Antitrust Grows Up
Platform cases test access, exclusion and market structure alongside price effects. The hardest question may be what remedy courts will accept.

Editor's note, September 20, 2026: This piece originally named no cases, courts or companies; on 20 September 2026 it was anchored to the proceedings it describes (the United States v. Google search and ad-tech cases, Epic v. Google and FTC v. Meta) with links to the opinions, and a bracketed update notes that in September 2026, after publication, Judge Brinkema declined to order a divestiture in the ad-tech case. The argument is unchanged. Further review September 20, 2026: Clarified that the 2023 Merger Guidelines are non-binding and retain concentration analysis; multiple platform roles do not mechanically determine legal exposure.
For most of the last two generations, American antitrust law ran on a single diagnostic test. If a merger or a business practice made consumer prices go up, it was suspect. If prices stayed flat or fell, the conduct was presumptively fine, whatever else it did to rivals, workers, or the shape of a market. This was the consumer-welfare standard — the Supreme Court borrowed the phrase from Robert Bork in 1979 — and it had the virtue of being measurable: an economist could put a number on it, a judge could apply it, and a company’s lawyers could plan around it years in advance. It also had a blind spot roughly the size of the modern digital economy, where the most consequential products are given away free and the harm shows up not in a receipt but in what disappears from the shelf, the feed, or the labor market. The current wave of cases against large platform companies is best understood as enforcers trying to patch that blind spot without a legal architecture built to hold the patch in place.
From price effects to structure
The doctrinal shift is real, and it did not start with the current administration. The intellectual groundwork goes back to law review arguments from the mid-2010s — Lina Khan’s 2017 Amazon’s Antitrust Paradox is the usual citation — that a price-only framework simply could not evaluate a firm that subsidizes one side of a market to extract rents from the other, or that uses a dominant position in one layer of the stack to foreclose competition in an adjacent one. What changed over the following decade is that this critique moved from academic sympathy to prosecutorial practice. The government’s search case against Google, filed in October 2020 under the first Trump administration and won on liability in August 2024 (“Google is a monopolist, and it has acted as one,” Judge Amit Mehta wrote); Epic Games’ app-store case against Google, whose jury verdict and injunction the Ninth Circuit affirmed in July 2025; and the Justice Department’s ad-tech case, in which Judge Leonie Brinkema found in April 2025 that Google had monopolised the publisher ad-server and ad-exchange markets and unlawfully tied the two — all have leaned less on “did this raise the price consumers pay” and more on structural questions: does the firm control both sides of a transaction and referee its own participation in it, does it own the pipes and compete with the traffic running through them, and can it use data or default positions accumulated in one product to entrench dominance in another. The remedies sought have moved in parallel — not fines calibrated to overcharge, which is the traditional currency of price-based cases, but divestitures, mandated interoperability, and in the more aggressive filings, outright separation of a business line from the parent: the government asked Judge Mehta to make Google sell Chrome, and asked Judge Brinkema to make it sell its AdX exchange.
That is a genuinely different theory of harm, and it is worth being precise about why it emerged where it did. Advertising technology is the clearest illustration. A firm that operates the tool publishers use to sell inventory, the tool advertisers use to buy it, and the exchange that clears the trade between them — which is what Brinkema found Google to be, its DFP ad server and AdX exchange unlawfully tied together — is not merely a large player. It is positioned to set the terms of a market it also participates in, in ways that a price-effects test is not well equipped to see, because the distortion shows up as fee structure and auction mechanics rather than list price. Regulators concluded, reasonably, that policing conduct like this piecemeal — one settlement over auction rigging, another over self-preferencing — could not keep pace with how quickly the underlying architecture could be reconstituted. Structural separation was the response: if the conflict of interest is inherent to owning all three legs of the stool, take away one of the legs.
The doctrinal question courts are now wrestling with is not whether platform power is real, but whether the remedy for it fits inside the antitrust statute at all.
What the platform cases actually signal
For platform businesses more broadly, the signal is not “get bigger, expect a breakup.” It is narrower and, for compliance planning, more useful: multiple roles in a value chain can create theories of competitive harm that market share alone does not resolve; liability still depends on the facts and the applicable law. A marketplace acting as intermediary, seller and buyer raises questions about conflicts, access and exclusion. Counting those roles is not a legal test or a reliable ranking of litigation exposure. This is a break from the merger-guidelines logic that dominated the 1980s through the 2000s, where horizontal share was the master variable; the agencies’ 2023 Merger Guidelines broaden the stated analytical frameworks while retaining concentration analysis. They are non-binding agency guidance, not a replacement for statute or binding judicial doctrine. It rewards companies for staying in their lane and penalizes vertical integration in ways that will surprise executives who built their moat on exactly that integration, because for a long stretch of the consumer-welfare era, owning more of the stack was the textbook way to lower costs and therefore prices — precisely the thing the old test rewarded.
It also changes how these cases get litigated, because structural remedies invite a different kind of defense. Under a price-effects regime, a defendant’s best argument is usually empirical: show the econometrics, show that output rose and prices fell. Under a structural theory, the defense shifts to a design argument — that the conflict the government describes as inherent is actually manageable through firewalls, information barriers, or contractual commitments short of divestiture. Expect to see a lot more of that argument in the next few years, because it is the one most likely to find a sympathetic judge, for reasons that get to the second half of this story.
The gap between ambition and what courts will uphold
Here is the tension nobody entirely wants to say out loud: American antitrust statutes were not written with structural break-up as their default remedy, and the courts asked to bless these newer theories are, by composition and training, more comfortable with the price-effects framework than with the market-architecture one. A judge asked to order a divestiture has to be persuaded not just that a conflict of interest exists, but that severing a business line is a proportionate response under the specific remedial powers Congress granted — a much higher bar than finding liability in the first place. The historical base rate for successful structural relief in monopolization cases, outside of the handful of landmark examples — Standard Oil in 1911 — usually cited precisely because they are exceptional, is thin; the last time a court ordered a technology company broken up, in Microsoft in 2000, the D.C. Circuit vacated the order within a year. Courts have repeatedly shown they will find conduct unlawful while declining to order the remedy the government actually wants, opting instead for behavioral commitments — the firewalls and reporting obligations the defense bar prefers — because those are easier to write into a consent decree and easier for a court to supervise without effectively running a company. That is exactly what happened in the search case: in his remedies opinion of September 2025 Judge Mehta refused to order Chrome sold (the plaintiffs, he wrote, had overreached) and instead barred exclusive default contracts and ordered Google to share search data with qualified rivals — a judgment entered in December 2025 and appealed by Google in January 2026. The FTC’s attempt to unwind Meta’s purchases of Instagram and WhatsApp did not even reach the remedy stage: Judge James Boasberg ruled in November 2025 that the agency had failed to prove Meta currently holds monopoly power. (Since this piece was published, Judge Brinkema has completed the set: on 2 September 2026 she declined to order the AdX divestiture, calling structural relief neither realistic nor needed, and imposed behavioral rules instead — interoperability with rival ad servers, bid-data sharing, a six-year technical monitor — in an opinion unsealed in mid-September.)
That gap between enforcement ambition and judicial appetite is where this era of antitrust will actually be decided, and it cuts against tidy narratives on both sides. It is not simply that regulators are overreaching against a hostile judiciary, nor that courts are captured apologists for concentrated power. It is that structural theories of harm are conceptually sound responses to a real problem — markets where a single firm plays several conflicting roles at once — while remaining underdeveloped as litigated law, tested in only a handful of full trials rather than settled by consent decree. Every case that goes the distance rather than settling adds real precedent in either direction, and the appellate courts hearing the current cohort of appeals will do more to define enforceable antitrust doctrine for the next decade than any settlement, however large the headline penalty. The prices question, in other words, has not disappeared. It has simply stopped being the only question that matters, and the law is still working out how to ask the others without the answer collapsing back to the one test everyone already knows how to grade.
Sources
- Reiter v. Sonotone Corp., 442 U.S. 330 (1979); Standard Oil Co. of New Jersey v. United States, 221 U.S. 1 (1911); United States v. Microsoft Corp., 253 F.3d 34 (D.C. Cir. 2001)
- Lina M. Khan, Amazon’s Antitrust Paradox, Yale Law Journal (January 2017)
- United States v. Google LLC (D.D.C., No. 20-cv-3010, search): Memorandum Opinion on liability (5 August 2024); Plaintiffs’ revised proposed final judgment (7 March 2025); Memorandum Opinion on remedies (2 September 2025); Final Judgment (5 December 2025); DOJ case page; Tech Xplore on Google’s appeal (January 2026)
- United States v. Google LLC (E.D. Va., No. 1:23-cv-108, ad tech): Memorandum Opinion on liability (17 April 2025); DOJ press release (17 April 2025); DOJ case page; AdExchanger on the remedies decision (16 September 2026); National Law Review on the remedies decision (18 September 2026)
- Epic Games v. Google, Ninth Circuit opinion (31 July 2025)
- Sullivan & Cromwell on FTC v. Meta (decision of 18 November 2025)
- DOJ and FTC, 2023 Merger Guidelines (18 December 2023)