AWS evaluates selling Trainium chips to third parties and fully entering the AI business
Although a business unit valued at $50 billion would be gigantic for any tech company, it would still be far from the numbers that Nvidia handles.
The company led by Jensen Huang currently maintains an annualized revenue rate close to $326 billion, primarily driven by the global explosion of generative artificial intelligence.
Amazon is preparing a move that could disrupt the artificial intelligence market. The company confirmed that it is considering selling its own Trainium chips to other companies for use in data centers, a strategy that would bring it directly into the territory dominated by Nvidia.
The information was revealed by Bloomberg following statements from Peter DeSantis, one of the key figures in AWS's AI business. Although potential clients are not yet known, the initiative already exposes Amazon's ambitions in one of the most competitive technology sectors at the moment.
So far, Trainium chips have been used exclusively within AWS's infrastructure to power artificial intelligence services in the cloud.
However, Amazon's CEO, Andy Jassy, had already hinted at this possibility in his annual letter to shareholders published in April. He stated that the demand for these processors is so high that the company could start selling them directly to third parties.
As the executive explained, if the chip business operated as an independent company, it would currently have an annual revenue close to $50 billion.
However, Amazon's eventual entry as a direct chip supplier would represent one of the most significant challenges Nvidia has faced so far in this market.
Amazon's historical strategy has been to use its processors as a competitive advantage within its cloud services ecosystem.
In addition to charging for processing artificial intelligence models, AWS generates additional revenue through storage, networking, monitoring, security, and other associated services.
For this reason, selling the chips independently could alter part of the business logic that currently supports Amazon's cloud business.
