There is no questioning Nvidia’s early dominance of the AI chip market, but that does not mean other big-name chip firms have been left out of the action. Microsoft reminded the world of this at the recent Microsoft Build event, where it claimed to be “the first cloud provider” to offer its customers AMD’s Instinct MI300X AI accelerator GPU for their AI training and inferencing needs.
It is doing so in the form of the Microsoft Azure ND MI300X v5 series of virtual machines, for which it announced general availability last week. Hugging Face also was announced as one of the first customers leveraging the new virtual machines to increase its AI model computing performance.
Julien Simon, chief evangelist officer, Hugging Face, stated, “The deep collaboration between Microsoft, AMD and Hugging Face on the ROCm open software ecosystem will enable Hugging Face users to run hundreds of thousands of AI models available on the Hugging Face Hub on Azure with AMD Instinct GPUs without code changes, making it easier for Azure customers to build AI with open models and open source.”
In addition, AMD President Victor Peng commented, “The AMD Instinct MI300X and ROCm software stack is powering the Azure OpenAI Chat GPT 3.5 and 4 services, which are some of the world’s most demanding AI workloads. With the general availability of the new VMs from Azure, AI customers have broader access to MI300X to deliver high-performance and efficient solutions for AI applications.”
That announcement put Microsoft on a short list of AI infrastructure companies that have put the MI300X into action since the GPU became available at the end of 2023. An AMD spokesperson told Fierce Electronics that LaminiAI and GigaIO currently have the MI300x as part of their AI infrastructures, and that Dell, Google, Lenovo, Meta, Supermicro, OpenAI, and Oracle “are gearing up to deploy it in the future.”
These deployments and others will be significant to AMD as it looks to gain traction in the AI chip market, and also serve as a reminder that not everyone will be able to afford or get their hands on Nvidia GPUs.
Jack Gold, founder and principal analyst at J. Gold Associates, said Microsoft is in the habit of supporting multiple chip vendors to serve its customers’ capacity needs, and that the MI300X is “also a great alternative to a supply shortage that is pushing out delivery of Nvidia chips by up to a year by some accounts. If AMD can provide a large supply of chips, it can fill a gap that many cloud providers need (as well as enterprises). It’s also likely that AMD will be attractively priced next to the premium priced Nvidia devices, which is a potential advantage in some scenarios.”
Gold added that AMD is in a good position to gain from the “insatiable demand for AI chips and the fact that Nvidia can’t supply everything the market needs.”
He continued, “I expect AMD to continue to push at the Nvidia juggernaut in AI, and it's likely it will gain a significant market share over the next 1-2 years, similar to what it did in servers against Intel several years ago. That doesn’t necessarily mean bad things for Nvidia, as the market continues to expand and can handle more than one supplier. So, it’s not like Nvidia's sales will tank or anything like that. They will continue to increase their sales with the growth of the AI market, at least for the next couple of years until the market stabilizes supply/demand.”
Ultimately, the AI boom is an opportunity for many companies to grab more market share, as Gold noted in a recent Fierce Electronics analysis piece.