TL;DR: Amazon is in talks to buy Decart, a three-year-old AI startup, for around $7 billion, The Wall Street Journal reported. Decart makes software that helps AI models run faster on different chips, including Amazon’s own Trainium, which could make AWS a much easier alternative to Nvidia. Neither company has confirmed the talks.
Amazon may be about to make one of the biggest acquisitions in its history, and it isn’t a retailer or a studio this time. The Wall Street Journal reported on Sunday that Amazon is negotiating to buy Decart, an AI startup, in a deal that could value it at around $7 billion. Citing people familiar with the matter, the Journal said the deal could be wrapped up soon, unless the talks fall apart.
Nothing is official yet. An Amazon spokesperson declined to comment to The Seattle Times, and Decart didn’t immediately respond to the paper’s request for comment.
What Decart actually does
Decart was founded in 2023 by brothers Dean and Orian Leitersdorf and Moshe Shalev. It’s based in San Francisco with a development center in Tel Aviv. As GeekWire explains, its main business is software that makes AI models run faster on chips from Nvidia, Google and Amazon, which it licenses to cloud providers and AI labs.
It also builds its own models. Lucy edits live video in real time, and Oasis generates interactive virtual worlds. Twitch streamers already use its technology, according to GeekWire, and Lucy can show online shoppers how clothes would look on them. Twitch, of course, is owned by Amazon.
Decart hasn’t disclosed its revenue. The Israeli business publication Globes has reported that it’s been profitable almost from the start, and Bloomberg has reported that eBay is both a customer and an investor. Still, CTech reports that the business is far smaller than the valuation being discussed.
Why Amazon would pay up
The short answer is Nvidia. Nvidia dominates AI chips in large part because of CUDA, the software developers have used for years to build and run models. Moving that work onto other hardware, including Amazon’s Trainium chips, usually takes extra engineering. Software that smooths that move is exactly what Amazon needs if it wants more AI developers renting Trainium instead of Nvidia GPUs.
The two companies already know each other well. The Journal reported that AWS gave Decart early access to its Trainium3 chip in 2025, and Decart has moved its own models onto Trainium. At AWS re:Invent last December, Decart’s real-time video model turned AWS senior vice president Peter DeSantis into a live cartoon of Amazon CTO Werner Vogels on stage, running on Trainium3. Amazon said that week that Decart was running real-time video four times faster on Trainium at half the cost of GPUs.
GeekWire draws an interesting historical line here. Amazon bought Israeli chipmaker Annapurna Labs in 2015 for a reported $350 million, and that team went on to design Trainium and Amazon’s other custom chips. A Decart deal would be roughly 20 times bigger, aimed at the software layer that makes those chips usable.
Amazon isn’t the first suitor
Decart’s sale process has been unusually crowded. According to CTech, Nvidia held acquisition talks with the company earlier, and Nvidia also invested in Decart’s $300 million round in May, which valued the startup at about $4 billion. Decart has raised roughly $450 million in total.
Then came Anthropic. Bloomberg reported in August that the Claude maker was in talks to buy Decart for about $6 billion, and Bloomberg Law notes Anthropic decided against the deal in September after doing due diligence. Globes, citing sources close to Decart, reported that the founders had refused Anthropic’s push to move much of the team from Tel Aviv to San Francisco. Anthropic’s reasons haven’t been publicly confirmed.
That makes Amazon’s interest a little awkward. Anthropic is one of Amazon’s closest AI partners. Amazon has invested about $18 billion in Anthropic and could put in up to $15 billion more, and Anthropic trains and runs Claude on Trainium chips. According to Bloomberg, Anthropic’s goal in pursuing Decart was to get more out of the chips it already uses, and GeekWire notes the startup’s software could do the same for Amazon.
How big would this deal be?
Big by Amazon standards. It paid $13.7 billion for Whole Foods in 2017 and about $8.45 billion for MGM, and in April it agreed to buy satellite company Globalstar for $11.57 billion, GeekWire notes. A $7 billion Decart deal would sit right in that tier, and The Seattle Times says it would be one of the largest exits in Israeli tech.
It also fits a broader pattern of tech giants trying to loosen Nvidia’s grip. Microsoft unveiled Maia 300, the third generation of its custom AI chip, on Monday and said it plans to make Maia the main hardware for running AI in its own products starting in 2027. Google has its TPUs, and OpenAI has developed a chip with Broadcom.
One open question, as GeekWire points out, is whether Decart’s software would keep supporting Nvidia and Google chips under Amazon ownership. That will matter to the AI labs and cloud providers that license it today.
Related: Nvidia’s $20 Billion Groq Deal Just Got Hit With a Shareholder Lawsuit and Nvidia Is Reportedly in Talks to Buy the Open-Source AI Lab It Already Backs for $25 Billion.
Bottom Line: Amazon is reportedly close to paying around $7 billion for Decart, whose software helps AI models run on chips other than Nvidia’s, including Amazon’s own Trainium. Anthropic and Nvidia both circled the startup first. If the deal happens, it would be one of Amazon’s largest ever and a clear bet on making its chips easier to choose.
Follow Teck Hustlers: Facebook · X · Instagram · LinkedIn · TikTok · Pinterest
The Teck Hustlers Editorial Team covers technology, business, finance and gaming news for readers who want the story behind the headline. Every article is researched, written and edited by our in-house team, checked against company announcements, filings and reputable reporting, and updated when new facts emerge. We focus on what each development means for founders, professionals, investors and everyday users. Have a tip or a correction? Reach us through our contact page.
