The new Microsoft chip, which is similar to Nvidia GPUs, is designed for data center servers that train and run large language models, the software that powers conversational AI features like OpenAI’s ChatGPT. Presently, Microsoft's data center servers use Nvidia GPUs to support advanced LLMs for cloud customers, including OpenAI and Intuit, as well as for AI features in Microsoft’s productivity apps.
Key takeaways:
- Microsoft is planning to unveil its first chip designed for artificial intelligence at its annual developers’ conference next month.
- The new chip could help Microsoft reduce its dependence on Nvidia-designed AI chips, which are currently in high demand and short supply.
- The Microsoft chip, similar to Nvidia GPUs, is designed for data center servers that train and run large language models.
- Microsoft's data center servers currently use Nvidia GPUs to power AI features in Microsoft’s productivity apps and for cloud customers, including OpenAI and Intuit.