amd
The hardware that has enabled and driven the recent explosion of generative AI products has been primarily manufactured by Nvidia. That’s partly due to its dominance in the consumer and workstation graphics markets, plus its established CUDA libraries and its AI-accelerating Tensor cores.
While Microsoft uses “tens of thousands” of Nvidia GPUs in its Azure data centers, Bloomberg reports that the company is also working with AMD to improve the AI capabilities of its GPUs. While extremely light on details, the report says Microsoft is contributing “support” and “technical resources” to AMD to improve the performance of its products in AI workloads.
Bloomberg also says AMD is working on a Microsoft-developed AI accelerator called “Athena,” details of which The Information reported last month. However, a Microsoft spokesperson flatly denied that AMD was involved in the development of Athena.
The upside for Microsoft is that a more competitive market for AI-accelerating hardware can save on server costs as the company looks to build AI features into more and more of its products. For example, Microsoft is reportedly working on a private version of ChatGPT to offer to businesses that could cost “10 times” as much to use as the regular version of ChatGPT; if the server hardware needed to run these generative AI models is cheaper, Microsoft could lower its prices, costs, or both to make these products more attractive.
The two companies have more or less worked together before; “Surface Edition” Ryzen processors have been used in some older versions of Microsoft’s Surface PCs, but not in any of the most recent Surface devices. Like the Qualcomm SQ-series chips that Microsoft uses in the Arm versions of the Surface Pro, the Surface Edition Ryzen processors weren’t much different from regular Ryzen processors: they’re the same chips that AMD sold to other manufacturers, but with Microsoft-supported “firmware, drivers, and software stack” optimizations.
While not representative of how each company’s server-grade GPUs perform, recent testing from Tom’s Hardware illustrates how AMD’s GPU architectures struggle with AI workloads. When tested with the Stable Diffusion imager, AMD’s current flagship Radeon RX 7900 XTX performed slower than Nvidia’s RTX 4090, 4080, and 4070 Ti, though the RX 7900 XTX is faster than the RTX 4080 in most games. Even the entry-level RTX 3050 managed to beat all previous generation RX 6000 series cards from AMD (software optimizations could help these cards perform better, but the fact that the software was optimized for Nvidia by default says a lot about the current state of things) .
