Broken-exit watchlist: re-pricing a failed $6B deal

The spread is the entry fee

Anthropic walked away from a roughly $6 billion deal to acquire Decart after completing due diligence.

Decart had raised a $300 million Series B in May 2026, led by Radical Ventures, at a valuation of approximately $4 billion, so the price was a roughly 50% premium.

The spread is the useful number. It shows how much a strategic buyer was willing to pay for speed, certainty, and a solved cost problem. It also shows how much of that price was never in the company's last funded value.

By the end of July, Anthropic's annualized revenue run rate passed $65 billion, up from $47 billion in May and about $9 billion at the end of 2025.

Anthropic's computing costs were 71 cents per revenue dollar in Q1 2026, and the company projected the ratio would fall to 56 cents in Q2 2026, with a projected Q2 operating profit of about $559 million on roughly $10.9 billion in revenue.

A strategic purchase made sense if the target could close the cost gap quickly enough. A walk made sense if diligence showed the gap was wider than the model assumed.

The diligence gap is the discount

Decart was registered on September 7, 2023, by Dean Leitersdorf, Moshe Shalev, and Orian Leitersdorf.

Radical Ventures said Decart's DOS 2.0 can process more than 1,600 tokens per second, while Decart placed the industry average at about 200 tokens per second.

Performance claims are the numbers a strategic buyer tests. Those numbers can survive a term sheet and still fail in diligence.

The discount is the part of the difference between the last anchor and the failed premium that diligence failed to prove.

Speed claims that fail in diligence pull the re-price toward the last anchor. Integration problems can leave it higher. A buyer's changing cost curve may put it in the middle. The diligence gap tells you which case you are in.

Live relationships decide whether the discount is real

A broken term sheet changes one buyer's model. The next buyer may have a different cost curve, a different product need, or a different deadline. The investor who can call a few of those buyers soon has a better re-price than the one who waits for a public signal.

Before you call, write down the buyer's likely objection in plain language, then match the request to it: a benchmark test for speed, a reference from a similar deployment for integration, and the buyer's own cost curve for a cost objection, including the point at which it changes.

Then ask each buyer what would make a similar deal work for them. The answer is a condition you can test, price, and explain to the company.

The watchlist makes the re-price testable

Use the Decart record as a template, and keep the entries together.

• Last anchor: can you name the last funded valuation and the date it was set?
• Failed premium: can you calculate the gap between that anchor and the collapsed offer?
• Diligence gap: can you list the specific finding that changed the buyer's view?
• Live relationships: can you name buyers who would still talk?

The anchor entry keeps you from pricing on hope, the premium entry from pricing on an abandoned model, the gap entry from repeating the same mistake, and the relationships entry from treating a broken exit as a market verdict.

Update the page after every call. A credible speed claim moves the re-price toward the failed premium. An open cost gap moves it toward the last anchor. If no one will talk, the re-price is not ready.

Write the number down only after you can say which entry moved it. When the next strategic buyer calls, your re-price should already be on the page. It should be explained by the diligence gap. If you cannot explain it, do not put the number in front of a new buyer.