Microsoft has signed some $33 billion in compute capacity deals with new AI cloud firms (aka neoclouds) to meet AI demand.

As reported by Bloomberg, this figure includes the $19.4bn deal signed with Nebius in September, which will give Microsoft access to more than 100,000 Nvidia GB300 chips, according to people familiar with the matter.

Outside of Nebius, the company has signed capacity agreements with multiple GPU-as-a-Service providers, including the likes of CoreWeave, Nscale, and Lambda.

According to the Bloomberg report, this is part of a strategy to help cope with the shortage of AI capacity, and will enable Microsoft to free up its data centers for use by customers.

This week, Microsoft's CTO, Kevin Scott, emphasized the severity of the capacity crunch during the Italian Tech Week event.

Scott told the audience: "Massive crunch is probably an understatement. I think we have been in a mode where it’s been almost impossible to build capacity fast enough since ChatGPT... launched.”

He added: "Even our most ambitious forecasts are just turning out to be insufficient on a regular basis. And so... we deployed an incredible amount of capacity over the past year, and it will be even more over the coming handful of years.” This is despite 2GW of capacity being stood up in the last 12 months alone.

This was emphasized by Scott Guthrie, who leads Microsoft's cloud efforts, in the Bloomberg report. “We are in very much land-grab mode in the AI space,” said Guthrie. “We’ve made the decision that we don’t want to be constrained in terms of capacity."

Guthrie told Bloomberg that by using neoclouds' capacity, the company is freeing up its own data centers.

According to unnamed sources, Microsoft's first foundation AI models built under consumer AI chief Mustafa Suleyman were trained at a CoreWeave data center near Portland, Oregon.

In addition, the neocloud capacity will be used to help Microsoft deliver AI services in those regions. Nscale, for instance, will be used for AI inference in Europe.

In comparison, Amazon and Google have not signed deals with neoclouds to the same degree. Google has leased some capacity from CoreWeave under a contract with OpenAI, but Microsoft is by far the largest adopter.

Microsoft's Guthrie put this down to surging AI demand for Microsoft products, including GitHub Copilot, and OpenAI's ChatGPT. “Amazon doesn’t have anywhere near that scale. Nor does Google, in terms of total users with intensity, especially during business hours,” he said.

In terms of its own expansion plans, Microsoft unveiled plans for its Wisconsin data center earlier this month, which it claims will be the "world’s most powerful data center," and will invest some $7.3bn in the project in total.