Microsoft is developing its own AI chip 'Athena'

Microsoft is developing AI-optimized chips to reduce the cost of training generative AI models, such as the ones that power the OpenAI ChatGPT chatbot.

The Information recently quoted two people familiar with the matter as saying that Microsoft has been developing a new chipset code-named "Athena" since at least 2019. Employees at Microsoft and OpenAI already have access to the new chips and are using them to test their performance on large language models such as GPT-4.

Training large language models requires ingesting and analyzing large amounts of data in order to create new output content for AI to imitate human conversation. This is a hallmark of generative AI models. This process requires a large number (on the order of tens of thousands). ) The cost of computer chips optimized for AI can be very high.

Microsoft is said to be developing new chips for use in its own products to reduce reliance on third-party chips and reduce costs. Currently, computer chip designer Nvidia dominates the AI chip market, releasing its latest H100 GPU last year. Microsoft will join Amazon, Google and Meta Platforms in developing and building their own customized chips for AI.

SemiAnalysis principal analyst Dylan Patel told The Information: "Microsoft wants to use large-scale language models in all its applications, including Bing, Microsoft 365 and GitHub, and to deploy them at scale using off-the-shelf hardware, then every year The cost will reach tens of billions of dollars.”

Microsoft invested $10 billion in OpenAI in early 2023 and quickly began integrating its AI technology into various Microsoft services, including Bing search with Bing Chat functionality. Engine, Microsoft 365, Dynamics 365 for enterprise users, and Security Copilot for security professionals.

The report pointed out that the purpose of these Microsoft chips is not to replace Nvidia's chips. On the contrary, these chips are to enhance Microsoft's existing infrastructure. According to The Information, Microsoft has planned future generations of the chip.

It’s unclear whether Microsoft plans to release these chips for Azure cloud AI customers or if they will be for internal use only, but Microsoft does offer AI-optimized cloud instances based on the Nvidia H100 chipset. It's unclear what impact this will have on the AI supercomputer partnership Microsoft and Nvidia announced late last year.

According to reports, Microsoft expects the new chip to be unveiled in 2024.

The above is the detailed content of Microsoft is developing its own AI chip 'Athena'. For more information, please follow other related articles on the PHP Chinese website!