Amazon, Nvidia step up AI bets: Inside the reported $7B Decart deal and Reflection AI talks

Amazon is in talks to acquire artificial intelligence startup Decart in a deal that could value the company at around $7 billion, while Nvidia is exploring a potential acquisition or deeper investment in US AI startup Reflection AI, according to reports.

The discussions highlight how major technology companies are seeking to strengthen their positions in the artificial intelligence market through acquisitions, investments and partnerships with emerging AI developers.

According to a WSJ report dated October 11, Amazon's negotiations with San Francisco-based Decart could result in a deal soon, although the discussions could still fall apart or attract another potential buyer.

Separately, the FT reported on October 10 that Nvidia was in early-stage talks with Reflection AI about several possible arrangements, ranging from a full acquisition to an acqui-hire, a further equity investment or an agreement to supply additional chips and computing capacity.

Neither potential deal has been confirmed, and the discussions remain subject to change.

Amazon's proposed $7 billion deal for Decart

Decart develops software that helps developers switch between different AI chips, a capability that is becoming increasingly important as technology companies compete for computing resources to train and operate AI models.

Founded in 2023 by Israeli engineers and brothers Dean and Orian Leitersdorf, along with Moshe Shalev, the startup has raised more than $400 million in venture capital funding from investors including Sequoia Capital, Benchmark and Radical Ventures, according to PitchBook data cited by the WSJ.

Decart was valued at nearly $4 billion in May, the report said. A potential acquisition at around $7 billion would represent a substantial increase over that valuation.

The company is also developing world models, including Lucy and Oasis. These models can generate and simulate three-dimensional environments representing aspects of the real world.

Amazon Web Services (AWS) gave Decart early access to its Trainium3 AI chip in 2025 after meeting the startup's founders, the WSJ reported. Amazon has developed its own AI chips to compete with Nvidia, whose graphics processing units (GPUs) remain central to the AI computing market.

A deal would add to Amazon's investments in technology and content businesses. Its major acquisitions in recent years include the $6.5 billion purchase of MGM in 2022 and a roughly $11 billion deal for satellite communications company Globalstar announced earlier this year, according to the WSJ.

Nvidia explores acquisition or investment in Reflection AI

Nvidia is considering several ways to deepen its relationship with Reflection AI, a US startup developing open-weight artificial intelligence models, the FT reported.

The discussions are at an early stage and could lead to an acquisition, an acqui-hire, a fresh equity investment or an agreement under which Nvidia provides additional chips and computing power, according to people familiar with the matter cited by the publication.

An acqui-hire typically involves a company hiring a startup's employees and obtaining rights to its technology, rather than acquiring the entire business. Such arrangements have attracted attention as technology companies pursue AI talent and capabilities while navigating potential regulatory scrutiny of large acquisitions.

Nvidia has already invested $800 million in Reflection AI, the FT reported. The startup was last valued at $25 billion in a March funding round.

Reflection is backed by venture capital firms including Sequoia Capital, Disruptive and 1789 Capital, where Donald Trump Jr, the US president's son, is a partner.

The FT reported that an agreement could be reached in the coming weeks, although people familiar with the discussions cautioned that the talks might not result in a deal. Nvidia and Reflection declined to comment, according to the news outlet.

Why Nvidia is interested in open-weight AI models

Reflection is developing open-weight AI models that can be customised and run on users' own infrastructure. Such models can offer businesses greater control over deployment and potentially lower operating costs than some proprietary AI systems.

The startup is also strategically relevant to US efforts to compete with Chinese AI developers, including DeepSeek, according to the FT.

Reflection recently unveiled its first public, customisable AI product, which runs on Nvidia chips. The company claims its model is comparable to China's Z.ai GLM-5.2, a leading competitor.

Its founders, Misha Laskin and Ioannis Antonoglou, previously worked on Google's Gemini AI models. Reflection has also partnered with the Pentagon and the Department of Energy and reached agreements to develop AI models for US allies, including South Korea, the FT reported.

Nvidia has increasingly backed open-weight AI development as part of its broader strategy to expand the market for AI computing. In July, chief executive Jensen Huang called for a domestic open-weight AI ecosystem, saying it was important for innovation and economic growth in the United States.

Nvidia's AI investment strategy

Nvidia has invested tens of billions of dollars in AI laboratories, cloud-computing providers and smaller technology companies over the past two years, seeking to strengthen its position across the AI ecosystem.

The strategy also supports demand for Nvidia's chips, which remain widely used to train and operate advanced AI models.

The FT reported that Nvidia agreed in September to acquire Hugging Face, a platform hosting millions of AI models and datasets, for $13 billion. It also cited a $20 billion deal involving chip designer Groq in December, structured in a way similar to an acqui-hire. That arrangement subsequently became the subject of a lawsuit brought by former Groq employees, according to the report.

Taken together, the potential Decart and Reflection deals illustrate two different approaches to AI expansion: Amazon could gain software that helps customers use different AI chips, while Nvidia could deepen its exposure to open-weight models, talent and technology that may generate further demand for its computing infrastructure.

Both discussions remain unconfirmed. The reported valuations and potential deal structures are subject to change, and neither company has announced a completed transaction.