NVIDIA and Oracle are accelerating AI and data processing by bringing together Oracle’s OCI Supercluster with NVIDIA’s Blackwell platform.
- The 15 Biggest AI Companies to Watch Right Now
- The Top Artificial Intelligence Events for 2024 and 2025
- Ethical AI: The Biggest Challenges with AI and Ethics
The technological partnership will help businesses to train and deploy AI models with over 100,000 NVIDIA’s graphics processing unit (GPU).
According to US GPU designer and manufacturer, NVIDIA, there is an increasing appetite among enterprises for more powerful compute for AI and data processing.
Dani Yogatama, cofounder and CEO of Reka, shared his review of OCI Supercluster and NVIDIA: “Reka’s multimodal AI models, built with OCI and NVIDIA technology, empower next-generation enterprise agents that can read, see, hear and speak to make sense of our complex world.
“With NVIDIA GPU-accelerated infrastructure, we can handle very large models and extensive contexts with ease, all while enabling dense and sparse training to scale efficiently at cluster levels.”
NVIDIA-Powered OCI Superclusters
OCI Superclusters give customers the opportunity to choose from various NVIDIA GPUs and deploy them on public cloud, sovereign cloud, or on premises.
Due to become available in the first half of 2025, Blackwell-based systems can scale as much as 131,072 Blackwell GPUs with NVIDIA ConnectX-7 NIC for RoCEv2 or NVIDIA Quantum-2 InfiniBand networking.
The result is described by NVIDIA as ‘astounding’ as it can deliver 2.4 zettaflops (a unit of computer processing power) of compute to the cloud.
Oracle has previewed the NVIDIA GB200 NVL72, which can be used to power generative AI apps.
These can be trained on a large scale with Quantum-2 InfiniBand and real-time inference of trillion-parameter models within 72-GPU NVIDIA NVLink domain, which can reportedly be used as a one big GPU.
OCI announced the general availability of NVIDIA L4OS for midrange AI workloads, visualisation, and NVIDIA Omniverse.
Accelerated GenAI and Data Workloads
Oracle’s Autonomous Database now has NVIDIA GPU support for Oracle Machine Learning notebooks so that customers can speed up their data processing workloads on the Oracle Autonomous Database.
NVIDIA and Oracle are bringing three capabilities to demonstrate the NVIDIA accelerated computing platform.
This included a showcase of NVIDIA GPUs being used to accelerate bulk vector embeddings from the Oracle Autonomous Database Serverless to connect data with AI.
The second demonstration uses a proof-of-concept prototype comprising NVIDIA GPUs, NVIDIA cuVS, and an offload framework to speed up vector graph index generation.
A final demo gives an insight into how NVIDIA NIM can improve generative AI performance for text generation and translation use cases.
Sovereign AI
The two companies are also working together to produce sovereign AI infrastructure around the world to meet the needs of governments and enterprises.
Wide Labs is a Brazilian start-up which trained one of the first LLMs for Brazilian Portuguese using NVIDIA H100 Tensor Core GPUs and NVIDIA NeMo framework in OCI data centres to retain data sovereignty.
Nelson Leoni, CEO of Wide Labs, commented on the results:
“Developing a sovereign LLM allows us to offer clients a service that processes their data within Brazilian borders, giving Amazônia a unique market position.
“Using the NVIDIA NeMo framework, we successfully trained Amazônia IA.”
Zoom is also reportedly planning to implement NVIDIA GPUs in OCI’s Saudi Arabian data centre to support local data compliance requirements.
NVIDIA will be attending Oracle CloudWorld 2024, where it will be delivering a joint solution keynote with Oracle on Wednesday 11th September in Las Vegas.
