Microsoft (MSFT.O) plans to launch the company's first chip designed for artificial intelligence at its annual developer conference next month, a person familiar with the matter said, The Information reported. The move is the culmination of years of work and could help Microsoft reduce its reliance on artificial intelligence chips designed by Nvidia Corp (NVDA.O). As demand surges, these chips have been in short supply.
access:
Microsoft China Official Mall-Home Page
Microsoft's chip is designed for data center servers that train and run large language models (LLM). Microsoft's data center servers currently use Nvidia's GPUs to power advanced LLM for cloud customers, including OpenAI and Intuit, as well as support artificial intelligence capabilities in Microsoft productivity applications.
Microsoft chips, similar to Nvidia GPUs, are designed for data center servers that train and run large language models, the software behind conversational AI features such as OpenAI's ChatGPT.
Earlier, other media pointed out that Microsoft has been developing its own AI chip, codenamed "Athena", since 2019, aiming to reduce its dependence on NVIDIA. Microsoft's real purpose is not to directly replace NVIDIA chips, but to reduce costs and continue to add AI functions to multiple services.
An updated risk factor in Microsoft's annual report released in July showed that Microsoft data centers depend on licensed and buildable land, predictable energy, network supply and servers, including GPUs and other parts, highlighting the importance of GPUs to data centers.
UBS said in August that Microsoft being limited by the number of GPUs it can use is a reality that could impact their ability to generate AI revenue next year. If it can build its own GPU, it can reduce this risk.
There were rumors in the market in September this year that Microsoft began to revise orders for Nvidia H100 chips and slowed down shipments. This shows that Microsoft's AI chips have matured in research and development and testing and will get rid of their heavy dependence on Nvidia.
If Microsoft launches its own AI chip, it will be on the same level as Amazon (AMZN-US) and Google (GOOGL-US), both of which have self-developed AI chips, namely Inferentia and TPU.
Microsoft (MSFT-US) rose 2.47% to $327.26 per share on Friday. Nvidia (NVDA-US) closed at a 2.40% dividend to $457.62 per share. Amazon (AMZN-US) rose 1.59% to $127.96 per share. Google (GOOGL-US) closed up 1.86% at $137.58 per share.