Microsoft and AMD Partnering to Develop Custom AI Chips
Earlier this week, news surfaced that Microsoft might be interested in partnering with AMD to develop custom silicon for its AI ambitions. As we see tech giants like Apple, Facebook, Amazon, Google, Baidu, and others working on their own chips, Microsoft’s interest in custom silicon is hardly surprising.
In Microsoft’s case, this interest started small with its partnership with Qualcomm to fine-tune Snapdragon chipsets for Surface PCs. Last year, this collaboration expanded to include purpose-built chips for augmented and virtual reality glasses. The company has already developed expertise in creating low-power digital signal processors that aggregate data from various gesture and environment sensors in HoloLens headsets.
According to a new report from Bloomberg, Microsoft has now enlisted the help of AMD to build AI chips. The project, codenamed “Athena,” involves Microsoft providing AMD with engineering resources to spearhead the development of specialized hardware for AI training and inference. This partnership aims to help AMD gain a foothold in a market that is dominated almost entirely by Nvidia.
The processing power needed to train and run artificial intelligence systems has increased exponentially in recent years, with AI startups often spending over 80 percent of their capital on compute costs alone. Nvidia’s annual data center revenue reached the $13 billion mark last year, while AI-specific companies like SambaNova Systems, Cerebras, and Graphcore made $5.1 billion, $4 billion, and $2.8 billion, respectively, in 2022. The partnership with Microsoft could be a real opportunity for AMD to cash in on the AI craze.
AMD CEO Lisa Su mentioned the company’s data center segment saw a drop in operating income during Q1 2023 primarily due to “increased R&D investments to address large opportunities ahead of us.” This is likely related to the company’s upcoming Bergamo silicon. Su explained that the company has also created a new AI group from teams scattered across various divisions. She pointed to generative AIs as AMD’s number one strategic priority and an area where the Xilinx acquisition and the “very capable semi-custom team” can bring a lot to the table.
Microsoft has reportedly spent $2 billion on its chip efforts and currently employs around 1,000 engineers in the Microsoft Silicon team. The company poached high-profile Apple chip architect Mike Fillipo and put Rani Borkar, a former Intel executive, in charge of the group. Microsoft acknowledges the existence of the Athena project but says AMD is not involved in it. However, this doesn’t exclude the possibility that the two companies are working on custom AI accelerators based on AMD’s MI300 APUs.
The first version of the Athena silicon is expected to be ready in 2024 at the earliest. It will be interesting to see what capabilities this custom silicon will bring to the table.
