Oracle Diversifies AI Hardware and Partners with Snowflake, OpenAI
Oracle is expanding its cloud infrastructure by integrating Google TPUs, Cerebras chips, and Snowflake services while launching a new AI security mesh with OpenAI.
During the Oracle AI World event in Las Vegas, Oracle Corp. announced a massive expansion of its cloud infrastructure and strategic partnerships to diversify its artificial intelligence capabilities. The company is shifting toward a multi-vendor hardware strategy, scaling its infrastructure to multi-gigawatt levels using Nvidia's Vera Rubin AI servers while integrating its own Acceleron networking chips. In a first for a hyperscale cloud platform outside of Google, Oracle will offer Google's eighth-generation tensor processing units (TPUs) starting in 2027.
Oracle is further diversifying its hardware by integrating chips from Cerebras Systems and SambaNova Systems to target the AI inference market. Cerebras hardware is scheduled for deployment in the first half of 2027. On the software side, Oracle announced that Snowflake Inc. will now be available as a native, first-party service on Oracle Cloud Infrastructure (OCI), allowing customers to provision the data platform directly through Oracle Universal Credits.
The company also expanded its collaboration with OpenAI, launching the Oracle AI Security Mesh to identify enterprise vulnerabilities using OpenAI's cyber models. Oracle revealed that OpenAI's GPT-6 Astra model was trained at its Stargate data center in Texas. These advancements come as Oracle maintains a capital spending target of $90 billion to $95 billion for the 2027 fiscal year, despite facing local opposition to its Project Jupiter data center in New Mexico.
