The Wall Street Journal reported on October 11 that Amazon is negotiating to buy AI startup Decart for about $7 billion, citing people familiar with the talks. Reuters said it could not independently verify the report. Amazon declined to comment outside regular business hours, and Decart did not respond to a request for comment.

No deal has been announced, and the talks could still collapse. The reported price has drawn attention to Decart’s software for running AI workloads across different chips, not only to its video and simulation models.

The reported value is about 75% above Decart’s nearly $4 billion valuation in May, but it is not a confirmed offer price.

Why the software matters

Decart’s DOS platform is designed to optimize AI training and inference across hardware from companies including Nvidia, AMD and Amazon. That could be relevant as Amazon builds its own Trainium chips. Software that helps workloads run across different processors may give cloud customers more options, though that is a potential strategic fit, not a confirmed reason for the talks.

Decart also makes Lucy, which edits video in real time, and Oasis, which generates simulated environments for robotics and autonomous-driving tests. The company describes DOS as infrastructure behind its models. Its software therefore spans both live video editing and the training and testing of physical AI systems.

What the reported figures show

The reported numbers put the potential deal in context:

  • Decart was valued at nearly $4 billion in May, The Wall Street Journal reported.
  • The company raised $300 million in that round, according to SiliconANGLE.
  • The reported Amazon talks value the company at about $7 billion. That is roughly 75% above the May valuation, based on the reported figures, not a confirmed offer price.
  • Anthropic explored an acquisition worth about $6 billion, then decided against pursuing it after due diligence, Bloomberg Law reported.

For now, the $7 billion figure describes reported negotiations, not a signed deal. Whether Amazon is willing to pay that much for software designed to work across chip platforms remains unsettled.