The deal, as reported
Price: around $7 billion. Source: Wall Street Journal. Unconfirmed by either company.
Target: Decart, founded 2023, Israel. ~$450M raised. Last valued at ~$4B in May 2026.
The twist: Anthropic was preferred bidder in August at ~$6-7B and withdrew in September after due diligence.
Status: could be signed soon, could collapse. Nothing is confirmed.
What Decart actually builds
Decart has two product lines, and only one of them explains a $7 billion price.
The infrastructure software. Decart builds software that makes AI models run more efficiently and lets them move between chips from different manufacturers. In practice that means getting more useful compute out of hardware you already own, and not being locked to one vendor's silicon.
The real-time generative products. Lucy is a real-time video editor. Oasis generates interactive simulated environments. Both are technically striking and both are, in acquisition terms, almost certainly not the reason.
Why Amazon wants it
AWS is already a strategic Decart customer, and gave the company early access to its Trainium3 chip in 2025. That relationship predates any acquisition talk and tells you what Amazon values.
Amazon designs its own AI silicon. The single largest obstacle to anyone adopting it is that the entire machine learning ecosystem is built around Nvidia's CUDA. Software that moves workloads across architectures without a rewrite attacks that obstacle directly.
The detail worth sitting with. Nvidia participated in Decart's $300 million round in May 2026. If Amazon buys Decart and uses it to make Trainium a credible alternative, Nvidia will have helped fund the company whose software loosens its own grip. That is not a prediction about what happens — it is simply what the cap table and the product do when you put them next to each other.
The question nobody has answered
Anthropic was the preferred bidder in August, at a price reported around $6 to $7 billion. It completed due diligence. In September it withdrew. No reason has been made public.
Three readings are available, and the honest position is that we cannot tell which is right.
- Something turned up in diligence. The ordinary reason bidders walk after looking at the books. If so, Amazon has either found the same thing and priced it in, or has a different risk appetite.
- The strategic fit was weaker than it looked. Chip-portability software is worth far more to a company with its own silicon than to one that buys compute. Amazon has Trainium; Anthropic does not make chips. The same asset is simply worth more to Amazon.
- Capital allocation. $7 billion is a different proportion of Anthropic's balance sheet than of Amazon's. A price can be fair and still be the wrong use of the money.
The second reading is the most economically coherent, and it is also the least dramatic. Beware the instinct to read a withdrawal as a red flag — buyers walk away from good companies regularly, for reasons that have nothing to do with the company.
The price trajectory
| When |
Event |
Value |
| 2023 | Founded | — |
| May 2026 | $300M round, Radical Ventures leads, Nvidia participates | ~$4B |
| Aug 2026 | Anthropic preferred bidder | ~$6-7B |
| Sept 2026 | Anthropic withdraws after diligence | — |
| Oct 2026 | Amazon in talks (WSJ) | ~$7B |
Roughly 75% appreciation in five months, on a company three years old with no public revenue figures. That is either a correct read on how valuable chip portability becomes, or it is what the top of a cycle looks like. Both have looked identical from the inside every time.
What it means if it closes
- For AWS customers: a credible path to running existing workloads on Trainium without rewriting them. That is a cost argument, and cost arguments move slowly but do not reverse.
- For Nvidia: no immediate effect. CUDA's advantage is a decade of accumulated tooling, and one acquisition does not undo it. The direction of travel is the story, not the quarter.
- For anyone self-hosting models: the interesting question is whether Decart's portability layer stays available to non-AWS customers. Acquisitions of neutral infrastructure rarely leave it neutral.
- For the market: this is a second data point, after Cloudflare buying Deno the same week, that the premium right now is on infrastructure software rather than on models.
FAQ
Has Amazon confirmed it is buying Decart?
No. The Wall Street Journal reported the talks. Neither Amazon nor Decart has publicly confirmed an agreement, and reporting indicates the talks could still fall apart.
What does Decart do?
It builds software that makes AI models run more efficiently and lets them move between chips from different manufacturers. It also makes Lucy, a real-time video editor, and Oasis, which generates interactive simulated environments.
Why did Anthropic pull out?
Not disclosed. Anthropic was the preferred bidder in August at a reported $6 to $7 billion and withdrew in September after completing due diligence. Nobody outside the two companies knows why.
How much has Decart raised?
Around $450 million in total. The most recent round was $300 million in May 2026 at roughly a $4 billion valuation, led by Radical Ventures with Nvidia participating.
Who founded Decart?
Dean Leitersdorf, Orian Leitersdorf and Moshe Shalev, in 2023. The company is Israeli.
What is Trainium?
Amazon's own AI training chip. AWS gave Decart early access to Trainium3 in 2025, which is the clearest public signal of why Amazon values the company.