OpenAI in talks with Amazon for $10 billion, may adopt Trainium

OpenAI entered early stage discussions with Amazon on December 16 about a potential financing round of roughly $10 billion and a commercial agreement to run significant training workloads on Amazon Trainium chips. People familiar with the negotiations said the talks were confidential and very fluid, and that terms remained subject to change. The talks, if finalized on reported terms, could value OpenAI at well above $500 billion.

Executives on both sides discussed a package that would marry a large capital infusion with a technology commitment, according to sources. The chip component would give Amazon, which has built Trainium as an in house accelerator for artificial intelligence training, a marquee customer for its semiconductor efforts. For OpenAI, the arrangement would offer an alternative supply of compute that could reduce part of its long term dependence on Microsoft cloud infrastructure.

The potential financing would be one of the largest private investments into an AI developer, and would come as the industry broadly rethinks vendor relationships for models that demand enormous compute capacity. Industry figures underscore Nvidia’s dominant position in AI accelerators, with a market capitalization roughly in the range of $4.3 trillion, and observers say Amazon’s silicon business remains far smaller. Still an OpenAI commitment to Trainium would be a major strategic and reputational gain for Amazon’s semiconductor division.

OpenAI’s capacity to seek large outside investment followed an internal restructuring last October that loosened some constraints in its prior governance arrangements and opened the door to broader capital partnerships. That change, combined with rapid investor appetite for frontier AI, helped set the stage for the conversations with Amazon. People briefed on the matter cautioned that the reported valuation and the size of the round were preliminary, with some characterizations calling the amount at least $10 billion and others saying it could exceed $10 billion.

A deal that combines capital and infrastructure raises many open questions that were not resolved in the talks. There was no definitive timetable, nor were precise terms disclosed for equity ownership, governance changes, pricing of compute, or the scope and timing of any migration to Trainium. Observers also noted potential regulatory and competitive implications, because Microsoft has been a central technology partner to OpenAI and a large purchaser of AI compute capacity. Shifts in OpenAI’s vendor choices could ripple through cloud contracts and long term infrastructure planning across the industry.

If consummated the arrangement would alter fundraising dynamics for OpenAI and could position the company to attract further investment. It would also force a reckoning among hardware suppliers and cloud providers over how to serve the next generation of large scale AI models. For now the negotiations remain at an early and confidential stage, and the sector is watching to see whether an agreement materializes and how it could reshape the balance of power in AI compute and capital markets.