Judge P. Kevin Castel of the Southern District of New York on September 30, 2026 denied most of Google's motion for summary judgment against Gannett and the publishers of the Daily Mail, leaving claims about Project Bernanke, Enhanced Dynamic Allocation and Project Poirot for a jury to weigh, while dismissing both publishers' New York consumer protection claims and every federal antitrust claim brought by the defunct video company Inform.

The 88-page opinion, Document 2041 in the multidistrict litigation 21-md-3010, rules on summary judgment motions and expert-exclusion motions, known as Daubert motions, in four coordinated cases. Ali Hortacsu, a University of Chicago economist retained by the two publishers, calculated damages of $599,955,198 for the Daily Mail and $900,771,489 for Gannett, plus interest. Einer Elhauge, the economist for a certified class of publishers that sold through Google's AdX exchange, puts class-wide damages at $1.72 billion through March 31, 2024. Google asked the court to set aside all of that expert work. It failed on every count.

In Short

A judge in New York decided that most of the money claims two big news publishers brought against Google cannot be thrown out before trial, so jurors will hear them. The publishers say secret Google auction programs cost them roughly $600 million and $900 million, and a group of websites that sold ads through Google's exchange says it was overcharged $1.72 billion. Two other sellers lost: a publisher that barely used the product it sued over, and a video company that could not show Google controlled online video advertising. For anyone selling ads through Google, what changes is that programs the earlier Virginia case never ruled on, such as Project Bernanke, now get tested in front of a jury, along with how much Google may owe.

How the four cases came out

The four cases share a docket but not a theory. Associated Newspapers Ltd. and Mail Media, Inc., which the court calls Daily Mail, sued individually, as did Gannett Co., Inc. A class of United States publishers that sold open-web displayimpressions through AdX, represented by The Nation Company and Genius Media Group, litigates alongside The Progressive and Mikula Web Solutions, whose claims concern Google's basic AdSense product. Inform Inc., which sold online video advertising, sued separately.

For Gannett and the Daily Mail, according to the opinion, Google's motion to exclude Hortacsu and Shengwu Li, a Harvard auction theorist, is denied, and its summary judgment motion is granted only on the New York General Business Law claims. The publishers' own motion against Google's failure-to-mitigate defense is granted in part, and Google withdrew a defense of unclean hands.

In the class case, Google's motion is denied as to the AdX Class and Mikula and granted as to The Progressive, and a motion to exclude Elhauge and a second class expert, Anand Das, is denied. Inform lost Counts I through V, its Sherman Act and Clayton Act claims, and the court asked for letter briefs on its two remaining Georgia law claims.

The posture explains why the ruling carries weight. Judge Leonie Brinkema of the Eastern District of Virginia found on April 17, 2025 that Google monopolized the publisher ad server and ad exchange markets and unlawfully tied the two together. On October 27, 2025, Castel gave those findings preclusive effect in New York, barring Google from relitigating the market definitions, their worldwide scope, the tie between DoubleClick for Publishers (DFP) and AdX, and the practices known as First Look, Last Look and Dynamic Revenue Share, plus Unified Pricing Rules for Gannett, Daily Mail and the AdX Class. That ruling carried a caveat the new opinion repeats: preclusion does not relieve plaintiffs of proving antitrust injury and damages. The September 30 opinion is where that burden met Google's challenges claim by claim.

The practices Virginia never ruled on

Google's main line of attack, according to the opinion, targeted conduct outside the Virginia findings, which the parties call the "Non-Precluded Acts." Google argued that no reasonable jury could find these practices harmed competition in the ad server or ad exchange markets. Castel disagreed on each of them.

Enhanced Dynamic Allocation

Under Enhanced Dynamic Allocation, implemented in 2014, impressions publishers had reserved for negotiated direct deals were made available for programmatic bids on AdX. Before EDA, according to the opinion, DFP prioritized fulfilment of direct sales. Under EDA, AdX stopped prioritizing those deals and assigned a temporary price lower than the one the publisher had already attached to the direct sale. Li calls the result a "cream-skimming problem": AdX could take an impression no matter how high the direct price, and the adverse selection that followed made advertisers less willing to buy direct at all. Li also concludes that EDA "probabilistically throttled" bidders arriving through header bidding, since only AdX could reach the high-value impressions.

Google's own correspondence is blunt about the mechanics. In a September 2016 email chain quoted in the opinion, a Google employee wrote that "with EDA on we set the reserve price at $1, and any AdX buyer including GDN [Google Display Network] can win over the query as long as they bid more than $1." The next sentence reads: "This transaction happens before we actually call Pubmatic [a rival ad exchange] and get their actual bids." Internally, Google described EDA as a way to let "AdX/AdSense buyers cherry-pick higher revenue impressions while respecting the delivery constraints of the reservation ads."

Hortacsu concludes that EDA reduced the global value of direct sales by approximately 24.1%. For Gannett, his figures put EDA damages at $348.9 million in the United States and $434.8 million worldwide, about 48% of the publisher's total claim.

Google's central legal argument was that direct sales sit outside any relevant antitrust market, so losses there cannot be antitrust injury. Castel rejected that framing. The harm, on the publishers' theory, arose when Google used its ad server monopoly to push direct sales onto AdX at manipulated price floors, which places it, in the language of Blue Shield of Virginia v. McCready, within the area of the economy endangered by the breakdown of competition.

The computational weight of the feature has already been documented in the Virginia remedies proceedings. Google engineering director Glenn Berntson testified that EDA runs on a seven-day window of historical bid data, recalculated daily in a job taking roughly 10 hours across 4,000 computers.

Google also attacked Li's counterfactual, an optimized EDA that publishers could have applied exchange by exchange, as a product that never existed. The court noted that one of Google's own experts had patented a method intended to prevent cream skimming while still letting direct deals compete, and quoted Li's view that "as Google enforced EDA across all publisher inventory, with no ability to opt out, exchanges had no incentive to develop innovations that would have solved the cream-skimming problem." The jury will decide whether that world is plausible.

Minimum Bid to Win

Google adopted Minimum Bid to Win in 2019, when AdX moved to a first-price auction and dropped Last Look. After each auction, DFP told AdX bidders and exchange bidding participants the lowest bid that would have won, in effect telling the winner how much it overpaid. A Google engineer's declaration describes Google Ads modelling "the bid landscape on any given auction based on bid data from past auctions, including (since 2019) minimum-bid-to-win data."

According to Li, other exchanges did not share that information, so only AdX buyers could predict the winning price with precision. He opines that the feature depressed publisher prices, discouraged bidding on rival exchanges and was "allocatively inefficient." The court found the evidence sufficient for a jury.

Bernanke, Global Bernanke, Bell and Alchemist

The longest-running program carries the broadest label. According to Li's report, the original Project Bernanke ran from 2013 to 2015, Global Bernanke from 2015 to 2019, Project Bell v2 from 2016 to 2019, and Alchemist from 2019 to the present. In the publishers' account, Google used its Google Ads buying tool to inflate and deflate advertiser bids on AdX, charging winners more than the price paid to AdX and banking the difference in pools that subsidized Google Ads bids where outside competition was strong. The pools were publisher-specific at first and became a single pool under Global Bernanke.

An internal Google memorandum quoted in the opinion states that "the current Bernanke increases the AdX win rate from 37% to 44% (match rate = 50%) and increases revenue by $290m." The same memorandum estimates that removing a fixed publisher margin constraint could increase Google Display Network revenue by up to $200 million and push the AdX win rate to about 51%.

Other documents use the language of cartels. An email of February 17, 2017 says the buyers were "currently implementing a 'bidding ring' which the FTC has declared illegal in many instances." A draft presentation dated July 18, 2017 states: "The strategy is reminiscent of a bidding ring in an auction: one member of the ring bids just enough to win the item, while the other players sit on their hands." And an email of June 15, 2017 complains of hitting "a 'the first rule about Bernanke is we don't talk about Bernanke' situation."

Li opines that Global Bernanke functioned like a bidding cartel, while clarifying that it "generates the outcomes of a cartel, not that Google Ads bidders were literally engaged in cartel behavior themselves." Google wanted those opinions excluded because the publishers bring no cartel or predatory pricing claim. Will jurors hear an economist compare Google's bidding system to a bidding ring? The court left that for later, holding only that the labels do not make Li's account of how Bernanke worked unreliable. The publishers do not claim a separate market for ad-buying tools; their theory is that Google used control of Google Ads to steer bids toward AdX.

PPC Land has walked through the Bernanke arithmetic as pleaded by Teads in August 2026, and Raptive's October 2025 complaint listed Bernanke and Alchemist among its targets. This is the first ruling in the New York litigation to say evidence on those programs can reach a jury.

Poirot

Poirot worked on the buying side. According to the opinion, Google adjusted bids from its DV360 buying tool to cut volume on rival exchanges and raise it on AdX. Li describes a form of bid shading aimed at exchanges "that do not run a 'clean' second-price auction." The publishers challenge 2019 to 2022, when Poirot used what Li calls a "pretextual 'risk aversion' parameter" that raised DV360's bids on AdX; he found no evidence that bidders actually held different risk aversion across exchanges.

Internal Google documents summarized by Li record that Poirot had "been quite effective" at counteracting header bidding while DV360 spent 7% more on AdX. Li calculates that it increased AdX revenue by 7.5% and cut spending on rival exchanges by 14.7%. Other filings cite different numbers: Magnite's 2025 complaint, drawing on trial data, said DV360 advertisers spent 9% more on AdX and 10% less elsewhere after the first version, and Index Exchange alleged a second version shaded bids by as much as 90%. The versions and measurement windows differ, and the opinion does not reconcile them.

What the court would not stretch

Not every argument went the publishers' way. They are not suing over the Admeld and DoubleClick acquisitions, and Castel rejected their claim that the October 2025 ruling precluded Google from disputing an "AdX/Google Ads tie." Brinkema's finding that Google Ads demand was "largely limited" to AdX is "not a finding of an unlawful tie between Google Ads and AdX," according to the opinion. Teads' August complaint in the same court pleads a tying count covering Google Ads to AdX; on Castel's reading, it has no Virginia finding behind it.

How the two publishers count their losses

Hortacsu builds each publisher's total from separate models for separate practices, and the opinion supplies several components.

For Last Look, comparing historical data with 2024 first-price data, he finds bid depression worth $8.36 million to the Daily Mail and $18.04 million to Gannett. For the take rate Google charged on AdX, he finds overcharges of $81.32 million and $106.10 million respectively. His benchmark is the 5% fee Google charged on AdX for programmatic guaranteed sales, where a publisher and buyer fix price and volume in advance. Those sales do not compete with open auctions, Hortacsu concedes, but they run through the same "AdX Plumbing," so cost does not explain the gap. The court accepted the benchmark as sufficiently similar.

A footnote shows how large some effects are relative to price. Matching the 1.59% win-rate gain from Dynamic Revenue Share would have required cutting the AdX fee from 20% to 15.65%, according to Hortacsu, and matching the 6.48% gain from Unified Pricing Rules, to 3.82%. For Bernanke, Last Look and Alchemist, no fee cut alone could have done it. In Virginia, another expert estimated that removing Unified Pricing Rules alone would cut the AdX take rate to roughly 16.6%.

Another component, an "Auction Pressure Model," rests on the principle that auctions with more bidders clear higher. It finds that since 2017, reduced participation by third-party exchanges cost the Daily Mail $234 million and Gannett $116 million. Google challenged its opportunity-cost assumptions; Hortacsu says they rest on more than 2,000,000 data points from 800,000 auctions. Google's own expert, Judith Chevalier, testified that her criticisms were limited to "informational assumptions" and that his models draw on "very famous papers." The dispute goes to cross-examination.

The rest of the world

Google also argued that losses outside the United States were barred by the Foreign Trade Antitrust Improvements Act. The split is significant for the Daily Mail, whose claimed foreign losses exceed its domestic ones. Hortacsu calculates its United States damages at $216,880,035 and its rest-of-world damages at $383,075,163. For Gannett, the figures are $797,831,266 and $102,940,233.

Those Gannett components add up to $900,771,499, ten dollars more than the $900,771,489 total the opinion reports twice. The opinion does not address the gap, which looks like a transcription error.

Castel rejected the foreign-commerce argument. Alphabet and Google LLC are United States companies, there is evidence the monopolization was planned and implemented from the United States, and six of AdX's major rivals are American against two abroad. Unlike the precedents Google cited, there is no foreign defendant and no wholly foreign conduct, and even if the statute applied, a jury could find a direct, substantial and reasonably foreseeable effect on domestic commerce.

Where Google won: sophistication

Google's one clean victory against the two publishers came on their claims under the New York General Business Law. The Daily Mail sued under section 349, which prohibits deceptive business practices, and Gannett under sections 349 and 350, which add false advertising. Both require the challenged conduct to be consumer-oriented.

Google pointed to the publishers' own descriptions. Each called itself a "large publisher." Gannett's corporate witness described the company as "quite sophisticated" at digital ad selling, and the Daily Mail said it "operates a sophisticated tech stack" with more than a dozen engineers in its advertising division. Castel concluded that no reasonable jury could treat the alleged deceptions about EDA and Bernanke as consumer-oriented. These were "complex [and] unique arrangements" used to sell billions of impressions for hundreds of millions of dollars, not the "modest" transactions the statute targets, according to the opinion.

The reasoning reaches other plaintiffs in the same courthouse. Teads' complaint includes a count under General Business Law sections 349 and 350.

Elsewhere Google fared worse. Gannett's unjust enrichment claim survives because Google identified no contract terms governing the alleged deceptions. The limitations defense failed too: class complaints filed on December 15, 2020, December 23, 2020 and December 4, 2022 tolled the four-year clock, and the tie first imposed in 2010 was renewed through combined contracts in 2018. A jury could treat those renewals, later changes to EDA and Dynamic Revenue Share, and each affected auction as fresh overt acts. Evidence that EDA's workings stayed unknown until government investigations around 2020 could also support tolling for fraudulent concealment.

Mitigation: tabloid content out, ad server settings in

The publishers also asked the court to strike Google's defense that they failed to mitigate their losses. Part of that defense was unusual. Google argued that advertisers avoided the Daily Mail because its sites carried "salacious, explicit, and violent" content, and blamed Gannett's losses on its merger with GateHouse Media, which it called a "Hail Mary play," on four campaigns where ads were not placed or viewable, and on nine months of inaccurate placement information. Castel found no discernible connection between any of that and the claimed antitrust damages, and struck those arguments.

The defense survives on the ad server itself. Google relies on Hortacsu's own numbers: if Gannett had stopped using AdX for indirect sales, its indirect revenue would have fallen 27%, a drop of $193.5 million in the United States and $217.4 million worldwide, against EDA damages of $348.9 million and $434.8 million. Google argues that Gannett could have reduced its EDA damages by $191.4 million by leaving AdX. The opinion does not show how that figure is derived; the numbers it cites produce differences of $155.4 million in the United States and $217.4 million worldwide.

Professor Milgrom, whose testimony Google cites, adds that publishers could have limited exposure by setting price floors that recreated the pre-EDA process, by assigning direct deals to 100% sponsorship line items outside EDA's reach, or by disabling Dynamic Allocation for favored exchanges after Unified Pricing Rules arrived. The publishers argued Google was estopped by Brinkema's finding that forgoing AdX was not financially viable for large publishers. Castel disagreed: Google claims only that leaving would have cost less than staying, which is consistent with an unlawful tie. Evidence that AdX was unavoidable will now be used to argue over the price of avoiding it.

The AdX class and a tie running the other way

The AdX Class consists of United States publishers that paid Google, directly or through reduced revenue, for selling impressions on AdX. Elhauge opines that two product ties allowed Google to charge a 19.67% take rate on AdX transactions against a 10% benchmark, producing the $1.72 billion figure. Unlike Hortacsu, he does not isolate damages for individual bidding practices.

The Act 1 tie is the one Brinkema found unlawful: publishers had to use DFP, the tied product, to receive real-time bids from AdX, the tying product. That claim is precluded, and Castel denied Google's motion on it, repeating a December 2025 ruling that a model measuring the combined impact on DFP and AdX can be reliable.

The Act 2 tie reverses the direction. The class alleges that Google coerced DFP customers into selling their impressions on AdX through EDA, Dynamic Allocation with First Look and Last Look, Dynamic Revenue Share and Unified Pricing Rules. No such tie was alleged in Virginia, and no party argued for preclusion. Elhauge testified that his damages figure would be the same whether a jury finds liability on Act 1, Act 2 or both, and he credits EDA with raising AdX revenues by more than $250 million a year.

Google's defense was that nothing forced publishers onto AdX. Nitish Korula, a Google engineer, testified in Virginia that "the publisher has to affirmatively take some step to provide that access like creating an AdX line item," and that AdX could be switched off for parts of an inventory. Elhauge himself wrote that the restraints "did not impose an absolute requirement that DFP publishers use AdX for all their sales." The class answered that AdX competed for 80% to 90% of DFP-served impressions and that publishers saw turning it off as an "option to starve to death."

Castel sided with the class at this stage. A contract for AdX and DFP, he wrote, was not like buying a discrete good or a parcel of land but a step toward potentially billions of later sales, and a jury could find the restraints coerced DFP customers to route impressions to AdX on Google's terms. He likened it to a railroad coercing existing land lessees to ship on its line, a variation on the Supreme Court's Northern Pacific Railway tying case.

In July, the Second Circuit held in Nielsen's dispute with Cumulus that a supplier can violate antitrust law through pricing designed to achieve a tie, not only through an explicit refusal to sell separately. Castel does not cite that decision, but both rest on the idea that a tie can arise from conduct, not only from contract terms.

Two further rulings favor the class. Google's expert calculated that $152.05 million of Elhauge's figure relates to websites of foreign "child" publishers managed by United States firms under Google's Multiple Customer Management program. Because the class consists of United States entities and the conduct was domestic, the court held the foreign-commerce statute does not apply. The timeliness challenge failed for the same reasons as against the two publishers.

The class also confirmed what it dropped in a December 12, 2025 letter: claims under California's unfair competition law and Cartwright Act, and reliance on Project Bernanke, Line-Item Capping, Redaction of Auction Data, Project Poirot, Project Elmo, Minimum Bid to Win and discriminatory risk aversion coefficients. Those practices remain live only in the Gannett and Daily Mail cases. One plaintiff, JLaSalle Enterprises, was terminated.

AdSense: one plaintiff out, one through

The Progressive and Mikula challenge what the court calls the "AdSense Package," Google's basic ad server, installed as a piece of code called the AdSense Tag, sold together with access to its basic ad auction platform. They allege two ties. Act 3 claims Google used power in basic auction platforms to force publishers onto its ad servers. Act 4 claims it tied the AdSense Tag to the basic auction platform, impairing competition among ad networks. The Virginia case never addressed AdSense, so nothing is precluded.

The Progressive sold more than 99.9% of its impressions through DFP, not AdSense, and there is no evidence its website carried the AdSense Tag. The court had already refused to let it represent an AdSense class in December 2025 for that reason. With no evidence that it was subject to either tie, it could not show antitrust injury, and judgment was entered for Google.

Mikula is different. Its sole employee, Pamela Mikula Paolino, testified that the firm would have preferred to reach Google's ad demand without the AdSense ad server. Google argued the publishers never pleaded a separate market for basic ad servers; Castel found it had sufficient notice, since the complaint distinguished AdSense's small-publisher customers from DFP's, and Elhauge splits ad servers into advanced and basic submarkets.

Google attacked Elhauge's reliance on data from a website called 6sense for AdSense's market share, noting large swings over four months. A jury could give that data little weight, the court said, and rely instead on evidence such as what Mikula calls a fifteen-year supracompetitive take rate, high entry barriers and an absence of significant rivals. A Google deponent made a heavily qualified observation that AdSense might be "the only major player" in its "small market."

On whether the AdSense Package is one product or two, Google noted the server and auction platform have never been sold separately and that the only identified rival, Microsoft pubCenter, is integrated the same way. Mikula pointed to internal documents showing Google perceived demand for an unbundled product, and to evidence that Google considered a "yield management" function giving access to non-Google platforms but dropped it because publishers would sell elsewhere. That was enough for a jury on separate products and actual coercion.

Inform: a video company in a display case

Inform's loss is the clearest illustration of what preclusion does not do. The company, in business from roughly 2008 to 2017, syndicated video clips to publishers including CBS, Reuters and the Associated Press and sold the ads inside them. It signed an AdX agreement on October 5, 2012 and moved from the Auditude ad server to DFP in December 2013. Inform argues Google lured it onto its tools and then starved it of ad sales. Its expert, David Kennedy of Berkeley Research Group, values the business at $479 million as of 2014.

Inform had obtained preclusion in October 2025 on the market definitions and four practices. It still lost, because those markets cover open-web display and Inform sold video. According to Google's expert Jeffrey Prince, 97.3% of Inform's AdX-filled impressions from December 2012 to May 2017 were instream video, 99.6% from December 2013, and 99.9% of its DFP impressions were video. AdX filled 1.3% of Inform's DFP-served impressions between December 2013 and November 2019 and produced about 1% of its revenue to May 2017. Inform earned more from rival exchanges, among them Rubicon, SpotXChange and BrightRoll, than from AdX.

Inform alleges Google failed to serve as many as billions of its ads per day, left $23,542,500 of impressions unserved each year and sold premium inventory valued at $15 to $20 per thousand impressions for as little as $0.80. Accepting those figures as true, the court said, they describe an initiative aimed at Inform in video ad technology, not harm flowing from Google's power in the display stack.

The online video claim failed on market power. Inform's economist, Pablo Peña, reasoned that because Google had power in ad exchanges, "it necessarily follows that Google had market power in the online video advertising market. This simply is a corollary." Castel called that "not a self-proving proposition," noting that Peña calculated no market share. (The opinion labels Peña Google's expert at one point, but its analysis treats him as Inform's.) A Google document calling YouTube "the world's great infrastructure for web and mobile video" was "internal puffery," according to the opinion. A YouTube tying claim failed for want of evidence on market power and anticompetitive effect.

Procedure did not help. Inform's opposition arrived a day late, on March 7, 2026, was revised without explanation on March 14, and was replaced on April 2 by a third version filed without leave six days after Google's reply, with 417 changes in the redline. The court struck it, while noting it would not have changed the outcome. Inform has seven days to argue for keeping its Georgia law claims for tortious interference and fraud in federal court; with Inform and Alphabet both Delaware corporations, there is no diversity jurisdiction to fall back on.

Remedies stay in Virginia

The AdX Class also seeks injunctive relief against EDA, Dynamic Allocation and Dynamic Revenue Share, which Google says it no longer uses. Castel declined to rule, citing an October 7, 2024 stipulation deferring remedies discovery and the September 2, 2026 Virginia opinion, which enjoins tying DFP to AdX, bars Google from reimplementing First Look, Last Look and Unified Pricing Rules, and requires AdX and DFP to integrate with header bidding.

PPC Land's reporting found that Brinkema rejected a sale of AdX and imposed a six-year worldwide set of conduct rules, including Prebid integrations for AdX and DFP, with a joint proposed final judgment due on October 2, 2026. Google had already removed Unified Pricing Rules from Ad Manager in December 2025. Castel said New York relief is better weighed against a trial record and what Virginia has ordered.

Why this matters for the marketing community

The Virginia remedies left Google's ad tech business intact. If there is to be a financial reckoning for conduct a federal court has already found unlawful, it runs through New York, and the September 30 opinion is the first detailed indication of how that reckoning will be measured.

The queue is long. Brinkema's remedies opinion listed private actions by PubMatic, OpenX, Magnite, Equativ, Index Exchange and Gannett, and Raptive, Dotdash Meredith, The Atlantic, Vox Media and Teads have sued too. Many lean on the same programs Google tried to keep from a jury here. The opinion shows that expert models built around those practices can survive Daubert scrutiny in this court and that foreign revenue is recoverable where the conduct was domestic. It also shows the limits: self-described sophisticated publishers will struggle with New York consumer law claims, and a business outside open-web display cannot borrow the Virginia findings.

For Gannett, which now trades as USA TODAY Co., the stakes are financial in the plainest sense. The company's digital advertising revenue fell 9.2% to $79.8 million in the second quarter of 2026, and it carries the Google suit as a gain contingency, with chief executive Michael Reed having said he expected a ruling on Google's summary judgment motion around September. For the Daily Mail, whose chief digital officer Matthew Wheatland testified in Virginia that display produces 50% to 60% of the publisher's digital revenue, more than 60% of the claimed damages fall outside the United States, which is why the foreign-commerce ruling matters to it.

Advertisers have a stake too. Bernanke adjusted the bids of Google Ads advertisers, which Li says Google concealed from them, and Poirot adjusted the bids of DV360 buyers, according to the evidence described in the opinion. The ruling does not address the advertiser class in the multidistrict litigation, but it puts internal documents about how advertiser bids were altered on a path toward a public trial, at a time when advertisers are separately pursuing mass arbitration over alleged ad overcharges.

For ad operations teams, the mitigation ruling is the detail with operational texture. Google's remaining defense depends on whether publishers could have blunted EDA with price floors, sponsorship line items or Dynamic Allocation settings. How DFP was configured years ago will now be argued over in front of jurors, line item by line item if necessary!

No trial date appears in the opinion. Google has said it will appeal the Virginia liability ruling, which underpins the preclusion these plaintiffs rely on. For now, two publishers, a class and one small AdSense customer have cleared the last major pretrial barrier, and one video company has not.

Timeline

Summary

Who: Judge P. Kevin Castel of the United States District Court for the Southern District of New York, ruling on motions by Google LLC and Alphabet Inc. against Gannett Co., Inc. (now USA TODAY Co.), Associated Newspapers Ltd. and Mail Media, Inc. (Daily Mail), a certified class of United States publishers that sold through AdX, The Progressive, Mikula Web Solutions and Inform Inc. The experts at issue include Shengwu Li, Ali Hortacsu, Einer Elhauge and Anand Das.

What: An 88-page opinion denying Google's motions to exclude plaintiffs' experts, denying summary judgment against Gannett, the Daily Mail, the AdX Class and Mikula on their antitrust claims, and granting it on the two publishers' New York General Business Law claims, on The Progressive's AdSense claims and on Inform's Sherman Act and Clayton Act claims. Claims about Enhanced Dynamic Allocation, Minimum Bid to Win, Project Bernanke and Poirot proceed toward a jury, as does the AdX Class's reversed "Act 2" tying theory.

When: The opinion was dated and filed on September 30, 2026, following the April 17, 2025 Virginia liability ruling, the October 27, 2025 preclusion ruling in New York and the September 2, 2026 Virginia remedies decision. Inform's letter brief on its Georgia claims is due within seven days.

Where: The Southern District of New York, in multidistrict litigation 21-md-3010, with the coordinated cases 21-cv-3446 (Daily Mail), 23-cv-5177 (Gannett), 21-cv-7034 (publisher class) and 23-cv-1530 (Inform). The damages claimed cover both United States and worldwide ad sales.

Why: Preclusion settled that Google broke the law in Virginia, but private plaintiffs must still prove what that cost them. The ruling determines which damages theories, worth $599,955,198 and $900,771,489 to the two publishers and $1.72 billion to the AdX Class by their experts' calculations, can be put to a jury, and it sets limits that will shape the many follow-on suits pending against Google over the same conduct.