Oracle is Expanding the Accelerator Options in OCI for Enterprises
AI-summarised brief · reviewed before publication
Oracle announced a suite of new AI accelerator options for its Oracle Cloud Infrastructure (OCI) at the Oracle AI World 2026 conference in Las Vegas. Starting in 2027, OCI will offer Google’s Tensor Processing Units (TPU‑8t for training and TPU‑8i for inference), Cerebras’ wafer‑scale CS‑4 accelerator, and SambaNova’s SN50 Reconfigurable Dataflow Units, expanding hardware choices beyond Nvidia and AMD. The move follows a surge in Oracle’s cloud revenue, which rose 121 % year‑over‑year to $7.4 billion in Q1 FY 2027, and reflects the company’s strategy to attract enterprise and high‑performance computing customers seeking flexible, best‑in‑breed AI chips. Oracle also introduced its Deep Trust AI Architecture for integration with its AI Database, emphasizing data security and operational efficiency.
💡 Why It Matters
- · By opening OCI to rival chip ecosystems, Oracle positions itself as a neutral platform, giving enterprises the freedom to match workloads with the most suitable accelerator and reducing reliance on any single vendor.
