Jeff Bezos says custom chips could become Amazon’s next major business, as the company expands its AWS AI infrastructure with new processors. Speaking about the potential of Amazon’s Trainium, Graviton, and other custom chips, Bezos highlighted their growing role in powering machine learning and cloud computing workloads. He noted that these in-house designed chips could eventually form a significant revenue stream, separate from Amazon’s core e-commerce operations. The comment came during a recent discussion on the company’s long-term strategy for dominating the cloud AI market.
Amazon has invested heavily in developing custom silicon to reduce reliance on third-party suppliers like NVIDIA. Jeff Bezos says custom chips such as Trainium are specifically optimized for training large AI models, offering better price-performance ratios for AWS customers. Graviton processors, built on Arm architecture, are already used for general-purpose computing and have shown energy efficiency gains. According to Bezos, this internal chip development aligns with Amazon’s goal to provide affordable, high-performance cloud services at scale.
The potential of Amazon’s custom chips extends beyond internal use. Jeff Bezos says custom chips could be marketed as standalone products or services, much like AWS has evolved from an internal tool to a public cloud platform. This shift would position Amazon as a competitor to major chip designers such as Intel and AMD. Bezos emphasized that the transition would require years of continued investment and customer adoption. However, he believes the long-term payoff could be substantial as AI workloads continue to grow worldwide.
In conclusion, Jeff Bezos says custom chips represent a strategic opportunity for Amazon to diversify its revenue and reduce supply chain dependencies. While still in early stages, this business segment could reshape the cloud computing landscape. The company’s focus on cost-effective, tailored hardware positions it well for the AI-driven future.