Amazon in talks to buy Decart in a deal reported around $7 billion, indicating continued push into AI software

Amazon is again drawing attention from investors and analysts after a report said the company is in advanced talks to acquire Decart for roughly $7 billion. If completed, the deal would extend Amazon’s strategy of building out its artificial intelligence stack, not just through cloud infrastructure but also through software that can translate AI demand into efficient execution across different hardware options.

The reported target, Decart, builds software designed to help developers distribute or “route” AI workloads across competing chips. In practical terms, this kind of technology sits between model developers and the underlying hardware, aiming to improve performance or cost by matching workloads to the available accelerators rather than forcing all AI operations onto a single type of chip.

By pursuing software that manages where AI computation runs, Amazon would also be positioning itself for a market in which customers increasingly have choices among GPUs, custom accelerators, and other compute platforms. Even when companies buy AI through a cloud provider, the question of how workloads are scheduled and executed can heavily influence latency, throughput, and overall cost.

The size of the potential transaction is notable in itself, particularly because acquisitions at that scale typically suggest Amazon sees more than a small product tuck-in. The report’s framing implies Amazon believes Decart’s tooling could help it compete more effectively as enterprises and developers look for ways to get better results from heterogeneous compute environments.

Amazon has not publicly detailed the reported Decart talks in the materials provided for this story. The company also did not disclose deal terms, timing, or whether any definitive agreement has been reached, beyond the fact that the report characterizes negotiations as “advanced.”

For Amazon, the strategic logic would fit with a broader push to strengthen AWS’s role in AI. While AWS is widely associated with cloud-based compute, the competitive boundary increasingly includes the software and tooling around AI workloads, including orchestration, performance optimization, and developer-facing platforms that make it easier to deploy and scale models.

Sector-wide, the reported focus on distributing AI workloads across “competing chips” highlights a growing industry problem: AI is not running on one uniform platform anymore. Organizations are increasingly building stacks that combine different accelerators or transition between them over time, making software that can manage those differences a valuable capability.

Investors will likely watch whether Amazon confirms or denies the talks, and if so, whether the company can provide additional details about product integration, customer impact, and how the technology would be made available through AWS. The report did not provide those specifics, so the immediate question is not only whether the acquisition happens, but what Amazon plans to do with the technology once it is in-house.