NVIDIA and Microsoft have collaborated to launch the AI Foundry service, which offers enterprise customers tailored generative AI model solutions. This partnership also introduces H100 and H200 Tensor Core GPU instances. The H100 instance features 4P AI computing capabilities and 188GB of HBM3 memory, while the H200 instance supports larger model inference with significant increases in memory capacity and bandwidth. Additionally, Microsoft has rolled out the NCC H100 v5 instance, providing Azure customers with accelerated performance that ensures confidentiality and integrity.