Full Breakdown
Oracle Unveils OCI Zettascale10: The Largest AI Supercomputer in the Cloud
10/20/2025, 12:19:21 PM
Introduction to OCI Zettascale10
Oracle has announced the launch of its OCI Zettascale10, which it claims to be the largest AI supercomputer in the cloud, capable of achieving 16 zettaFLOPS of peak performance. This ambitious project utilizes approximately 800,000 Nvidia GPUs distributed across various data centers, including the infrastructure supporting OpenAI’s Stargate cluster in Abilene, Texas. The architecture is designed to handle demanding AI workloads, with a focus on scalability and reliability.
Technical Specifications and Design
The Zettascale10 system is built on Oracle's Acceleron RoCE networking technology, which aims to enhance performance for data-intensive AI operations. This architecture employs network interface cards as mini switches, linking GPUs across isolated network planes to minimize latency and ensure operational continuity in case of network failures. Ian Buck, vice president of Hyperscale at Nvidia, emphasized that the system provides the necessary compute fabric to advance AI research and facilitate the transition from experimentation to industrialized AI.
Market Position and Competitive Landscape
Oracle's Zettascale10 positions the company among major players in the AI infrastructure market, competing with other cloud providers that are also developing large-scale GPU clusters. Analysts are keen to observe whether Oracle's configuration can deliver performance on par with existing AI clusters from competitors. The system is expected to roll out next year, and its success will depend on real-world efficiency, which is influenced by both network design and software optimization.
Financial Implications and Growth Strategy
The announcement comes at a time when Oracle is experiencing significant growth, with co-CEOs Clay Magouyrk and Mike Sicilia highlighting a projected $0.5 trillion revenue pipeline from AI infrastructure. The company’s cloud infrastructure business is anticipated to generate $144 billion in sales by fiscal 2030, contributing to an overall revenue target of $225 billion. This growth is attributed to the integration of enterprise solutions, distributed cloud capabilities, and AI infrastructure, creating a synergistic effect across different business segments.
Criticism and Concerns
Despite the impressive claims surrounding the Zettascale10, some industry observers have raised concerns regarding the lack of independent verification for Oracle's performance figures. The company's reliance on theoretical peak performance rather than sustained rates has led to skepticism about its ability to meet the demands of scalable and efficient AI computation. Additionally, as other providers continue to enhance their own AI capabilities, Oracle's competitive edge may diminish.
Official Statements
Mahesh Thiagarajan, executive vice president of Oracle Cloud Infrastructure, stated, “With OCI Zettascale10, we’re fusing OCI’s Oracle Acceleron RoCE network architecture with next-generation Nvidia AI infrastructure to deliver multi-gigawatt AI capacity at unmatched scale.” This reflects Oracle's commitment to advancing its AI capabilities while addressing the growing demand for cloud-based AI solutions.
Conclusion
Oracle's OCI Zettascale10 represents a significant advancement in AI supercomputing, with the potential to reshape the landscape of cloud infrastructure. However, the true test of its capabilities will come with its rollout and subsequent performance evaluations against established competitors in the market.
