Amazon is in talks to acquire Decart, an Israeli-founded AI startup, in a deal worth around $7 billion, the Wall Street Journal reported on October 11, citing people familiar with the matter. If completed, it would be one of the largest AI infrastructure acquisitions of the year — and a clear signal of where the AI arms race is heading: chips.
What Decart Does
Founded in San Francisco in 2023 by three Israeli engineers, Decart builds two things buyers are desperate for. The first is chip-portability software: most AI models are built for specific chips and need costly tuning to run elsewhere. Decart’s software removes that burden, letting developers move AI workloads across different chip architectures with minimal friction — a capability that grows more valuable as enterprises try to escape dependence on any single GPU supplier.
The second is world models: AI systems that learn the rules of the physical world to simulate 3D environments. Its Lucy model edits live video in real time; Oasis generates interactive simulations used to train and test robots and autonomous vehicles.
Why Amazon Wants It
The logic, as Reuters summarised, is vertical integration: Amazon’s custom Trainium chips plus Decart’s portability software would strengthen AWS against Nvidia in the AI infrastructure market. Amazon gave Decart early access to its Trainium3 chip back in 2025 — this courtship has been building for a while.
The price tag tells its own story. Decart was valued at nearly $4 billion in May after raising $300 million led by Radical Ventures, with Nvidia joining as an investor. Anthropic explored a $6 billion acquisition in August but walked away after due diligence in September. At $7 billion, Amazon would be paying almost double the May valuation — the kind of premium hyperscalers now routinely pay for scarce AI engineering talent and proprietary software.
What to Watch
Deals at this stage can still collapse, and another bidder could emerge — Decart has reportedly fielded interest from Nvidia, SpaceX and Nebius this year alone. But the direction is unmistakable: the AI boom’s scarcest resource is no longer models or data, but the infrastructure to run them. Whoever controls the chips — and the software that makes them interchangeable — controls the economics of the entire industry.
1 Comment
Pingback: AI Is Less Regulated Than Milk, Says Microsoft President — And He Wants That to Change - Weekly Business Records