Amazon announced on Tuesday that Uber has decided to expand its cloud services contract with AWS, aiming to migrate a larger portion of its ride-hailing infrastructure to Amazon’s proprietary chip architecture. Under the agreement, Uber will increase its use of AWS Graviton—Amazon’s energy-efficient, ARM-based server CPUs—while simultaneously launching trials for the company’s latest AI-focused silicon, Trainium3.
This partnership has drawn significant industry attention, as it signals a notable pivot in Uber’s cloud infrastructure strategy. Previously, in 2023, Uber signed major multi-year agreements with Oracle and Google Cloud, outlining plans to move the bulk of its IT infrastructure away from its own private data centers and onto those two platforms.
Reshaping the Competitive Landscape
Although Uber’s blog post from last December highlighted its ongoing migration to Oracle Cloud Infrastructure (OCI) and Google Cloud—specifically noting the use of Ampere ARM chips within Oracle’s environment—this deeper integration with Amazon is being viewed as a direct competitive challenge to Oracle and Google. Analysts suggest that this move not only weakens Oracle’s influence in this specific segment but also underscores Amazon’s push to close the gap with traditional hardware giants like Nvidia through its own custom silicon.
The relationships within the chip industry are notoriously complex, and the background of Ampere Computing serves as a prime example. Founded by former Intel executive Renee James, Ampere counts Oracle as a major investor, with the latter holding roughly a one-third stake in the company. By choosing to embrace Amazon’s Graviton and Trainium chips, Uber has effectively bypassed the technical ecosystem barriers Oracle had previously established through its reliance on Ampere hardware.
For AWS, the introduction of Trainium3 is aimed at lowering AI computing costs and reducing reliance on third-party hardware suppliers. As one of the world’s largest ride-hailing platforms, Uber’s massive workloads will serve as a critical proving ground for Amazon to validate the performance of its custom silicon. With competition in the cloud market intensifying, the ability to design proprietary chips has become a key bargaining chip for major cloud providers vying for enterprise-level clients.
