Microsoft invests millions in supercomputer to run ChatGPT

Microsoft spent hundreds of millions of dollars building a giant supercomputer to run OpenAI’s ChatGPT chatbot. To do this, the company has connected thousands of Nvidia’s graphics processing units (GPUs) to its Azure cloud computing platform, allowing OpenAI to train more powerful models than ever before.

The company also recently launched the ND H100 v5 VM, a scalable and powerful artificial intelligence (AI) virtual machine that enables the creation of artificial intelligence models with much faster performance than the previous generation.

With the ND H100 v5, customers will be able to choose the number of NVIDIA H100 graphics processing units (GPUs) for their needs, with an NVIDIA Quantum-2 InfiniBand interconnect network ranging from eight to a thousand.

Also read: Microsoft says new ChatGPT you can ‘see’ is coming next week

The new virtual machine includes NVSwitch and other technological innovations such as NVLink 4.0, Intel Xeon Scalable processor and DDR5 memory. With this new offering, Microsoft aims to bring AI supercomputing capability to startups and businesses of all sizes without the need for significant investments in hardware or software.

Microsoft Azure with ChatGPT

ChatGPT is available in preview on Azure OpenAI Service. Developers can integrate the chatbot into their applications for faster issue resolution, summarizing call center conversations, creating personalized ads, automating order processing.

Companies such as The ODP Corporation, Smart Nation Digital Government Office of Singapore, and Icertis are already using Azure OpenAI Service to improve their operations and services. Microsoft also uses artificial intelligence models in its products such as GitHub Copilot, Microsoft Teams Premium, Microsoft Viva Sales and Microsoft Bing.