Google lost the adtech antitrust case. On 17 April 2025, a federal court found that it had illegally monopolised the publisher ad server and open-web display exchange markets, and had unlawfully tied DFP to AdX. Read the court record at the US Department of Justice.
On 2 September 2026, the same court declined to make Google sell AdX. There will be behavioural remedies, although the full opinion remains sealed at the time of writing. Google Ad Manager will still be connected to publishers. AdX will still run auctions. Most media buyers will turn up for work next week and use the same machinery they used last week. Associated Press, 2 September 2026.
It feels like an extraordinary result. A court can find that a company unlawfully protected a monopoly, then leave the commercial structure behind it largely intact.
Google gets behavioural remedies and carries on.
Any independent adtech company facing the same finding would spend the next year reassuring customers, calming investors and wondering whether it still had a business.
Meta and Amazon make the same point from different directions. The cases aren't comparable, so I won't bundle them into one accusation. What the companies share is an ability to take a blow that would flatten a smaller business without losing their place on the plan.
The penalties are serious
Meta's recent child-safety cases are not a slap on the wrist. In New Mexico, a jury and judge imposed a combined $942 million in penalties and remedies after findings under the state's Unfair Practices Act and public-nuisance law. The order also requires five years of court-supervised changes to Facebook and Instagram for young people in the state. New Mexico Department of Justice, 7 August 2026.
A separate settlement with a bipartisan coalition of attorneys general requires Meta to pay at least $12.1 billion, potentially rising to $17.1 billion, and introduces restrictions on how minors use its platforms. That settlement is subject to court approval. It resolves allegations rather than producing another trial finding, which is an important distinction when discussing what Meta has or has not been proved to have done. Office of the New York Attorney General, 26 August 2026.
Amazon's case is different again. The FTC and 22 states allege that Amazon quietly added a “soft reserve price” to advertising auctions while continuing to describe them as second-price auctions. According to the complaint, Sponsored Products advertisers were paying their own winning bid about 80% of the time by 2024. More than one million brands and sellers may have been affected. The FTC says the practice extracted tens of billions of dollars. Federal Trade Commission, 31 August 2026.
Amazon denies that it deceived advertisers and argues that the FTC has misunderstood an auction model designed to reward relevance. That case has just started. “Caught gouging” is a satisfying headline, but it is not yet a legal conclusion. Read Amazon's response.
These are consequential actions. They shouldn't be minimised simply because the companies can afford them. A company can pay a serious price while its customers barely alter their plans.
A billion isn't what it used to be
Meta generated $196.2 billion in advertising revenue during 2025. The minimum multistate settlement is roughly six per cent of that figure, before tax or accounting treatment, and payments will not necessarily arrive as a single annual charge. Painful, certainly. Existential, no. Meta 2025 results.
Amazon's advertising business produced $68.6 billion in 2025, up from $56.2 billion the year before. It has its own shopping data, Prime Video, Twitch and devices, then extends those advantages across thousands of third-party sites and apps through Amazon DSP. Amazon 2025 annual report and Amazon DSP.
The usual language of punishment can become misleading. A fine may be huge in isolation and still fit inside the annual economics of the platform. Legal costs can be absorbed over several years. Product changes can be contained by geography. Appeals stretch the timetable while the underlying business continues selling.
An independent company does not get that luxury. A serious allegation can stop a sale before any court reaches a view. Agencies add another approval. A client asks whether there is a safer supplier. Investors mark down the risk and employees start taking calls.
The largest platforms can spend years separating the legal event from the customer decision.
Everybody goes back to work
The uncomfortable part is that advertisers aren't behaving irrationally. Google, Meta and Amazon are on plans because they provide something buyers struggle to reproduce elsewhere.
Google sits inside the publisher workflow and connects an enormous pool of demand. Meta reaches people at scale through products they use habitually, with an optimisation system that can turn a relatively simple instruction into measurable activity. Amazon sits close to the point of purchase and increasingly carries its data into premium video and the open internet. The planner sees performance, reach and operational ease. An antitrust judgment sits somewhere else in the organisation.
We have watched this before. More than 1,000 advertisers joined the Stop Hate for Profit boycott of Facebook in July 2020. The campaign created genuine public and reputational pressure, but most of the boycott didn't continue into August. Axios, 2 August 2020. Facebook's third-quarter advertising revenue still grew 22% year on year to $21.2 billion. Facebook's Q3 2020 filing.
Large advertisers paused and many later returned. Smaller businesses kept spending because Facebook remained one of the few places where they could find customers without building a media department. Moral pressure met a practical dependency, and the practical dependency won.
Media planning also works against collective action. A chief marketing officer may care deeply about market concentration. Their team is still measured against this quarter's sales target. Nobody wants to explain that performance declined because the company volunteered to make the market healthier over the next decade.
Washington isn't doing nothing
I'm sceptical about describing the current US administration as indifferent. The Department of Justice asked the court to force a sale of AdX and DFP. The court rejected the structural remedy. The FTC has just sued Amazon with 22 states, while attorneys general from both parties secured the Meta settlement.
You can reasonably dislike the outcomes and question the pace. I do. It's fair to scrutinise the relationship between government and very large technology companies as well. What I can't evidence is that corporate donations explain these particular results. The known problem is already bad enough: enforcement arrives years after the conduct, courts are reluctant to dismantle integrated products and behavioural remedies are difficult to police in systems outsiders can't see.
By the time a remedy lands, the market has moved. Customers have trained teams, integrated data and designed measurement around the platform. Switching becomes a business transformation rather than a media decision.
What would actually move the market
Independent adtech won't win by reminding buyers that it is independent. That may open a conversation, but it doesn't improve an outcome or make a campaign easier to run.
The first requirement is a product good enough to justify the inconvenience of choice. Independent companies need to combine access and decisioning without recreating the frankenstack buyers are trying to escape. I don't mean another bundle of logos under one contract. The work needs to feel coherent to the person planning the campaign.
Publishers also need to make their value easier to buy. Direct relationships are useful, but a buyer can't negotiate a different workflow with every media owner. Common commercial definitions and comparable outcome evidence would let independent supply behave more like a market and less like a collection of exceptions.
Planning teams have a part in this. If concentration risk is real, it needs to appear in procurement and investment decisions before the next controversy. That could mean testing a credible independent route against a platform, preserving access to publisher data or refusing auction mechanics that cannot be explained. A one-month boycott followed by business as usual changes very little.
Regulators can create room for that competition, although the remedies have to touch the advantages that hold the market in place. Data portability matters. Non-discrimination between owned and third-party inventory matters. Independent auction audits would help when one company controls the rules and reports the result. Interoperability can reduce switching costs, provided it is implemented rather than admired in a policy document.
None of this will sink Google, Meta or Amazon, and that isn't a sensible objective for independent adtech anyway.
The useful goal is a market where losing trust has a commercial consequence because a buyer can take their money elsewhere without accepting a worse product and six months of operational pain.
I suspect the immediate future is much less dramatic. AdX continues. Meta absorbs the settlements and adjusts its youth products. Amazon defends the case while its DSP reaches further into third-party supply. A year from now, most plans will look familiar.
The only durable answer is to make the alternative easier to buy and good enough to keep. Regulation can remove some of the obstacles. It can't make advertisers choose it.