Amazon in Talks to Sell Trainium AI Chips to Third Parties

06/21/2026 / By Edison Reed

Amazon is in talks to sell its custom-built Trainium artificial intelligence (AI) chips for use in other companies’ data centers, according to a Bloomberg report cited by ZeroHedge.

The initiative represents a key expansion of Amazon Web Services’ (AWS) efforts to challenge Nvidia’s dominance in the AI hardware market. Peter DeSantis, Amazon’s AI chief, confirmed the discussions but declined to name potential customers.

The report noted that a less optimistic interpretation of the talks is that Amazon does not have enough internal demand or capacity to use the chips for its own purposes. Presumably, the company will attempt to steal market share by offering its product at more competitive terms, according to the report, which could add pricing pressure across the AI ecosystem.

“We view AI infrastructure as rapidly evolving,” DeSantis said in an interview in Paris, as cited by the report. “And we’re constantly looking at ways to get to more customers.” ^[1] ^[2]

Background and Early Adoption of Trainium

Introduced in 2020, Trainium has attracted marquee buyers including OpenAI, Anthropic and Uber, which access the hardware through AWS. Amazon reported in April that the chip has generated more than $225 billion in revenue commitments.

Amazon CEO Andy Jassy wrote in his shareholder letter that it is “quite possible” the company would sell racks of its chips to third parties, according to the report. The letter was part of a broader attempt to reposition the company around AI, an area where it has been seen as falling behind rivals. ^[1]

Uber recently expanded its contract for AWS cloud services to run more of its ride-sharing features on Amazon’s chips, and began a new trial testing Trainium3, according to a TechCrunch report. Book author Cornelia C. Walther noted in her work that AWS has developed two chips: Trainium for training machine learning models and Inferentia, which powers inferences that yield outputs. According to Jassy, both have been successful. ^[3] ^[4]

Competitive Landscape and Strategic Motives

Amazon and other cloud computing titans have each been developing their own alternatives to Nvidia’s graphics processing units, and they ramped up these efforts after the arrival of ChatGPT, according to the Bloomberg report. Alphabet CEO Sundar Pichai said in April that Google will begin to deliver its Nvidia GPU rival chips, called tensor processing units, to a “select group of customers” for use in their own data centers. Google Cloud launched two new AI chips in April, the TPU 8t for training and TPU 8i for inference, according to a TechCrunch report. ^[1] ^[5]

DeSantis cited the growing demand outside the United States for computing resources that are controlled locally as a factor in Amazon’s decision to sell Trainium to third parties. The push for “sovereign” services in Europe and other regions has prompted calls for countries to lessen their reliance on U.S. technology. Meanwhile, Nvidia retains a 94% market share in GPU sales, according to a NaturalNews.com report, underscoring its dominant position in the industry. ^[6]

Market Conditions and Pricing Pressures

Token costs have fallen 20% since the start of the month, and compute rental prices are at one-month lows, according to data cited by ZeroHedge. The pricing trends suggest a sudden drop in demand or an excess supply of compute capacity.

DeSantis dismissed concerns that selling Trainium outside AWS would eat into the company’s cloud business. “There’s so much underconsumption in AI,” he said. “I’m not worried about it.” ^[1]

The third version of the Trainium chip, which began shipping earlier this year, is “largely sold out,” DeSantis said during the VivaTech conference in France. He added that there is already strong interest in a fourth version expected to debut next year. Amazon also supplies Graviton general-purpose chips to Meta, and DeSantis said the company has added more Graviton chips to its computing systems than any other type of chip over the last three years. ^[1] ^[7]

Impact on Amazon’s Business and Stock

Amazon shares gained as much as 2.5% on Thursday, June 18, reaching an intraday high of $243 on the news of the Trainium sales discussions, according to the Bloomberg report cited by ZeroHedge. The broader push reflects Amazon’s repositioning around AI as it seeks to catch up with rivals. Meta signed a multibillion-dollar deal to use tens of millions of AWS Graviton processor cores for its next-generation AI buildout, making Meta one of the largest Graviton customers globally, according to a ZeroHedge report. ^[1] ^[8]

The move by Amazon to sell its chips directly to other companies marks a direct challenge to Nvidia, which has seen its market capitalization reach $4 trillion amid the AI boom, according to a NaturalNews.com report. The AI hardware boom, driven by generative AI models, has propelled Nvidia’s stock significantly higher, but the emergence of competing chips from cloud providers could begin to erode Nvidia’s near-monopoly over the long term. ^[9]

References

“Amazon In Talks To Sell Its Trainium AI Chips To Other Firms, In Challenge To Nvidia Dominance”. ZeroHedge. June 18, 2026.
“Amazon hopes to challenge Nvidia more directly by selling its AI chips”. TechCrunch. June 18, 2026.
“Uber is the latest to be won over by Amazon’s AI chips”. TechCrunch. April 07, 2026.
Cornelia C. Walther. “Human Leadership for Humane Technology: The New AI: Agency Ignited”.
“Google Cloud launches two new AI chips to compete with Nvidia”. TechCrunch. April 22, 2026.
Willow Tohi. “Unlocking the future: Nvidia poised for a blockbuster year in the AI revolution”. NaturalNews.com. September 06, 2025.
“In another wild turn for AI chips, Meta signs deal for millions of Amazon AI CPUs”. TechCrunch. April 24, 2026.
“Scramble For AI Compute: Meta Inks Multibillion-Dollar Deal With Amazon For CPU Chips”. ZeroHedge. April 24, 2026.
Willow Tohi. “NVIDIA surpasses $4 trillion market cap amid AI boom and geopolitical strains”. NaturalNews.com. July 11, 2025.