Microsoft has canceled leases for large-scale data center capacity in the U.S., signaling a potential oversupply as it continues building AI infrastructure. Analysts at TD Cowen suggested that Microsoft is adjusting its strategy after previously making substantial investments to support AI demand.
Microsoft earlier announced plans to invest $80 billion in AI during its current fiscal year, focusing heavily on building out AI data centers and infrastructure to support its AI development and cloud services.
The move comes amid growing investor skepticism about the massive spending on AI infrastructure by U.S. tech firms, especially as Chinese startup DeepSeek has demonstrated AI capabilities rivaling Western competitors at a significantly lower cost, Reuters news report said.
Microsoft, headquartered in Redmond, Washington, reportedly scrapped leases amounting to “a couple of hundred megawatts” with at least two private data center operators. Additionally, the company has paused converting statements of qualifications, which typically precede formal leasing agreements. Analysts noted that other major tech firms, including Meta, have also taken steps to reduce capital expenditures.
Microsoft, which is the primary backer of OpenAI, did not immediately respond to requests for comment.
However, brokerage firm Jefferies, currently hosting Microsoft’s investor relations team in Sydney, reported that the company strongly denied any changes to its data center expansion plans. Despite this denial, Microsoft’s stock declined by approximately 1.8 percent in early U.S. trading, reflecting market concerns.
This development marks a sharp shift in strategy for a company that recently allocated $80 billion in capital spending for the fiscal year, primarily focused on AI. Microsoft has previously emphasized that these investments were essential to overcoming supply bottlenecks that hinder its ability to meet AI-driven demand.
Bernstein analyst Mark Moelder suggested that the lease cancellations could signal weakening demand, particularly following underwhelming earnings from major cloud providers. However, he also noted that Microsoft may have initially overcommitted to securing data center and GPU capacity, even at premium prices, in response to previous difficulties in meeting demand. The situation highlights the complexities of balancing AI infrastructure investment with market fluctuations and evolving competitive pressures.
