At GTC 2025, Oracle and NVIDIA announced a series of collaborations designed to
accelerate enterprise AI adoption, agentic AI inference, and sovereign
AI—empowering organizations to scale AI workloads securely and efficiently.
The two companies unveiled a first-of-its-kind integration that embeds NVIDIA AI
Enterprise natively in Oracle Cloud Infrastructure (OCI), providing direct access to
over 160 AI models and more than 100 NVIDIA NIM™ microservices through the
OCI Console. This integration eliminates deployment complexity, enabling
enterprises to quickly harness agentic AI and simplify production AI inference
pipelines by deploying pre-optimized models in minutes, with unified billing and
support. NVIDIA cuVS further accelerates AI vector search in Oracle Database
23ai, delivering fast vector queries and reduced operational costs as data volumes
grow.
(Read the official press release)
OCI & NVIDIA: Powering Secure and Cost-Efficient AI Agents for
Government
Oracle and NVIDIA deliver a comprehensive AI ecosystem designed for
government and federal agencies to build and deploy AI agents securely and cost-
effectively. NVIDIA NIM on OCI Data Science enables real-time intelligence and
automated workflows, while deployments in OCI’s FedRAMP High and DISA IL5
environments offer end-to-end encryption, zero-trust architecture, and continuous
monitoring. Agencies can scale workloads based on demand with OCI’s cost-
effective NVIDIA GPU instances, maximizing AI performance and budget
efficiency.
Oracle and NVIDIA Deliver Sovereign AI Anywhere
With Sovereign AI capabilities, organizations can retain data control while building,
tuning, and inferring models wherever their data resides. OCI and NVIDIA DGX™
Cloud support flexible deployments—on-premises, in OCI regions, OCI Dedicated
Region, or via OCI Alloy partner clouds—helping ensure compliance with residency
and privacy requirements.
OCI AI Blueprints: Accelerating AI Adoption on OCI
Oracle’s AI Blueprints provide validated deployment patterns—including
architecture diagrams, Terraform code, and automation—for rapidly launching AI
workloads on OCI. This reduces setup time from weeks to minutes while helping
ensure best practices and scalability.
(Learn about OCI AI Blueprints)
First Principles Zettascale OCI Superclusters
Oracle is building zettascale superclusters that combine high-performance NVIDIA
GPUs, RDMA cluster networking, and scalable Lustre storage—engineered to run AI
models with trillions of parameters and deliver extreme-scale AI training and
inference workloads efficiently in OCI.
(Learn about OCI superclusters)
Accelerating AI Vector Search in Oracle Database 23ai with NVIDIA GPUs
Oracle Database 23ai integrates NVIDIA GPU acceleration to enable fast and
efficient AI vector search. It supports hybrid queries combining semantic vector
search with traditional SQL, powering generative AI applications and retrieval-
augmented generation (RAG) workloads.