Microsoft warns of service disruptions if it can’t get enough A.I. chips for its data centers

Microsoft is emphasizing to traders that graphics processing items are a important uncooked materials for its fast-growing cloud business. In its annual report launched late Thursday, the software program maker added language about GPUs to a danger issue for outages that may come up if it may’t get the infrastructure it wants.

The language displays the rising demand on the prime expertise corporations for the {hardware} that is needed to supply synthetic intelligence capabilities to smaller companies.

AI, and particularly generative AI that entails producing human-like textual content, speech, movies and pictures in response to folks’s enter, has change into extra common this 12 months, after startup OpenAI’s ChatGPT chatbot turned successful. That has benefited GPU makers similar to Nvidia and, to a smaller extent, AMD.

“Our datacenters depend on the availability of permitted and buildable land, predictable energy, networking supplies, and servers, including graphics processing units (‘GPUs’) and other components,” Microsoft mentioned in its report for the 2023 fiscal 12 months, which ended June 30.

That’s considered one of three passages mentioning GPUs within the regulatory submitting. They weren’t talked about as soon as within the earlier 12 months’s report. Such language has not appeared in current annual experiences from different giant expertise corporations, similar to Alphabet, Apple, Amazon and Meta.

OpenAI depends on Microsoft’s Azure cloud to carry out the computations for ChatGPT and varied AI fashions, as a part of a posh partnership. Microsoft has additionally begun utilizing OpenAI’s fashions to reinforce present merchandise, similar to its Outlook and Word purposes and the Bing search engine, with generative AI.

Those efforts and the curiosity in ChatGPT have led Microsoft to hunt extra GPUs than it had anticipated.

“I am thrilled that Microsoft announced Azure is opening private previews to their H100 AI supercomputer,” Jensen Huang, Nvidia’s CEO, mentioned at his firm’s GTC developer convention in March.

Microsoft has begun trying outdoors its personal knowledge facilities to safe sufficient capability, signing an settlement with Nvidia-backed CoreWeave, which rents out GPUs to third-party builders as a cloud service.

At the identical time, Microsoft has spent years constructing its personal customized AI processor. All the eye on ChatGPT has led Microsoft to hurry up the deployment of its chip, The Information reported in April, citing unnamed sources. Alphabet, Amazon and Meta have all introduced their very own AI chips over the previous decade.

Microsoft expects to extend its capital expenditures sequentially this quarter, to pay for knowledge facilities, normal central processing items, networking {hardware} and GPUs, Amy Hood, the corporate’s finance chief, mentioned Tuesday on a convention name with analysts. “It’s overall increases of acceleration of overall capacity,” she mentioned.

WATCH: NVIDIA’s GPU and parallel processing stays important for A.I., says T. Rowe’s Dom Rizzo