Blackstone’s $68 billion flagship property fund is selling assets and preparing to spend billions on data centers.
The Financial Times reports that the Blackstone Real Estate Income Trust, or Breit, plans to pump $8bn into building new data centers through its asset QTS Data Centers, as it predicts a surge in artificial intelligence workloads.
Breit spent around $60 billion on properties from hotels, warehouses, and self-storage facilities from the beginning of 2021 to the third quarter of 2022. But then the trust reversed course, and has since made no large purchases, instead selling more than $10bn in assets.
The company plans to sell a Texas resort for $800m, and self-storage properties for $2.2bn, as it raises cash to build data centers for a potential AI wave.
“Large technology companies are in the midst of an AI arms race which we believe will be a once-in-a-generation engine for future growth in data centers and is driving tremendous demand on the ground,” Breit told investors.
The FT reports that Blackstone has committed to spending more than $8bn to build new data centers, likely for hyperscalers. The investment is through QTS Realty Trust, which Blackstone acquired for $10bn in September 2021.
The company has already spent more than $1bn to buy land to build new data centers in five states across the US, and QTS has tripled its leased space since it was acquired. The company is preparing to further double its size, the FT reports, and QTS’ valuation has risen to about $20bn.
DCD spoke to QTS' chief growth officer earlier this year about the company's plans to build multiple gigawatt campuses. “All of the power in the state of Arizona today is seven gigs,” Tag Greason told us. “And we're going to them and saying, 'we'd like to build one gig.'”
The new investment comes as analysts and industry figures predict a wave of AI investment. Analyst outfit Dell'Oro Group this month said that data center capex will hit $500 billion by 2027 to support AI workloads, while TD Cowen said that around ~2.1GW of data center leases were signed in the last 90 days driven by AI requirements - a new record.