Microsoft Cancels US Data Center Leases, Sparking Concerns Over AI Capacity Oversupply

Microsoft Cancels US Data Center Leases, Raising AI Capacity Concerns

Microsoft Corp. has canceled some leases for US data center capacity, according to TD Cowen, sparking concerns about whether the tech giant is securing more AI computing capacity than it needs in the long term. The US brokerage reported that Microsoft has voided leases totaling “a couple of hundred megawatts” of capacity, equivalent to roughly two data centers, and has pulled back on converting statements of qualifications into formal leases.

The exact reasons for Microsoft’s lease cancellations are unclear, but TD Cowen suggested that the company may be in an oversupply position. The brokerage also posited that Microsoft could be reallocating some of its in-house investment to the US from abroad. Microsoft reiterated its spending target for the fiscal year ending June but declined to comment on TD Cowen’s note.

The potential lease pullback raises questions about whether Microsoft, a frontrunner in AI among Big Tech companies, is growing cautious about the outlook for overall demand. The company has said it expects to spend $80 billion this fiscal year on AI data centers and has pledged to sustain spending to meet “exponentially more demand.”

European stocks tied to the energy sector dropped on the report, which may suggest that Big Tech companies will need less power to run their data centers. Critics have consistently pointed out a dearth of practical, real-world applications for AI, even as Microsoft, Meta Platforms Inc., Alphabet Inc., and Amazon.com Inc. have pledged to spend billions on the data centers needed to train, develop, and host AI services.

Microsoft executives have played down concerns about AI overcapacity, stating that the company’s plans to spend over $80 billion on infrastructure this fiscal year remain on track as it continues to grow at a record pace to meet customer demand. Rivals have also doubled down on their AI spending commitments, with Amazon, Alphabet, and Meta pledging to spend $100 billion, $75 billion, and up to $65 billion, respectively, on AI infrastructure. Chinese e-commerce giant Alibaba Group Holding Ltd. said it would invest more than 380 billion yuan ($53 billion) over the next three years as it seeks to become a leader in the field.

TD Cowen’s analysts wrote that their channel checks had unearthed a number of signs that Microsoft is gradually retreating, including letting more than a gigawatt of agreements on larger sites expire and walking away from “multiple” deals involving about 100 megawatts each. Microsoft used facility and power delays as justification for the termination of leases, a tactic rivals such as Meta previously employed when curbing capital spending.

Large cloud hyperscalers typically use a mixture of leased and owned data centers across many locations, so investors should expect some degree of “tweaking” of plans, according to Mizuho Securities analyst Jordan Klein. Microsoft’s alliance with OpenAI may also be evolving in ways that mean the software giant won’t need the same kind of investments. In January, Microsoft said it would alter its multiyear deal with OpenAI so the AI startup could use cloud-computing services from rival providers. Microsoft, which had been the company’s exclusive cloud provider, still has a right of first refusal when OpenAI seeks computing horsepower to train and run its AI models.