The Federal Trade Commission is filing suit against Amazon in Seattle federal court today, joined by a bipartisan coalition of more than 20 state attorneys general, alleging that Amazon spent seven years secretly raising the minimum price businesses had to pay to advertise on its retail platform. The Wall Street Journal's Dave Michaels broke it at 2:00 p.m. Eastern, working from a briefing by FTC officials ahead of the filing. As of this writing the complaint is not public and the FTC has posted nothing to its own newsroom.

The number the agency is putting on it: tens of billions of dollars, captured by Amazon, over roughly seven years.

Here is the part that matters, and it is not the number.

Amazon has historically run its sponsored ad auctions on second-price logic.

You bid five dollars, the runner-up bids two, you win and you pay about two dollars and a penny.

That format exists for a specific reason. It exists to encourage advertisers to bid their honest maximum without being punished for it, and it has the effect of holding the clearing price down.

Starting in 2018, according to the FTC, Amazon began entering its own bid into those auctions. Internally it was called a soft reserve.

Amazon set it above the runner-up's price. The winning advertiser still won, but now paid Amazon's number instead of the competitor's number. Amazon could do this because Amazon, as the operator of the auction, could already see every bid in it.

And Amazon did not tell anyone it was doing it.

Read that once more, slowly, because the trade press is going to spend the next week calling this a story about price floors, and it is not.

A floor is a threshold. This is a participant. Amazon is alleged to have looked at the live bid stack on inventory it owns, on a marketplace it operates, in an auction it runs, and then inserted a bid of its own positioned just above the highest real competitor, so that the winner would pay more. The advertiser sees a CPC. The advertiser does not see the bid stack. There is no observable difference between "a rival bid $3.75" and "Amazon said a rival bid $3.75."

The FTC says Amazon's ad executives tracked the surcharge the strategy earned and worked to limit how many people knew about it. And it says they deployed it first on high-demand shopping days, when merchants would assume that a spike in cost-per-click was the ordinary consequence of everyone else bidding hard for the same Prime Day shopper.

That is the single most damaging allegation in the entire briefing, and it is the one nobody is leading with. It is not a pricing decision. It is a concealment schedule. You launch the mechanism on exactly the days when the anomaly is invisible, because the buyer already has a story ready to explain it to himself.

In recent years, the FTC says, Amazon intervened to raise the minimum price in 70% to 80% of auctions. On major shopping days, the agency will allege, the practice raised pay-per-click costs by 50%.

Set that against the growth curve. In 2018, the year the FTC says this started, eMarketer put Amazon's advertising business at roughly $4.6 billion. In 2019 the company's own reporting showed about $14.1 billion. Last year it was $68 billion, per Amazon's securities filings, making it the third-largest digital advertising platform on earth behind Google and Meta. Average Amazon CPC ran about $0.71 before 2020. It runs about $1.20 now.

Amazon will say that curve is demand. The FTC is going to argue that a measurable slice of it is a bid Amazon placed against its own customers.

Amazon currently tells advertisers, on a public page updated in April, that it uses reserve pricing and that reserve pricing may affect the cost of your ad. The company's help material goes further than that. It says reserves can be influenced by predicted likelihood of sale, predicted return on ad spend, competing bids, placement, context and predicted performance, and it acknowledges that the price charged may exceed the runner-up's bid, though not the advertiser's own authorized maximum.

That is a real disclosure. It is also, read in the right light, a confession with a date stamp on it.

Bloomberg revealed the existence of this probe in 2025. The FTC had a draft complaint by June of 2026. And in April of 2026, a page explaining that Amazon's reserves can push your price above the runner-up's bid gets updated. Somebody should pull every Wayback Machine capture of that URL going back to 2018 and publish the diff. The diff is the story about the story.

Because the FTC's theory does not require Amazon to have said nothing. It requires Amazon to have created a misleading impression that prices were competitively cleared while the operator was materially setting them. A disclosure written in 2026 does not retroactively inform a merchant who was buying keywords in 2019.

This becomes the FTC's third major action against Amazon, and the second consumer-protection case in twelve months. In September 2025 the company paid $2.5 billion to settle the Prime dark-patterns suit, $1 billion of it a civil penalty, the largest in agency history, $1.5 billion in consumer refunds. The separate monopolization case, which already contains unsealed allegations that Amazon deliberately accepted junk ads to raise revenue, is headed for trial. Amazon also faces a certified consumer class of roughly 288 million people in the same courthouse.

Amazon shares fell on the news.

States expected to join include New York, California and Florida. The presence of more than twenty attorneys general is not a press-release flourish. It is where the money is, and Part Two of this piece explains why.

We have not read the complaint. Nobody has. Everything above is the FTC's own characterization of its own case, delivered to one reporter under embargo, and it will look different in a filed document with exhibits attached.

Open questions that will decide how big this actually is: which ad formats are covered, sponsored products alone or sponsored brands, display and DSP inventory too. What the legal theory is, deception under Section 5, unfairness, state consumer-protection statutes, or some braid of all three. How the overcharge is calculated, whether by comparing charged price against the runner-up bid, by modeling counterfactual auctions, or out of Amazon's own internal experiments. Whose names are on the documents. And whether the FTC has an internal launch memo, because the difference between a bad auction design and a seven-year deception is entirely a question of who approved what, knowing what.

There is one more question, and it is the one the rest of this industry should be reading over its glasses about. The FTC opened a parallel investigation in 2025 into Google's advertising disclosures on materially the same theory. If Amazon's motion to dismiss fails, that case is next. And after that one, every platform running a dynamic floor it has never fully explained to a buyer has a calendar problem.

Breaking story. Based on the Journal's account of FTC officials' pre-filing briefing, Amazon's filings and help documentation, and the Texas AG complaint against Google; all characterizations of the case are the government's. Comment requests are with Amazon and the FTC; dated updates to follow. ADOTAT has no financial relationship with Amazon. I hold less than $100 in stock in Amazon