October 14, 2025 — At Oracle AI World 2025, Oracle unveiled its most ambitious AI infrastructure yet: the OCI Zettascale10 computing cluster, a high-performance AI system built on NVIDIA’s latest GPU architecture and Spectrum-X Ethernet. The platform promises up to 16 zettaflops of peak AI performance — a leap designed to meet the scale and complexity of modern large language models and enterprise AI workloads.

Pushing AI Performance Boundaries

The Zettascale10 cluster is engineered for AI training and inference at massive scale, connecting millions of GPUs with unprecedented efficiency. It leverages NVIDIA Spectrum-X Ethernet, the world’s first Ethernet platform purpose-built for AI, allowing hyperscalers and enterprises to train next-generation models faster and more efficiently.

“The AI market has been defined by partnerships like the one between Oracle and NVIDIA,” said Mahesh Thiagarajan, Executive Vice President of Oracle Cloud Infrastructure. “OCI Zettascale10 delivers multi-gigawatt capacity for the most challenging AI workloads, while the native availability of NVIDIA AI Enterprise on OCI brings a leading AI toolset right next to our 200+ cloud services.”

Ian Buck, NVIDIA’s VP of Hyperscale and High-Performance Computing, added that the partnership “marks new frontiers in accelerated computing — streamlining database AI pipelines, speeding data processing, and making inference easier to deploy and scale on OCI.”

Oracle Database 26ai Gets AI Acceleration

Oracle’s flagship Database 26ai is now optimized for AI vector workloads with NVIDIA NIM microservices and NeMo Retriever integration. Developers can build retrieval-augmented generation (RAG) pipelines directly within the database, using NVIDIA’s suite of AI microservices to handle data ingestion, embedding, reranking, and context-aware response generation.

The update also introduces the Oracle Private AI Services Container, a flexible deployment tool that will soon let users offload vector embedding and index generation to NVIDIA GPUs using the open-source cuVS library — speeding up one of the most computationally heavy parts of vector search applications.

Oracle AI Data Platform and NVIDIA RAPIDS for Apache Spark

Oracle also launched the Oracle AI Data Platform, a unified data and AI ecosystem that integrates NVIDIA accelerated computing to power high-performance analytics and machine learning pipelines.

The platform includes a GPU-powered option and native integration with the NVIDIA RAPIDS Accelerator for Apache Spark, which boosts data processing and ETL speeds without code changes by combining NVIDIA cuDF with Spark’s distributed framework.

Enterprise AI in Action: Media, Vision, and Document Intelligence

Oracle is already deploying NVIDIA NeMo and Nemotron models in real-world use cases. For example:

• Oracle Media and Entertainment is using NeMo Curator and Nemotron vision-language models (VLMs) for automated video understanding — from decoding and segmentation to captioning and annotation.

• Oracle Fusion Document Intelligence is enhanced with NeMo Retriever Parse, a transformer-based model that can interpret and extract structured information from complex enterprise documents.

Together, these capabilities streamline AI model development and multimodal data processing across industries.

NVIDIA AI Enterprise Now Native in OCI

Oracle and NVIDIA are also simplifying enterprise AI deployment through the native integration of NVIDIA AI Enterprise within the OCI Console. Customers can now access NVIDIA’s full software stack — including frameworks, libraries, and tools — directly from OCI when provisioning GPU instances.

The integration supports deployment across OCI’s distributed cloud, including sovereign and dedicated regions, helping enterprises meet strict data security and compliance requirements.

This move eliminates the need for separate Marketplace purchases and brings flexible pricing, enterprise-grade support, and priority security updates under one unified interface.

A Partnership Built for the AI Era

The collaboration between Oracle and NVIDIA continues to expand across infrastructure, data, and software layers — combining Oracle’s enterprise cloud scale with NVIDIA’s AI acceleration expertise. At Oracle AI World, NVIDIA was recognized as a 2025 Oracle Partner Award winner, underscoring the growing importance of this alliance in shaping the enterprise AI landscape.

With the Zettascale10 cluster and new integrations across OCI and Database 26ai, Oracle is positioning itself as a major player in AI infrastructure and data-driven enterprise computing — one designed to compete at the zettascale frontier.