Oracle announces partnership with Nvidia to purchase A100/H100 GPUs

Oracle Corporation announced on Tuesday that it plans to deploy tens of thousands of Nvidia's top A100 and H100 computing GPUs in its Oracle Cloud Infrastructure (OCI) . The A100 and H100 GPUs will be offered to Oracle's cloud customers for their AI workloads powered by Nvidia's AI software, and the exact terms of the deal are still being kept under wraps.

The new collaboration between Nvidia and Oracle will give all enterprise customers access to AI training, computer vision, data processing, deep learning inference, and simulation, and they won't have to invest huge sums of money to deploy their data centers with Nvidia's expensive compute GPUs. Oracle has already provided OCI customers with access to HPC instances and will now provide them with a variety of AI capabilities.

Enterprise customers of Oracle OCI will have access to all of Nvidia's AI platforms, including the following:

  • AI Enterprise: A suite of engines that can be used for AI model training, computational vision, conversational AI, data processing, recommender systems, simulation, and more.
  • RAPIDS: Accelerates Apache Spark data processing on Apache Spark services fully managed by OCI Dataflow, including bare metal instances like BM.GPU.GM4.8 with A100 Tensor Core GPUs.
  • Clara: Medical Imaging, Genomics, Natural Language Processing, and Drug Discovery (coming soon).

Nvidia's A100 and H100 computing GPUs are expensive, even the previous generation A100 computing GPUs cost $10,000 to $15,000 (about 108,000 yuan), depending on the specific configuration, while the next generation The H100 product will naturally be more expensive. So, while Oracle is unlikely to buy Nvidia's compute GPUs at retail prices, the total price is certainly not lower, and the deal between Nvidia and Oracle is expected to be worth hundreds of millions of dollars.

Post a Comment