Oracle expands cloud hardware with Google TPUs and OpenAI inference chips

Oracle Corp. announced plans on October 27, 2026, to offer Google LLC's Tensor Processing Unit accelerators through its public cloud, SiliconANGLE and The Register reported. SiliconANGLE noted that this is the first time Google has supplied its TPUs to a rival hyperscaler. CNBC and Quartz both confirmed that Oracle is expanding its cloud computing options to include chips from multiple external partners. [1][5][6][7][8][9]

The public cloud expansion includes additional hardware options alongside Google's TPUs. SiliconANGLE and CNBC reported that Oracle will add Cerebras Systems Inc.'s wafer-scale accelerators and OpenAI Group PBC's Jalapeño inference processor to its cloud platform. According to SiliconANGLE, Google's TPUs exchange data through inter-core interconnect and optical circuit switch technologies. [1][6][9]

OpenAI and Oracle are also expanding their software partnership. The OpenAI Blog and CNBC reported that the companies will integrate business applications and artificial intelligence agents into ChatGPT. CNBC and Quartz both added that Oracle plans to embed ChatGPT and Codex directly into its enterprise applications. [2][6][7]

Google developed the TPU as an application-specific integrated circuit for machine learning workloads, initially for internal use in 2015 before offering it to third parties in 2018. As FTMQ SI reported earlier, Oracle and OpenAI previously agreed to broaden cloud infrastructure chip selections alongside enterprise software integration. Following the hardware and software announcements, Stocks Down Under reported that Oracle stock rose five percent. [4][10][14]