.

The Complaint's tying allegations describe a course of conduct that was conceived and initiated in 2010. (Compl't ¶ 246.) It states that an internal Google study from 2013 observed that publishers who did not license DFP and received static bids instead of live, competitive bids from AdX saw a 20 to 40 percent revenue drop. (Compl't ¶ 247.) One publisher reached a similar conclusion in 2017, and abandoned plans to switch from using Google's ad server after concluding that the loss of AdX bids outweighed the benefits of changing subscriptions. (Compl't ¶ 247.) In 2018, Google renegotiated certain publisher agreements that had previously allowed publishers to access AdX without using the DFP ad server, which suggests that prior to 2018, at least some publishers were able to use AdX without being tied to the DFP ad server. (Compl't ¶ 251.)

Google's memorandum urges that laches apply because "advertisers and publishers necessarily must have been aware" of Google's conduct and because the Complaint does not identify a new and independent act that caused a new, accumulating injury. (Google Mem. at 13-14.) This is not apparent from the face of the Complaint. As late as 2018, some publishers continued to access AdX without the requirement that they subscribe to the DFP ad server. (Compl't ¶ 247.) Google's memorandum assumes that advertisers and publishers "must have been aware" of Google's purportedly unlawful conduct, but the Complaint also asserts that a publisher did not recognize the extent of the alleged anticompetitive effects until 2017, when it considered switching to a different ad server. (Compl't ¶ 247.) The Complaint does not describe conduct that is comparable to the public roll-out of a new initiative or product line, and instead suggests a type of piecemeal, escalating implementation of the product tie, the scope and effect of which may not have been easily recognized in real time. A more complete factual record is needed in order to adjudicate whether the States' tying claim is the product of unreasonable delay or whether the claim was timely brought.

Similarly, in the Complaint's telling, Google's rollout of EDA was opaque, and accompanied by misrepresentations about its intent and effects. The Complaint asserts that Google "introduced" EDA in 2014. (Compl't ¶ 282.) It asserts that Google "automatically turned on" EDA for publishers and falsely informed them that EDA would increase their inventory yields. (Compl't ¶ 291.) Internally, however, Google perceived EDA as a way to cherry-pick the highest-value inventory for AdX. (Compl't ¶ 292.) The Complaint asserts that currently, publishers "have no choice" other than to remain enrolled in EDA, or else DFP will not allow them to receive live, competitive bids. (Compl't ¶ 293.) These allegations describe a purported

scheme that enrolled publishers without their advance consent, which was misrepresented to publishers as a way to increase yield but internally understood by Google as a way to transact highest-value inventory on AdX. (Compl't ¶¶ 282, 291-92.) As described in the Complaint, publishers would have quickly been aware that EDA existed, but not had have been able to readily observe its anticompetitive effects or purpose.

The extent of Google's prejudice also is not apparent at this stage. Google states in its memo that it would be prejudiced because it has "continually improv[ed] its digital advertising tools over the past decade, only to have some of the core functionality of those tools challenged now ...." (Google Mem. at 15.) It also states that "Google could have deployed its resources differently" if plaintiffs had brought their claims earlier, and that decisionmakers may have "potentially stale recollections ...." (Reply at 13.) It is true that a defendant may be prejudiced if unfair delay precluded it from "effectively adopt[ing] an alternative" strategy, Conopco, 95 F.3d at 192, but Google's assertion here is vague and general. Conopco, by contrast, found undue delay and prejudice following a bench trial. Id. at 190.

The nature of Google's purported schemes also contrast with the underlying conduct in State of New York, which applied laches to bar various states' claims for injunctive relief. There, six years and eight years after Facebook made "highly publicized" acquisitions of Instagram and WhatsApp, various plaintiff states sought to compel the divestiture of the acquired companies. 549 F. Supp. 3d at 40. The transactions were well-known at the time they occurred and drew scrutiny from the FTC and European regulators. Id. at 43. The court concluded as a matter of law that the plaintiff states unreasonably delayed bringing claims for injunctive relief, and that Facebook faced prejudice from the years-late effort to undo its acquisitions. Id. at 49. A belated effort to compel the divestiture of high-profile acquisitions that drew contemporaneous regulatory scrutiny is significantly different from the more opaque activities described in the Complaint.

Other context-specific factors weigh against applying laches at this time. The allegations of Google's anticompetitive conduct and intent relating to the tying claim and the EDA claim are bolstered by internal Google materials that would not have been known to customers or rivals at the time. The Court also affords some weight to the fact that the claims are brought by states, as opposed to competitors.²⁸

28 See State of New York, 549 F. Supp. 3d at 40 ("Although the doctrine of laches therefore applies to parens patriae suits such as this one, the Court does not mean to suggest that the presence of state plaintiffs has zero effect on the analysis. Laches is an equitable doctrine, and in the balancing of the equities, it is of course relevant that this suit is brought not by a competitor hoping to seriously interfere with a rival's business operations, but rather by many of the states of the Union.") (quotation marks omitted).

The applicability of laches is more appropriately adjudicated on a more developed factual record, rather than on a Rule 12(b)(6) motion, where the Court is limited to consideration of the Complaint and any documents that are integral thereto. The Complaint is sufficient to aver that the tying claim and EDA claim are not the products of undue delay. Google's motion to dismiss these claims on laches grounds will therefore be denied.

CONCLUSION.

For the reasons explained, Google's motion to dismiss is GRANTED as to Count IV and otherwise DENIED. The Clerk is directed to terminate the motion. (21-md-3010, Docket # 217; 21-cv-6841, Docket # 176.)

SO ORDERED.