The Monopoly the Court Wouldn't Break: Google's Remedy and the Rules Era
A federal court found Google illegally monopolized the internet's ad economy — then refused to break it up, ordering interoperability, data sharing, and a six-year monitor instead. The second landmark case, the second behavioral remedy. Is this how antitrust works now — and does it work at all?
By MyAudioBooks.ai ·
Listen free: The Monopoly the Court Wouldn't Break: Google's Remedy and the Rules Era
On September sixteenth, twenty twenty-six, a federal judge in Alexandria, Virginia, unsealed the most consequential antitrust remedy of the platform era, and it contained, inside its hundreds of pages, a paradox that defines the age. Judge Leonie Brinkema of the Eastern District of Virginia had already found — in a liability ruling the government celebrated as historic — that Google had illegally monopolized two of the markets that power the internet's advertising economy, the software systems through which nearly every publisher on Earth sells ad space and nearly every advertiser buys it. The Department of Justice had asked her to break the machine: to force Google to sell its ad exchange, the central marketplace of the system, and to unwind the integration that made the machine so hard to compete with. On September sixteenth, Judge Brinkema gave her answer. Google had broken the law, in two landmark cases now. Google would keep the machine, in both of them.
The remedy she ordered instead is real, detailed, and consequential in ways that will take years to measure. Google must open its advertising systems to rivals — building the technical connections that let competing exchanges and publisher tools interoperate with its own, so that a publisher can route its inventory through a competitor's pipes without losing access to Google's enormous pool of advertising demand. Interoperability, in the vocabulary of these cases, means the machine must learn to talk to the machines it was built to exclude.
It must also share bid data with the publishers whose inventory runs through it — giving the sellers of ad space visibility into the auctions their own content generates — and it must stop the self-preferencing practices the court documented, the suite of design choices by which the auction house quietly advantaged its own paddle: the last-look advantages, the unified pricing rules, the information asymmetries the liability phase laid out in detail. And it must submit to an internal antitrust compliance monitor, embedded for six years, watching whether it actually does these things. The company called the outcome a win — which it mostly was — and said it would appeal parts of the ruling anyway. The government called it substantial relief. The publishers whose businesses run through Google's pipes read the order and asked the question this article asks: when a court finds a monopoly and declines to break it, what, exactly, has been remedied?
At My Audio Books dot A I, you can create your own audiobooks from prompts, turn your documents into audio, all with one subscription, and store your items in your own personal library.
To understand the paradox, you first need to see the machine, because most of the internet runs on it and almost nobody has ever looked at it. Every time you load a webpage with ads on it, an auction happens in the fraction of a second before the page renders: the publisher's ad server — the software that manages the site's ad space — asks the market what this impression is worth, an exchange runs a real-time auction among advertisers, the winner's ad appears, and money flows from the advertiser through the exchange and the server to the publisher, minus everyone's cut. Google's business was to own all of it at once: the dominant publisher ad server, the dominant ad exchange, and an enormous advertising buyer of its own. The liability ruling found the obvious structural truth in that arrangement: when one company is simultaneously the auction house, the auctioneer, and a bidder with insider knowledge, it can tilt every auction in ways no one outside can see — and the court found Google had done exactly that, tying the server and the exchange together so that leaving one meant losing the other, and using its positions on all three sides of the market to make competition structurally impossible. The economics of the machine explain the stakes with unusual clarity. For every dollar an advertiser spends through these systems, the publisher whose content hosted the ad keeps a fraction, and the machinery keeps the rest — the tolls on the toll road of the open web, levied at every stage: the server's cut, the exchange's cut, the buyer-side cut. Independent analyses over the years have put the machinery's combined share at thirty cents of the dollar or more. For the local newspaper, the niche publisher, the independent site whose journalism the ads are supposed to fund, the toll is the difference between survival and closure — and it is levied by a company that also competes with those publishers for the reader's attention, an arrangement the publishers' trade associations have described, with unusual bluntness for polite industries, as being mugged by the landlord. This was not a gray area of innovative design. It was, the court held, the deliberate construction and maintenance of a monopoly, in the pipes that carry a large share of the money that pays for the open web.
Now the remedy question, because it is the question American antitrust has been circling for forty years. When a court finds a monopoly, the law offers two families of answers. The first is structural: break the monopoly physically — divestiture, the forced sale of the monopolized assets, so that the market gets competitors instead of rules. It is the crowbar remedy, and it is the one with the famous victories.
The second family is behavioral: leave the structure intact and prohibit the abusive conduct — connection mandates, data-sharing duties, non-discrimination rules, monitors to enforce them — so that the monopolist keeps its house but may no longer lock the doors. The structural tradition has the famous victories: Standard Oil broken into thirty-four companies in nineteen eleven, AT&T split into the Baby Bells in nineteen eighty-two, the last great American breakup. The behavioral tradition has the modern record, and the modern record is what the Brinkema decision extends: since the Microsoft case a quarter-century ago, when the trial judge ordered Microsoft broken in two and the appeals court threw the breakup out, no American court has structurally dismantled a major technology company. The government asked, in the search case against Google last year and in the ad-tech case decided this month, and in both cases the courts found the monopoly and declined the breakup. The remedy era has a name, and the name is behavioral.
At My Audio Books dot A I, you can listen to this story and thousands of others that explore the hidden science and mechanics behind the headlines.
Why do judges keep declining? The legal answer traces to the Microsoft reversal, which taught every judge since that a remedy must fit the violation rather than the ambition — that breaking a company is a remedy of last resort, reserved for conduct no rule can reach, and that an appeals court will second-guess the surgery for decades. The institutional answer is subtler: judges are generalists, and a breakup is not an order so much as an industry redesign, executed by a court that must then supervise the pieces for years. Behavioral orders let the court believe it has done something calibrated rather than crude: open the pipes, ban the tilting, install a monitor, and let the market do the rest. And the defendants have learned to make that choice easy. Google's own filings, in both of its recent cases, leaned hard on the same themes: the market is dynamic, the technology is shifting under everyone's feet — artificial intelligence is already reshaping how ads are bought and sold — and a court-ordered disassembly would punish success and freeze a moving target. The judges, twice now, have agreed with enough of that framing to keep the machine intact while condemning the tilting.
Does behavioral remedy work? The honest answer, from the evidence of twenty-five years, is: sometimes, slowly, and never the way the order imagines. Consider the immediate predecessor, because it is the same defendant and the same logic: in the search case, decided the year before, the court found Google had illegally monopolized general search — the front door of the internet, held shut by tens of billions of dollars a year in default-placement payments to browsers and phone makers — and then declined to break anything, choosing instead to ban exclusive default deals and order limited data sharing with rivals. A year into that remedy, the search market's structure is recognizable: the front door still swings the same way, the defaults are merely non-exclusive now, and the rivals' gains are measured in single digits. The ad-tech order was written with that outcome in full view, which is part of why its specificities — the named connections, the six-year monitor, the phased rebuild — read as a judge trying to write the behavioral remedy that finally behaves. The Microsoft consent decree did not break Microsoft's operating-system monopoly, but it did open the documentation and the settlement space through which a small company called Google eventually walked — the remedy worked, just late and indirectly. The details are worth recalling because they are the behavioral tradition's strongest exhibit: the consent decree — the settlement, approved by the court, in which the company accepts rules without admitting everything — forced Microsoft to disclose its interfaces, to license its protocols on reasonable terms, and to stop punishing manufacturers that shipped rival browsers. None of it dethroned Windows. All of it made the world marginally safer for the alternatives that did the dethroning — the browser first, then the search engine, then the mobile operating system that Microsoft, busy complying, missed entirely. The lesson the behavioralists draw is that remedies work through doors, not wrecking balls: open the interface, and the market finds the door. The lesson the structuralists draw is that the door took a decade to open, and the monopolist collected the monopoly rents the whole time. The European Union's more aggressive behavioral regime offers the closest thing to a controlled experiment the remedy debate has. The Digital Markets Act's mandates — gatekeepers must connect their messaging systems to rivals, share search data, allow alternative app stores and payment systems, and report their compliance publicly — produced, in its first years, a mixed ledger that both sides mine: alternative app stores exist where none did, browser choice screens are real, and messaging connections are slowly opening; and at the same time the platforms' compliance filings run to thousands of pages, their regulatory teams outnumber the regulator's staff, and the measurable shift in market share toward rivals remains, on most metrics, small. The European lesson is neither that rules fail nor that they work — it is that rules work at the speed of enforcement, and enforcement is a staffing line in a budget, while the incentive the rules are meant to cage is a profit center measured in the tens of billions. And the monitorships — the embedded watchers — have a mixed record: a good monitor with real access can change a company's behavior; a boxed-in monitor becomes a very expensive filing cabinet. The Brinkema order's design acknowledges the failure modes: it is unusually specific about which systems must connect to which rivals, it names the auction-rule changes the court expects, it gives the monitor six years rather than the usual two or three, and it phases implementation over twelve to fifteen months so the pipes can actually be rebuilt rather than cosmetically re-labeled. It is, by the standards of behavioral remedies, an unusually strong and unusually specific one. It is also, by definition, a bet that you can cage the incentive without changing the structure that produces it.
The strongest case against the breakup reading — the case for behavioral over structural, stated at full strength — begins with the observation that breakups are not magic either. AT&T's dismemberment created the Baby Bells, which spent the next three decades re-merging into a duopoly-plus of the same kind; Standard Oil's dissolution created companies that grew larger than the original; the structural victories of legend look, on a century's evidence, less like permanent restructurings than like reshufflings that the market's logic eventually re-sorted. The ad-tech machine, the behavioral case continues, is not a steel trust: it is a software system whose value to publishers comes substantially from the same integration that makes it dangerous, and severing the exchange from the server would impose enormous transition costs on the very publishers the suit was brought to protect — the small news sites whose ad revenue already hangs by a thread. The technology is, genuinely, moving: the shift of advertising into AI-mediated channels is happening faster than any court order can track, and a breakup designed for the market of twenty twenty-three would be executed into the market of twenty twenty-eight, where it might address nothing. Better, the case concludes, to open the interfaces, kill the self-preferencing, embed the monitor, and let a faster-moving market than the court do the structural work itself.
And the strongest case that the remedy era has failed — stated with the anger the publishers feel — is that the monopolist has now been convicted twice, in two landmark cases, and the sentence, both times, is a compliance program. The whole point of a structural remedy is that incentives follow structure: as long as Google owns the auction house, the gavel, and the largest paddle in the room, every behavioral rule is a rule the owner has every incentive to interpret, route around, and out-wait, and the monitor, however good, leaves in six years while the structure is forever. The behavioral track record in tech is the defendant's résumé: Microsoft out-waited its decree and dominated the next era anyway; the platforms absorbed their European fines as a cost of doing business; and the companies learned the deepest lesson of the remedy era — that the worst realistic outcome of a lost antitrust case, in America, in this century, is a period of supervised inconvenience followed by a return to business as structured. The deterrent math, this case concludes, is now part of the business model: monopoly profits for a decade, a behavioral remedy at the end, and an appeal to shorten the monitor. If that is the equilibrium, the remedy era is not antitrust enforcement. It is antitrust theater, with the court in the role of conscience and the monopolist in the role of itself.
Three developments would disprove or confirm which reading of the Brinkema order history ratifies, and each is observable in the years directly ahead. First, the interoperability reality: if, within the implementation window, the connections to rival exchanges and publisher tools are genuinely built and genuinely used — if a mid-sized publisher can route inventory through a competitor without losing Google's demand — then the behavioral bet has produced a working market opening, measurable in the rivals' share. Second, the appeal: Google has said it will appeal parts of the ruling, and the appellate courts' treatment of the remedy package will either cement the behavioral era as settled doctrine or force the structural question back open. Third, the AI transition itself: if the advertising market's center of gravity moves to AI-mediated buying faster than the order's mechanisms can follow, the case becomes the test of whether twentieth-century remedies can govern twenty-first-century markets at all — and the answer will be written in whether anyone is still fighting over the pipes the order opened.
It is worth saying what this article has not claimed. It has not claimed the Brinkema order is toothless; it is unusually specific, unusually long, and backed by a six-year monitor, and dismissing it as nothing would be wrong. It has not claimed breakups are painless or always effective; the structural record is mixed, and the article says so. It has not claimed Google is uniquely villainous; the conduct the court documented is the conduct its structure invited, and other companies in the same position have behaved the same way. And it has not claimed the case is over; the liability ruling and the remedy both face appeal, and the final shape is years from settled. The claim here is narrower and more structural: the American courts have now established, through two landmark Google cases, that monopolies in the platform era will be found and not broken — and whether that doctrine is wisdom or surrender is the defining antitrust question of the decade.
Which returns to the paradox in the courtroom in Alexandria: the machine condemned, the machine kept. There is a reading of American legal history in which September sixteenth, twenty twenty-six, is the day the remedy era reached its logical conclusion — the courts as referees of conduct rather than architects of structure, the monopolies managed rather than dissolved, the largest companies in the history of the world instructed to share nicely and watched while they do. And there is a reading in which it is the day the theater was formalized: the monopoly established by evidence, condemned by law, and preserved by remedy, with the publishers whose businesses hang from the pipes told that the pipes will now be monitored. The truth, as is so often usual in this era, will not be decided in the courtroom. It will be decided in the implementation window, in the rivals' market share, in the monitor's reports — in whether a rule can ever do the work of a crowbar, when the thing being ruled is the machine itself.
At My Audio Books dot A I, you can create fiction, non-fiction, and turn your documents into audio, all stored in one place with a single subscription — plus get instant access to thousands of audiobooks and deep-dive investigations. Learn more today at My Audio Books dot A I.