Microsoft Scraps Data Center Leases Amid Possible Oversupply

Microsoft has reportedly canceled multiple U.S. data center leases and pulled back from other pending leases. The voided deals have raised concerns that Microsoft has more data center capacity than it needs and sparked new doubts about bullish data center demand forecasts. 

The Washington-based tech giant terminated leases totaling close to 200 megawatts in multiple U.S. markets with at least two data center providers, TD Cowen reported Friday in a note to investors. 

Citing supply chain inquiries known as channel checks, the firm’s analysts said Microsoft also pulled back from several agreements known as statements of qualifications that precede data center lease deals. 

According to the note, Microsoft also recently abandoned negotiations on data center deals each reaching more than 100 MW, allowed letters of intent to expire on large-scale development sites representing more than a gigawatt of capacity, and walked away from at least five land parcels it had under contract in major markets.

These decisions suggest Microsoft may have more data center capacity in its pipeline to support artificial intelligence than it needs, the analysts said. 

“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's Michael Elias, Cooper Belanger and Gregory Williams wrote.

The drawdown “indicates the loss of a major demand signal that Microsoft was originally responding to,” they wrote.

That drop-off in demand is likely tied to the company’s fraying infrastructure partnership with OpenAI, according to TD Cowen.

Microsoft is OpenAI’s largest shareholder and provides most of its data center capacity. But over the past six months, OpenAI leaders have publicly chafed at what they say is Microsoft’s slow pace of data center expansion. As a result, OpenAI has shifted its infrastructure planning toward Oracle and SoftBank, most notably partnering on the $500B Stargate data center joint venture. OpenAI expects Stargate to host three-quarters of its computing power by 2030, The Information reported last week.   

But Microsoft’s data center pullback also comes amid rising doubts about long-term data center demand in the wake of the seemingly more efficient AI model released by Chinese AI firm DeepSeek.

The model’s unveiling caused tech and energy stocks to plummet last month amid fears that more efficient AI training would lead to lower demand for data center capacity. 

Still, speaking on quarterly earnings calls in the weeks since, leaders at Microsoft and other major tech firms indicated plans to increase capital spending on data centers and other AI infrastructure. Microsoft plans to spend close to $80B on AI data centers — a target it reiterated in a statement to Bloomberg Monday. 

“While we may strategically pace or adjust our infrastructure in some areas, we will continue to grow strongly in all regions,” a Microsoft spokesperson said.