Oracle embeds Cohere
Reports say Oracle is partnering with Cohere to embed generative AI across Oracle Cloud and apps — and Larry Ellison is reportedly buying a lot of NVIDIA chips to support it. That creates a hyperscaler‑ISV channel moment with predictable GPU procurement cadence. (insidermonkey.com)
Oracle added Cohere Embed v4 (cohere.embed‑v4.0), a multimodal embeddings model, to the OCI Generative AI catalog with documentation updated Feb 25, 2026. (docs.oracle.com/en-us/iaas/Content/generative-ai/cohere-embed-4.htm). (docs.oracle.com) A technical case study from Oracle describes Cohere training and deploying its foundation models on OCI using Oracle Kubernetes Engine and cluster GPUs to support enterprise-scale model workflows (May 20, 2025). (oracle.com/cloud/technical-case-studies/cohere). (oracle.com) Cohere and Oracle customer pages say Oracle Fusion Applications have been augmented with more than 100 Cohere‑powered generative AI use cases to deliver in‑app automation and search over trusted data sources. (cohere.com/customer-stories/oracle). (cohere.com) DatacenterD and footage from Oracle AI World quote Larry Ellison saying the Stargate Abilene site will house on the order of 450,000 Nvidia GB200 GPUs for paired OpenAI workloads. (datacenterdynamics.com/en/news/openai-and-oracle-to-deploy-450000-gb200-gpus-at-stargate-abilene-data-center). (datacenterdynamics.com) Contemporaneous reporting in May 2025 placed Oracle’s initial purchase at roughly 400,000 GB200 (Grace Blackwell) processors with an estimated transaction value near $40 billion for the first Stargate installation. (siliconangle.com/2025/05/23/oracle-reportedly-buying-400000-nvidia-chips-first-stargate-data-center). (siliconangle.com) OpenAI’s Stargate announcement frames the broader deal as adding roughly 4.5 GW of data‑center capacity with plans that could run “over 2 million chips” across sites under development. (openai.com/index/stargate-advances-with-partnership-with-oracle). (openai.com)
