Microsoft has begun canceling leases for a substantial amount of data centre capacity in the US, a move that may reflect concerns about whether it’s building more AI computing than it will need over the long term, TD Cowen said in a report.
OpenAI’s biggest backer has voided leases totalling “a couple of hundred megawatts” of capacity, analysts from the American multinational investment bank and financial services division of TD Securities wrote on Friday, citing channel checks or inquiries with supply chain providers.
Microsoft has also stopped converting so-called statement of qualifications, which are agreements that usually lead to formal leases, TD Cowen said. That was a tactic rivals such as Meta Platforms employed previously, when it decided to cut back on capital spending, the brokerage wrote.
A pullback by Microsoft on spending and data centre construction raises questions about whether the company — one of the frontrunners among Big Tech in AI — is growing cautious about the outlook for demand.
The company has said it expects to spend US$80 billion ($106.84 billion) this fiscal year on AI data centres, and, on a late January earnings call, CEO Satya Nadella said Microsoft has to sustain spending to meet “exponentially more demand”.
Microsoft in a statement on Monday reiterated its spending target for the fiscal year ending June, but declined to comment on TD Cowen’s note.
See also: Vietnam-based fund seeks to develop US$1.5 bil data centre hub
“While we may strategically pace or adjust our infrastructure in some areas, we will continue to grow strongly in all regions,” a company spokesperson said in the statement. “Our plans to spend over US$80B on infrastructure this FY remains on track as we continue to grow at a record pace to meet customer demand.”
Critics have consistently pointed out a dearth of practical, real-world applications for AI, even as Microsoft, Meta and Amazon.com have pledged to spend billions on the data centres needed to train, develop and host AI services.
Wall Street stepped up its questions about the massive outlays after the Chinese upstart DeepSeek released a new open-source AI model that it claims rivals the abilities of US technology at a fraction of the cost.
See also: Singtel secures $643 mil five-year green loan to finance Tuas data centre
“While we have yet to get the level of color via our channel checks that we would like into why this is occurring, our initial reaction is that this is tied to Microsoft potentially being in an oversupply position,” TD Cowen analysts Michael Elias, Cooper Belanger and Gregory Williams wrote, stressing it was just their interpretation.
Microsoft executives have consistently played down concerns about AI overcapacity. It’s spending more than it ever has in its history, outlays that mostly go to the chips and data centres required to fuel power-hungry AI services.
In Friday’s report, TD Cowen’s analysts wrote that their channel checks had unearthed a number of signals about Microsoft’s gradual retreat from data centre construction and acquisition.
They learned that Microsoft had let more than a gigawatt of agreements on larger sites expire, and walked away from “multiple” deals involving about 100 megawatts each (data centre capacity is often stated in terms of the power they need to stay up and running).
“Our channel checks indicate that Microsoft is also re-allocating a considerable portion of their projected international spend to the US, which suggests to us a material slowdown in international leasing,” the analysts wrote.
