Key Takeaways

  • Amazon is reportedly negotiating to acquire Decart for about $7 billion, but no agreement is final.
  • Decart could strengthen Amazon Web Services by making AI workloads more portable across NVIDIA GPUs, Trainium chips, and other processors.
  • The reported price represents a roughly 75% premium to Decart’s nearly $4 billion valuation in May 2026.

Amazon.com Inc. is reportedly negotiating to acquire Decart, an artificial intelligence startup focused on chip portability and generative 3D environments, for approximately $7 billion. The potential transaction would give Amazon technology that could address two increasingly important parts of the AI market: reducing dependence on a single class of processor and supporting new kinds of interactive content.

No final agreement has been reached. Reuters reported it could not independently verify the negotiations, which originated with The Wall Street Journal. Terms could change, negotiations could continue for some time, or the parties could walk away.

The reported talks offer a revealing look at Amazon’s priorities. Bloomberg also reported the negotiations, while The Wall Street Journal described Decart as developing software that helps developers move AI workloads among different chips. Decart also builds models capable of generating 3D environments.

That first capability could be particularly valuable to Amazon Web Services. AI developers often optimize models and applications around specific hardware, especially NVIDIA GPUs. Moving those workloads to a different processor can involve engineering work, performance testing, and changes to software libraries. Decart’s technology could potentially lower some of that friction.

For Amazon, that creates an obvious strategic angle. AWS is investing in Trainium, its family of chips designed for AI training, while continuing to provide customers with access to NVIDIA hardware. Software that makes workloads easier to move could encourage customers to test Trainium without treating the decision as an irreversible architectural commitment.

Enterprises generally want options, even when they are not planning an immediate migration. Portability can improve negotiating leverage, reduce concentration risk, and let technical teams match different workloads with different combinations of cost and performance. Decart could serve as a bridge between the dominant NVIDIA ecosystem and Amazon’s own silicon, providing a plausible rationale for the reported interest.

The valuation underscores how much Amazon may be willing to pay for that position. Decart raised $300 million in May 2026 in a round led by Radical Ventures, with NVIDIA participating. PitchBook data valued Decart at nearly $4 billion at the time (source). A transaction near $7 billion would therefore represent a premium of roughly 75% within only a few months.

Such a jump comes amid rapid growth in enterprise AI spending. Menlo Ventures estimated that U.S. enterprise generative-AI spending reached $37 billion in 2025, up from $11.5 billion in 2024. Applications accounted for $19 billion of the 2025 total, suggesting that value is spreading beyond foundation models and infrastructure into tools that businesses can deploy more directly.

Demand is broadening as well. McKinsey’s 2026 global survey found that nearly nine in ten organizations regularly use AI in at least one business function, although only 44% are scaling AI across the enterprise. This gap underscores that while experimentation is common, reliable deployment across departments, data environments, and infrastructure remains challenging.

Gartner forecast worldwide AI spending of $2.67 trillion in 2026, including $42.3 billion for AI-optimized infrastructure as a service. Against that backdrop, Decart’s portability technology could help AWS compete for workloads that might otherwise remain tightly coupled to a particular chip ecosystem.

The second part of Decart’s portfolio, generative 3D environments, adds another dimension. Such models could support gaming, simulation, robotics, design, training, and interactive media. They could also require considerable computing capacity, potentially creating additional demand for AWS infrastructure.

Anthropic reportedly considered acquiring Decart before abandoning the idea. Amazon’s interest suggests Decart’s technology may have strategic value across several layers of the AI stack, not merely as a standalone developer tool.

An acquisition would bring integration and governance questions. Amazon would need to clarify how Decart supports multiple chip architectures while sitting inside AWS, which has strong incentives to promote Trainium. Enterprise customers would also examine model risks, data controls, and generated-environment safeguards using approaches such as the NIST AI Risk Management Framework, its Generative AI Profile, and ISO/IEC 42001:2023.

For now, the transaction remains a negotiation rather than a completed acquisition. If it proceeds near the reported price, Amazon would be making a sizable bet that the next phase of AI competition will involve not only more computing power, but also greater freedom to move workloads across it.