Amazon Eyes $7 Billion Acquisition of AI Startup Decart to Challenge Nvidia Dominance

Reviewed byNidhi Govil

13 Sources

Share

Amazon is negotiating a $7 billion deal to acquire Decart, an AI startup specializing in chip-agnostic software. Founded in 2023 and valued at nearly $4 billion in May, Decart's technology helps developers switch between AI chips seamlessly. The acquisition would strengthen Amazon's Trainium chips and reduce dependence on Nvidia.

Amazon Pursues Strategic AI Chip Software Acquisition

Amazon is in talks to acquire Decart in a deal that could value the AI startup at approximately $7 billion, according to

1

1

. The transaction could finalize soon unless negotiations collapse or another buyer emerges. This Amazon Decart acquisition would rank among the company's largest recent deals, trailing only its $13.7 billion Whole Foods purchase in 2017 and $11.57 billion Globalstar agreement announced earlier this year

2

.

Decart, founded in 2023 by Israeli engineers Dean and Orian Leitersdorf alongside Moshe Shalev, has raised over $400 million from investors including Sequoia Capital, Benchmark, and Radical Ventures

3

. The San Francisco-based company was most recently valued at nearly $4 billion during its $300 million funding round in May, with Nvidia joining as a new investor

1

. At the reported price, this would represent one of the largest exits in Israeli tech history

1

.

Source: SiliconANGLE

Source: SiliconANGLE

Breaking Nvidia's Grip on AI Infrastructure

Decart's core offering is DOS, a software layer that enables developers to move AI workloads between different chips with significantly reduced engineering effort

4

. This capability addresses a critical pain point in AI infrastructure: Nvidia dominates the market largely because of CUDA, software that developers have relied on for years to build and run AI models

2

. Moving workloads to alternative chips typically requires substantial additional engineering work.

With Decart DOS, the neural network optimization process that once took months can be compressed into just a couple of weeks

4

. This allows companies to shift AI workloads to different processors when requirements change or chip availability becomes constrained. For Amazon Web Services, owning this technology would make Trainium chips an easier sell to customers whose AI models were originally built for Nvidia hardware

2

.

Proven Performance on Amazon's Custom AI Chips

Decart already maintains strong ties with Amazon Web Services. AWS offered the startup early access to Trainium chips last year after meeting with its founders

1

. At the AWS re:Invent conference in December 2025, Decart's video model ran live on Trainium3, demonstrating the chip's capabilities in a high-profile keynote presentation

2

.

Source: GeekWire

Source: GeekWire

Amazon reported that Decart achieved "4x faster frame generation at half the cost of GPUs" when running on Trainium3

1

. This performance improvement validates Amazon's strategy of building custom AI chips to compete with Nvidia. The acquisition would bring the engineering team behind these optimizations in-house, potentially accelerating development of future Trainium generations.

World Models and Real-Time Video Capabilities

Beyond AI chip software, Decart develops its own world models that could integrate across Amazon's business portfolio. Lucy performs real-time video editing, transforming live video feeds to show people wearing different clothing or accessories

3

. This technology already powers features on Twitch, which Amazon owns, and could enable online shoppers to visualize how clothes look on them

2

.

Oasis generates interactive 3D environment generation for robotics training and autonomous vehicle development

3

. These synthetic environments provide critical testing grounds for physical AI systems without real-world risks. The company's customers include eBay, which is also an investor

2

.

Previous Acquisition Attempts Signal Market Value

Amazon isn't the first major player to pursue this AI startup. Anthropic was in talks to acquire Decart for approximately $6 billion in August, which would have been Anthropic's largest acquisition

3

. However, Anthropic walked away in September after conducting due diligence

1

.

Sources close to Decart reported that founders refused Anthropic's demand to relocate much of the team from Tel Aviv to San Francisco

1

. Before Anthropic, Nvidia held acquisition discussions with Decart, though the startup's founders turned down a higher offer from the chip giant before accepting their May funding round

1

. Nvidia's continued interest materialized through its participation as an investor in that $300 million round.

Broader Industry Push Away from Nvidia Dependence

The potential Amazon Decart acquisition reflects a wider industry movement to reduce reliance on Nvidia. Microsoft announced Monday that its Maia 300 chip, the third generation of its custom AI chips, will power AI services in its own products starting in 2027

1

. Google operates its TPU chips for cloud computing workloads, while OpenAI has developed processors with Broadcom

2

.

Meanwhile, Nvidia continues its own acquisition strategy. The Financial Times reported that Nvidia is in talks to buy or increase investment in Reflection AI

1

. This parallel maneuvering underscores how critical AI model performance and infrastructure flexibility have become as companies race to secure computing power.

Questions Remain About Post-Acquisition Strategy

One critical uncertainty is whether Decart's software would continue supporting Nvidia and Google chips under Amazon ownership

2

. Maintaining chip-agnostic capabilities could preserve Decart's customer base and market position, while restricting the technology to Trainium would maximize Amazon's competitive advantage but potentially alienate existing clients.

Source: PYMNTS

Source: PYMNTS

Decart has not disclosed its revenue, though reports indicate the company has been profitable almost from its inception

2

. Business publication CTech noted that Decart's current business remains far smaller than the $7 billion valuation being discussed

2

, suggesting Amazon is betting heavily on future potential rather than current revenue streams. Watch for how Amazon positions this acquisition relative to its broader AI strategy and whether it maintains Decart's multi-chip support or pivots exclusively to Trainium optimization.

Today's Top Stories

© 2026 TheOutpost.AI All rights reserved