Oracle and NVIDIA have significantly expanded their strategic partnership, aiming to make enterprise artificial intelligence (AI) services more accessible, powerful, and practical for businesses worldwide. These pivotal announcements, unveiled during Oracle AI World, encompass a spectrum of innovations from cutting-edge hardware to deeply integrated software solutions designed to embed AI at the core of enterprise data management.

A cornerstone of this collaboration is the introduction of the new OCI Zettascale10 computing cluster. This platform is engineered with NVIDIA GPUs and specifically designed to handle the most demanding AI training and inference workloads. Boasting an impressive 16 zettaflops of peak AI compute performance, the OCI Zettascale10 utilizes NVIDIA’s Spectrum-X Ethernet, a high-performance networking fabric that prevents GPUs from idling while waiting for data, enabling organizations to scale efficiently up to millions of processors. Mahesh Thiagarajan, Executive VP of Oracle Cloud Infrastructure, emphasized that this cluster provides “multi‑gigawatt capacity for the most challenging AI workloads with NVIDIA’s next-generation GPU platform.”

Beyond raw computational power, the true impact of this partnership lies in the profound software integrations that aim to seamlessly weave AI into every layer of business operations. A key component of this strategy is the Oracle AI Database 26ai, which fundamentally shifts the conventional approach to AI. Instead of moving data to AI models, Oracle advocates for bringing AI directly to the data, arguing that this method is inherently more secure and efficient. This “AI for Data” vision is fully realized in the latest database release.

Juan Loaiza, Executive VP of Oracle Database Technologies, highlighted the simplicity and ease of use of the Oracle AI Database 26ai, enabling customers to effortlessly deliver trusted AI insights and innovations across all their data, including both operational systems and analytic data lakes. A standout feature of the AI Database 26ai is its ability to run agentic AI workflows directly within the database. These AI agents can address complex queries by intelligently combining a company’s private, sensitive data with public information, all without ever requiring the private data to leave its secure environment. This is facilitated by advanced features such as Unified Hybrid Vector Search, which allows AI to contextualize information across diverse data types, whether they reside in relational tables, JSON files, or spatial maps.

Security remains a paramount concern for Oracle, with the new database implementing NIST-approved quantum-resistant algorithms for data both in-flight and at-rest. This proactive measure provides a robust defense against potential “harvest now, decrypt later” attacks, where adversaries might steal encrypted data today with the intent of decrypting it using future quantum computing capabilities. Holger Mueller, VP and Principal Analyst at Constellation Research, remarked on the synergy, stating, “Great AI needs great data. With Oracle AI Database 26ai, customers get both,” noting its role as a secure, consistent, single source of business data, and its innovative agentic AI capabilities accessible via a no-code visual platform.

The new database is meticulously designed for seamless interoperability with NVIDIA’s comprehensive toolset. Its programming interfaces now directly integrate with NVIDIA NeMo Retriever, a collection of microservices that streamline the complex infrastructure required for modern enterprise AI, particularly for retrieval-augmented generation (RAG). RAG empowers language models to search and utilize relevant facts from a company’s internal documents before formulating responses, significantly enhancing accuracy and utility. Furthermore, the Oracle Private AI Services Container will receive a GPU-powered upgrade, enabling it to offload the intensive task of generating vector embeddings—crucial for AI search—to powerful NVIDIA GPUs using the cuVS library, thereby drastically reducing data preparation time for AI applications.

Beyond the database, the partnership aims to simplify the entire AI pipeline, democratizing enterprise AI. The new Oracle AI Data Platform now offers a built-in NVIDIA GPU option and includes the NVIDIA RAPIDS Accelerator for Apache Spark. This allows data scientists and engineers to accelerate their data processing and machine learning workflows using GPUs, often without requiring any changes to their existing code. All these powerful tools and capabilities are consolidated within the Oracle AI Hub, providing organizations with a single, unified platform to build, deploy, and manage their AI solutions. From this hub, users can easily deploy NVIDIA’s NIM microservices—essentially pre-packaged AI skills—through an intuitive, no-code interface. To further lower the barrier to entry, the complete NVIDIA AI Enterprise software suite is now natively available within the OCI Console, allowing developers to quickly provision GPU instances and activate all necessary NVIDIA tools with minimal clicks, bypassing lengthy procurement processes.

This extensive collaboration clearly addresses the practical challenges businesses encounter when adopting AI. By synergistically integrating advanced hardware, sophisticated data management, and comprehensive software tools into one cohesive ecosystem, Oracle and NVIDIA are heralding the arrival of a new era of practical, secure, and highly scalable enterprise AI.