@2026 Finance News International. All Rights Reserved.

Blends Media
A Blends Media Group Production

Amazon Reportedly Eyes AI Startup Decart in Potential $7 Billion Deal

Arry Hashemi
Arry Hashemi
Oct. 13, 2026
AmazonDecart already uses Amazon’s Trainium chips to run real-time AI video models, giving the companies an existing technical relationship. (Shutterstock)

Amazon is in talks to acquire artificial intelligence startup Decart in a deal that could value the company at around $7 billion, The Wall Street Journal reported. The negotiations involve a developer of software for running AI models across different chips, alongside technology for real-time video and interactive environments.

An agreement could be reached soon, although negotiations could break down or another buyer could emergel. The report describes a potential transaction, rather than a signed acquisition.

Decart already has a working relationship with Amazon Web Services. Amazon has identified the startup as a customer of Trainium, its custom AI processors.

Decart describes its Decart Optimization Stack, known as DOS, as software that works across NVIDIA hardware, AWS Trainium and Google TPU. It is designed to improve processing performance, cost efficiency and hardware utilization.

The work addresses a practical challenge in deploying AI: securing computing capacity and using it effectively are separate problems. A processor’s capabilities depend partly on how software organizes calculations, moves information and uses available memory. Changing hardware can therefore involve changes to the software that runs a model.

Alongside DOS, Decart develops Lucy, a model for transforming video in real time. The company lists virtual try-on, live advertising, social platforms and gaming among its applications. Its Oasis product focuses on interactive environments for physical AI, including intended uses in robotics, autonomous vehicles, manufacturing and drones.

These products place Decart on two sides of the AI business. It develops infrastructure software intended to improve how models run, while also building models with visible applications.

The Journal reported that AWS gave Decart early access to Trainium3 in 2025. Public company disclosures subsequently showed the startup using Amazon’s chips for real-time generative video, offering a clearer picture of the technical relationship preceding the negotiations.

In its December 2025 Trainium3 UltraServers release, AWS said Decart was achieving four times faster inference at half the cost of GPUs. Inference is the processing performed when a trained model responds to an input.

An AWS customer account attributed to Decart co-founder and CEO Dean Leitersdorf reports early testing with up to four times higher frame throughput and latency falling from 40 milliseconds to 10 milliseconds. Both measures are relevant to live video: throughput concerns the volume produced, while latency concerns the time taken to respond.

Decart also has substantial venture backing. Its corporate overview lists a $300 million funding round led by Radical Ventures, bringing total funding to more than $450 million.

Amazon’s latest published quarterly figures provide the financial backdrop. In its second-quarter 2026 results, the company reported AWS sales of $42.2 billion, a 37% increase from a year earlier, and AWS operating income of $16.6 billion. The release also named Decart among startups making Trainium commitments.

Amazon said its AWS AI business and its chips business had each exceeded a $25 billion annual revenue run rate.

Decart advertises support for several hardware families, while Amazon operates its own chip platform and cloud services. Whether ownership would preserve that breadth, change distribution or alter product priorities cannot be determined from the reported negotiations.