Microsoft weighs fewer AI cloud suppliers as it prioritizes its own data centers

Microsoft is evaluating a smaller group of outside AI computing suppliers while giving priority to its own data centers, CNBC reported on November 6. The review could narrow an important customer base for specialist cloud providers, but Microsoft has not decided which providers it will stop using, according to people familiar with the discussions.

How outside capacity fits into Azure

The suppliers under review operate in the market known as neoclouds: specialist providers of AI computing capacity built around graphics processing units, or GPUs. Microsoft has established relationships with CoreWeave, Nebius, Chronoscale, IREN, Lambda and Nscale. It has also used Oracle, which is investing heavily in AI infrastructure.

Microsoft turned to these providers after ChatGPT’s 2022 launch drove a surge in GPU demand. CoreWeave was still privately held when Microsoft signed it up. The relationship became substantial: when CoreWeave went public in 2025, it disclosed that Microsoft accounted for 62% of its 2024 revenue. That history illustrates the stakes of losing business from a large customer.

Outside suppliers are one part of Microsoft’s infrastructure mix, alongside facilities it owns and leases. Responding to CNBC, a company spokesperson described that strategy as built around flexibility and choice. Microsoft said it reviews investments against customer demand, tracks delivery and adjusts its plans as circumstances change.

We combine Microsoft-owned datacenters, leased facilities and strategic partnerships with third-party providers to deliver the capacity our customers need, where and when they need it.

Microsoft spokesperson, in an emailed statement to CNBC

Fewer suppliers does not mean less spending

CNBC’s Jordan Novet and Seema Mody reported that Microsoft intends to keep working with some key suppliers as it builds its own capacity. Two confidential sources described the process as an evaluation; one said Microsoft wants to prioritize its own data-center footprint. The report does not identify a finalized list of providers to retain.

The review comes alongside continued investment. Microsoft reported $45.2 billion in capital expenditures and finance leases for its latest quarter. It also said capital spending would accelerate in the second half of its fiscal year. Those commitments distinguish a change in who supplies computing capacity from an overall spending pullback.

Getting that capacity online remains a constraint. Finance chief Amy Hood said on Microsoft’s October earnings call that some GPU-related cloud contracts had been delayed until the first half of 2027. Those delays were affecting Azure’s cloud growth, linking infrastructure delivery directly to the business’s ability to grow.

Contracts turn on power and delivery

Some outside-capacity decisions have already been made. Two people told CNBC that Microsoft opted not to renew a deal with Oracle. Microsoft also previously walked away from a planned Nscale site in England. In October, its UK and Ireland chief, Darren Hardman, attributed that decision to Nscale’s ability to obtain electricity.

IREN provides another example of the gap between handing over infrastructure and bringing it into service. In its November 5 earnings report, the company said it delivered data-center capacity to Microsoft in August. Microsoft then requested a networking-equipment upgrade, pushing the expected start of operations from this year to January or February.

IREN co-CEO Daniel Roberts described a tougher financing test on the earnings call. Capital markets had become more selective, he said, shifting attention from contract size to whether projects could be delivered, financed and operated profitably. That is the business pressure surrounding Microsoft’s supplier review—not simply the amount of computing capacity promised.