According to reports, people familiar with the matter revealed that Microsoft is cooperating with AMD to jointly develop artificial intelligence processors. This is part of Microsoft’s multi-pronged strategy to secure the supply of artificial intelligence processor chips.
The two companies are working together to explore alternatives to Nvidia’s GPU chips, the people said. Currently, Nvidia dominates the market for artificial intelligence processor chips. In the cooperation, Microsoft has provided financial support to AMD, and the two parties are cooperating to develop an artificial intelligence processor code-named “Athena”, which is self-developed by Microsoft.
AMD shares rose more than 6.5 percent Thursday on the news, while Microsoft shares rose about 1 percent. Both companies declined to comment for this story. Nvidia shares fell 1.9 percent.
Recently, ChatGPT and other artificial intelligence chatbots have received widespread attention, and the industry’s demand for artificial intelligence processors has also expanded. Microsoft is not only one of the most important cloud computing service providers but also the main driving force for the application of artificial intelligence. Microsoft previously invested $10 billion in OpenAI, the developer of ChatGPT, and announced that it will integrate the artificial intelligence technology behind ChatGPT in all its software.
This move also shows that Microsoft’s participation in the chip industry is gradually deepening. Over the past few years, under the leadership of former Intel executive Rani Borkar (Rani Borkar), Microsoft is building a chip division that now has nearly 1,000 employees. Last month, some media also exposed the specific situation of Microsoft’s development of the artificial intelligence chip “Athena”.
Hundreds of those employees are working on Project Athena, and Microsoft has invested about $2 billion in the chip project, according to a person familiar with the matter. However, this does not mean that Microsoft will part ways with Nvidia. At present, Nvidia’s GPU chips are still the main tool for artificial intelligence model training and inference, so Microsoft will continue to work closely with Nvidia. However, the entire industry is currently facing a shortage of artificial intelligence processors, and Microsoft is also looking for ways to obtain more Nvidia processors.
Microsoft’s partnership with OpenAI, as well as Microsoft’s own plans to launch a series of artificial intelligence services, means that Microsoft needs far more artificial intelligence computing power than the company previously expected when ordering chips and building data centers. After seeing the capabilities of ChatGPT, many companies are interested in integrating this technology in their own products or enterprise applications, and Microsoft has used ChatGPT’s technology to launch a new version of the conversational Bing search, as well as an artificial intelligence-enhanced version of the Office tools.
Microsoft is also updating older products, such as GitHub’s code generation tools, at the same time. All these artificial intelligence programs run on Microsoft’s Azure cloud computing platform, so they need more artificial intelligence processor support. Currently, these processors from Nvidia are expensive and in tight supply.
Artificial intelligence is also a key priority for AMD. “We are very excited about our opportunity in artificial intelligence, which is also our highest strategic priority. We are in the very early days of the era of artificial intelligence computing,” AMD CEO Lisa Su said on an earnings call on Tuesday. At this stage, the adoption rate and growth rate of artificial intelligence is faster than that of any other technology in recent history.”
Su also said there is an opportunity for AMD to produce some of its custom chips for its largest customer for use in artificial intelligence data centers.
Borka’s team at Microsoft is also prioritizing Project Athena, in addition to developing chips for servers and Surface computers. The team is developing a GPU chip that can be used for training and inference of artificial intelligence models. The chip is already being tested internally and could be widely available as soon as next year, a person familiar with the matter said.
But people familiar with the matter also said that even if the project progresses smoothly according to the timetable, the first version is only a starting point. It takes years to design and produce a powerful chip, and Nvidia has a big head start. Currently, Nvidia is the chip supplier of choice for many providers of generative artificial intelligence tools, including Amazon AWS and Google Cloud, and Tesla CEO Elon Musk is also hoarding chips for his own artificial intelligence business. thousands of Nvidia processors.
Developing an alternative to Nvidia chips will be a challenging task. Nvidia currently provides a complete set of supporting software and hardware around these chips, including chips, programming languages, network equipment and servers, so that customers can quickly build their own capabilities.
This is one of the reasons why Nvidia is currently in the lead. Plus, Microsoft isn’t the only company trying to develop its own AI processors. Amazon acquired Annapurna Labs in 2016 and has since developed two AI processors. Google also has its own artificial intelligence training chip.
