Oracle expands AI chip supply partnership with Nvidia while adding Google TPUs and Cerebras chips
Oracle will deploy Nvidia Vera Rubin AI servers at gigawatt scale, while also becoming the first hyperscaler outside Google to offer Google TPUs, with service expected from 2027. It will additionally offer Cerebras and SambaNova inference chips and replace some Nvidia networking chips in Nvidia-based systems with its own Acceleron chips. This matters because it breaks a single-vendor pattern: as inference demand diversifies, cloud providers are competing on chip choice rather than Nvidia alone.
SOURCES
- IT之家2026-10-28 00:20