Blackstone Founder Discusses Data Center Strategy After Conversation with Jack Ma
Stephen Schwarzman, founder of Blackstone, recalled during a Bloomberg TV program that in 2015, while sitting on a bus in Beijing with Alibaba founder Jack Ma, he could not understand the discussion about artificial intelligence.
Ma explained to him that AI would be a transformative technology that would change the world. After returning to New York, Schwarzman consulted with academics, including those from MIT, who believed Ma underestimated its impact. Subsequently, Blackstone formed a data science team to systematically study and strategize in this area.
In 2021, Blackstone acquired the American data center company QTS for approximately $10 billion, which was one of the largest transactions in the field at that time. After the acquisition, QTS significantly expanded, with some reports stating its scale increased to 15 times the original size, with a substantial capacity boost and a client base covering major cloud service providers.
In 2024, Blackstone led the acquisition of AirTrunk, the largest data center operator in the Asia-Pacific region, for about $16.1 billion (AUD 24 billion), marking its largest investment in the region. AirTrunk has assets in Australia, Japan, Singapore, Malaysia, and Hong Kong, promising a capacity of over 800 megawatts and holding land to support further growth.
Blackstone's scale of leased, operated, or under-construction data center assets is massive, previously disclosing about $55 billion for operational and under-construction parts, with an additional potential development pipeline of about $70 billion. The company leases data rooms, power, and land to global tech giants through long-term contracts, creating stable cash flow while also engaging in related debt and equity investments.
While AI model companies debate model capabilities, token pricing, and profitability, Blackstone locks in long-term leases using a traditional real estate model, with clients primarily being well-capitalized tech firms. Schwarzman stated that he did not understand AI at the time but chose to build a team and invest real money in infrastructure.
ABAB AI Insight
Since co-founding Blackstone with Peter Peterson in 1985, Schwarzman has consistently generated stable cash flow through private equity and real estate. After his encounter with Jack Ma in 2015, he validated his judgment with academic institutions like MIT and donated $350 million to establish the Schwarzman College of Computing, as well as over £190 million to Oxford to support AI ethics research, transforming personal insights into institutional capital allocation.
In terms of capital strategy, Blackstone first privatized QTS for $10 billion and then acquired AirTrunk for about $16.1 billion, shifting funds from traditional real estate to data centers, power, and cooling facilities. Leases typically last 15-20 years, with tenants including major cloud providers like Microsoft and Amazon, while also providing financing for companies like CoreWeave through debt, creating a multi-layered structure of equity and debt.
This approach is similar to Blackstone's past strategies in logistics, student housing, and hotels, as well as the long-term holding models of Equinix and Digital Realty in the data center sector. The industry is currently transitioning from a model training competition to an infrastructure control phase, where computing power and electricity have become bottlenecks rather than mere algorithm breakthroughs.
Essentially, this is a reconstruction of the industrial chain. Model companies compete for pricing power at the application layer, while owners of physical infrastructure lock in cash flow through long-term leases. The mechanism is that the demand for power and cabinet space for AI training and inference is growing exponentially, while land, grid access, and approval cycles are lengthy, giving early movers a pricing advantage on scarce resources, converting technological uncertainty into stable rental income akin to real estate.
ABAB News · Cognitive Laws
- If you don’t understand the technology, buy from those who do.
- Models are burning cash, landlords are collecting rent.
- Infrastructure locks in profits earlier than algorithms.