Amazon built AWS into the backbone of its business. Jeff Bezos believes custom computer chips could become its next major pillar.
The Amazon founder told Fortune that the company’s silicon operation, which produces Trainium AI accelerators and Graviton processors, is developing into an enduring business alongside Marketplace, Prime, and AWS. The chips are also central to Amazon’s effort to expand its AI infrastructure while reducing its reliance on Nvidia.
For AWS customers, the wager could mean more choice and potentially lower computing costs. For Amazon, it is a bid to control more of the technology — and economics — behind the AI boom.
“A few of our offerings have become durable pillars, things like Marketplace and Prime and AWS,” Bezos told Fortune. “What I see right now is that our chips business, our silicon business, is lining up to be our next pillar.”
Why Amazon is building an alternative to Nvidia
Bezos’ comments underscore Amazon’s expanding investment in custom AI chips as demand for computing capacity rises and cloud providers seek alternatives to Nvidia’s market-leading processors.
Technology giants are seeking to develop their own processors to reduce their dependence on Nvidia’s costly AI chips. Amazon has invested heavily in custom silicon through Annapurna Labs, the Israeli chip startup it acquired in 2015. The company now develops its own AI chips under the Trainium and Inferentia brands, which are designed to train and run LLMs while reducing costs for customers using Amazon Web Services.
AWS has positioned its chips as a lower-cost alternative for AI developers. Trainium is also used by Anthropic to train and runs its Claude models, and OpenAI has committed to consuming about 2 gigawatts of Trainium capacity, ramping in 2027.
With spending among the hyperscalers projected to exceed $700 billion this year, Amazon is expected to invest $200 billion in capital spending this year, primarily on AI infrastructure. The figures have prompted fears about an industry bubble.
What's hot at TechRepublic
- Blackpoint Cyber vs. Arctic Wolf: Which MDR Solution is Right for You?
- Why AWS Sellers Choose Deepgram Over Other Voice AI Tools
- SS&C Intralinks DealCentre AI vs. Datasite: Which platform is built for the future of dealmaking?
- SS&C Intralinks FundCentre AI vs. Juniper Square: Which platform better supports modern private markets fund managers?
- Verito vs. Rightworks: Which IT Provider Is Best for Your Firm?
Amazon may eventually sell its chips beyond AWS
Amazon President and CEO Andy Jassy has repeatedly made the case that demand for AI computing will remain strong for years to justify company investments in chips, servers, networking equipment, and power generation as critical to its long-term growth.
According to the Fortune article, Jassy said that AI will change “every customer experience that we know today and invent a whole host of new ones … I do think we’re living in a world where … the key to the compute is often the chips,” he said. “The growth in AI has been so significant, but we have a chips business that we built over the last decade here that is growing very quickly.”
Amazon’s chip ambitions may eventually extend beyond its own data centers. In his annual shareholder letter, Jassy wrote that it was “quite possible” the company would one day sell racks containing its custom chips directly to third parties.
That would move Amazon closer to the strategy pursued by Google, which develops custom tensor processing units with Broadcom.
AWS says customers are already committing to AI capacity
Jassy said in April that AWS’s AI business had exceeded a $15 billion annual revenue run rate. He argued that Amazon’s spending was supported by demand rather than speculation.
“We’re not investing … on a hunch,” Jassy said. He added that Amazon already had customer commitments covering a substantial portion of the AWS capital expenditure planned for 2026, much of which the company expects to monetize in 2027 and 2028.
The next test is whether demand for Trainium and Amazon’s other processors expands beyond a handful of major AI customers — and whether the company can turn custom silicon into a business that stands alongside AWS rather than merely supporting it.
Read more: Amazon Reportedly Plans to Consolidate Nova AI Models