Microsoft is collaborating with AMD to help the latter in its expansion towards the creation of processors equipped with Artificial Intelligence, according to Bloomberg. Both companies are therefore teaming up to offer an AI chip alternative to Nvidia, which dominates the market for AI-powered GPUs.
Those of Redmond are in charge of offering support to AMD in several areas, such as resource engineering. Also, according to various sources, they are working with AMD on the development of a processor designed by them and designed to work with Artificial Intelligence workloads, which is known as Athena. But according to Frank Shaw, a Microsoft spokesman, AMD is not involved in Athena.
This deal is part of a broader move to expand the processing power of Artificial Intelligence, which is in high demand after the popularization of chatbots, such as ChatGPT, among other AI-related services. Microsoft is both one of the leading providers of cloud computing services and one of the main drivers of the use of AI. As shown, the $10 billion they have invested in OpenAI, in addition to their promise to add AI capabilities to their entire software portfolio.
But with this step, in addition, Microsoft shows the increase in its involvement in the chip sector. The company has been creating a hardware division for several years now under the leadership of former Intel manager Rani Borkar. Currently, this division already has a thousand employees.
The first news about Ahtena began to appear a few weeks ago. Currently, several hundred of Microsoft’s hardware division employees are working on the Athena project, and Microsoft has invested some $2 billion in semiconductor-related initiatives and projects. However, this does not mean that the company will stop collaborating with Nvidia.
Microsoft intends to continue working closely with Nvidia, whose chips are currently the most widely used for training and running Artificial Intelligence systems. It’s also trying to get more processors from Nvidia, highlighting the shortages facing both Microsoft and other tech companies.
Microsoft’s relationship with OpenAI requires the Redmonds to have computing power at a very high level. In fact, it’s more than they bargained for when they placed orders for chips to power up many of their data centers. Hence, you need more chips capable of providing you with the necessary power.
The company recently introduced a version of its Bing browser with ChatGPT, and new AI-powered tools in its Office suite apps. In addition, it is also updating older products, such as the GitHub code generation tool. All of these AI-powered programs run in the Microsoft Azure cloud, and require the powerful, but also expensive, processors that Nvidia provides.
But AI is also a priority for AMD. According to its CEO, Lisa Su, right now is his main strategic priority. Su has also ensured that AMD has the opportunity to create partially customized chips for its principals, with the aim that they can use them in their data centers where they work with Artificial Intelligence.
Meanwhile, Borkar’s team at Microsoft, which has also worked on chips for servers and Surface computers, is now prioritizing the Athena project. He is developing a GPU that can be used to train and run Artificial Intelligence models. The chip is already undergoing internal testing, and could be more widely available over the next year.
Of course, although the availability of the chip increases next year, it is a first version that will be a starting point. Developing a good chip takes years, and Nvidia has quite a head start on it. It is the main chip supplier to many generative AI tool providers, such as AWS and Google Cloud, and Elon Musk has snapped up thousands of its processors for his future AI business.
As a consequence, developing an alternative to Nvidia’s AI products is going to be a difficult task. Furthermore, because it is capable of offering a package of software and hardware that work in combination, and that includes chips, a programming language, network equipment and servers.
Thanks to this, its customers can update their systems quickly, and Nvidia has become a company with a great dominance of the field of chips and equipment for tasks related to Artificial Intelligence. And Microsoft isn’t the only one trying: Amazon bought Annapurna Labs in 2016 and has developed two chips for AI, and Alphabet also has its own chip for training models.