Oracle has unveiled a series of AI-related announcements at Oracle AI World 2025, introducing new infrastructure, database technologies, and cloud partnerships aimed at integrating artificial intelligence into every layer of data management and enterprise computing.
At the core of the announcements is the new Oracle AI Data Platform, a unified environment designed to securely connect generative AI models with enterprise data, applications, and workflows. The platform combines automated data cataloging, contextualization, and vector indexing with built-in AI tools. According to Oracle, this simplifies the process from raw data ingestion to production-grade AI deployment, allowing organizations to operationalize generative AI more efficiently.
OCI Zettascale10: Cloud-Based AI Supercomputer
Oracle also introduced OCI Zettascale10, described as the largest AI supercomputer in the cloud. Built on Oracle Cloud Infrastructure (OCI), Zettascale10 links hundreds of thousands of NVIDIA GPUs across multiple data centers into clusters capable of 16 zettaFLOPS peak performance. The system uses Oracle’s new Acceleron RoCE network architecture to minimize latency between GPUs and maximize scalability. It forms the backbone of Stargate, a joint AI supercluster project with OpenAI, designed to handle next-generation generative AI workloads.
Partnership Expansion with AMD
Oracle announced an expanded collaboration with AMD to deliver large-scale AI infrastructure based on the upcoming AMD Instinct MI450 GPU series. OCI will serve as the launch platform for the first publicly available AI supercluster powered by these GPUs. The initial deployment will include 50,000 GPUs, expected to go online in the third quarter of 2026, with further expansion planned for 2027 and beyond.
AI Database 26ai
Another key announcement is Oracle AI Database 26ai, a new version of Oracle’s flagship database that embeds AI directly into its core. The company describes this as part of its “AI for Data” strategy—an evolution toward AI-native data management. The database integrates vector search, AI-assisted development, and generative workflows, allowing users to apply AI directly to their data without moving it outside secure environments. It enables the execution of agentic AI workflows that combine proprietary enterprise data with public sources for advanced reasoning and automated actions.
Expanded Multicloud Database Partnerships
Oracle is also extending its multicloud strategy with updates to its database services across AWS, Azure, and Google Cloud. Oracle Database@AWS now supports Autonomous AI Lakehouse, Zero Data Loss Autonomous Recovery Service, and Terraform, and is available through AWS and Oracle partners. Oracle Database@Azure is expanding to 28 regions, with five more planned within a year. Activision Blizzard was cited as a key user supporting cloud migration with this service. Oracle Database@Google Cloud has gained new AI capabilities, additional regional coverage, and a partner program to accelerate innovation and reduce costs for multicloud users.
A Unified AI Infrastructure Vision
With these announcements, Oracle positions itself as a full-stack AI infrastructure provider—from GPU-powered supercomputing to AI-native databases and multicloud integration. The company’s approach aims to give enterprises control over their data while enabling large-scale AI innovation within existing regulatory and security frameworks.
