If Amazon Web Services (AWS) has its way, the cloud giant is set to push deeper into Nvidia’s territory, potentially mounting one of the most significant challenges yet to Nvidia’s dominance in AI chips.
In a recent interview with Bloomberg, Amazon’s AI chief Peter DeSantis revealed that AWS is in early-stage talks to sell its custom AI chip, Trainium, to other companies for use in their own data centers. DeSantis declined to name potential buyers, but the move signals a major strategic shift for the e-commerce and cloud computing titan.
The discussions stem from Amazon CEO Andy Jassy’s annual shareholder letter in early April 2026, in which he acknowledged the high demand for Amazon’s in-house AI chips. “If our chips business was a standalone company, selling chips produced this year to AWS and other third parties (as other leading chip companies do), our annual run rate would be roughly $50 billion,” Jassy wrote. “There’s so much demand for our chips that it’s quite possible we’ll sell racks of them to third parties in the future.”
Could a $50 billion chip business seriously rattle Nvidia? Not exactly. Nvidia’s current revenue run rate stands at approximately $326 billion, fueled by record quarters and a commanding position in the AI accelerator market. However, Amazon’s potential chip revenue would rival that of Intel, which reported around $50 billion in annual sales in 2025. So while it wouldn’t “tank” Nvidia, it would establish Amazon as a formidable new competitor in the high-stakes AI semiconductor arena.
AWS has historically resisted selling its AI chips for several reasons. The primary one is that the cloud division profits not only from the chips themselves but also from the ecosystem of services they enable—including storage, security, networking, and monitoring—creating a “waterfall” revenue effect. Additionally, demand for Trainium has far outstripped supply. In his April letter, Jassy noted that the current generation of Trainium chips sold out almost immediately, as did pre-orders for the upcoming Trainium4, which won’t be available for more than a year—all before AWS formally added OpenAI’s models to its cloud platform.
Selling chips externally would likely mean pushing current AWS customers onto waiting lists unless Amazon can somehow ramp up production through its manufacturing partner TSMC. That’s a tall order: Nvidia recently surpassed Apple to become TSMC’s largest customer, making it difficult for Amazon to secure additional capacity without direct competition for fab space.
AWS spokesperson Doron Aronson confirmed the company’s evolving stance: “While we’ve historically declined requests to sell chips directly, Andy noted it’s quite possible we’ll sell racks of them to third parties in the future.” Aronson recently hosted TechCrunch on a private tour of Amazon’s chip design facility, underscoring the company’s deepening commitment to silicon development.
So, while Nvidia CEO Jensen Huang recently proclaimed his company’s lead unassailable, Amazon’s move to commercialize its Trainium chips—even on a limited basis—could reshape the competitive landscape in AI hardware. If AWS can navigate supply constraints and offer a viable alternative to Nvidia’s GPUs, 2026 may be remembered as the year the cloud giant became a direct chip rival.
via TechCrunch AI