Amazon is in advanced talks to acquire Decart, the Israeli AI startup backed by Nvidia, for roughly $7 billion, the Wall Street Journal reported Sunday. If the deal closes, it would be one of Amazon’s biggest acquisitions in recent years. The question worth asking this morning is not what Decart does. It is why Amazon is structuring this as a clean, outright purchase rather than the licensing and talent arrangements that have become the tech industry’s preferred way to absorb AI companies without triggering antitrust review.
The deal history here matters. Nvidia had entered exclusive talks to buy Decart outright. Those talks fell apart once a bigger offer arrived. Decart then entered acquisition talks with Anthropic at about a $6 billion valuation, and the two sides conducted due diligence. In early September, Anthropic walked away, Bloomberg reported. That left Amazon as the next major suitor, now bidding above Anthropic’s reported level.
So what does Decart actually offer? Its infrastructure technology was central to prior acquisition interest. The company develops the Decart Optimization Stack, which is designed to increase AI training and inference performance across different chips. It also develops real-time world models including Lucy and Oasis. The software solves a migration problem: it makes Amazon’s own Trainium chips a more credible alternative to Nvidia GPUs, lowering the barrier for AWS customers who want to switch away from Nvidia. Decart has publicly emphasized cross-chip support, but the specific claim that it pushed real-time video from 26 frames per second on GPUs to roughly 101 frames per second on Trainium has not been substantiated in public reporting, so treat performance anecdotes here as directional rather than definitive.
The antitrust angle is where this gets interesting. Amazon chose a full purchase over the acqui-hire or licensing structures that rivals have used to sidestep HSR filings. That is a deliberate signal. Amazon’s earlier Adept arrangement was criticized precisely because it was structured around hiring and a nonexclusive technology license that did not require a premerger filing. Antitrust scrutiny is a real consideration here. Regulators in the United States and European Union have shown increased willingness to examine large technology acquisitions affecting AI infrastructure markets. A deal giving Amazon exclusive control of chip-portability software used by competitors could attract that attention. The FTC’s approach to AI-adjacent acquisitions has been cautious since 2023, and a multibillion-dollar price tag will not escape notice.
Yet Amazon appears to be making a calculated bet that a transparent, fully filed acquisition is easier to defend than a creative structure that invites the kind of congressional and regulatory backlash its Adept deal produced. With an FTC monopolization bench trial currently scheduled for February 9, 2027, Amazon is already operating under heavy scrutiny in Washington. Adding a contested acqui-hire to that docket would be a different kind of exposure than a straightforward HSR filing on a pure acquisition.
What to Watch
- AMZN: A deal could be wrapped up in the coming days, assuming discussions do not break down or another suitor does not emerge. Watch for any formal announcement, which would trigger the HSR clock.
- NVDA: If Amazon internalizes more AI workload via Trainium and adds Decart’s chip-switching tooling, it reduces the urgency for customers to standardize on Nvidia-only stacks. Even if Nvidia remains dominant, incremental share pressure from hyperscalers building credible alternatives can compress expectations at the margin.
- Valuation gap: Decart’s proposed $7 billion price is a notable jump from its valuation of about $4 billion in May 2026. That is a roughly 75% premium over a five-month-old funding round, and it is the kind of step-up regulators will notice when assessing competitive impact.
