(March 11): Singapore’s Princeton Digital Group plans to raise as much as US$5 billion ($6.36 billion) in debt this year to fund a data centre buildout across several countries in Asia, another sign of the capital pouring into expanding artificial intelligence (AI) infrastructure.
The planned financing will support facilities in India, Indonesia, Malaysia and Japan, the company said in a statement on Wednesday. The Warburg Pincus-backed firm operates AI data centres in seven Asian markets, with a portfolio exceeding 1.8 gigawatts of capacity.
“We are in the midst of an investment supercycle in AI,” Rangu Salgame, a co-founder and the chief executive officer of PDG, said in a video interview from Singapore. “The long-term demand for AI and the kind of innovation that’s yet to come are driving demand for data centre capacity.”
A bank consortium including Barclays plc, BNP Paribas SA, Deutsche Bank AG and Standard Chartered plc will jointly fund a portion of the loan.
Training and deploying large AI models requires hyperscale data centres with specialised graphics processing unit chips and high-density computing clusters. Building these facilities can cost tens of billions of dollars, often far beyond what operators can singlehandedly fund. Operators worldwide with pre-sold capacity are leveraging customer commitments to raise billions in debt and build infrastructure, even as the industry grapples with power constraints, regulatory hurdles and geopolitical uncertainties.
Borrowing against future revenue locked in through signed customer contracts has become a norm among several firms. PDG is following that model, turning AI data centres into an asset class similar to toll roads or power plants, with predictable cash flows.
See also: Amazon returns to debt market with jumbo cross-border bond sale — Bloomberg
“We have rock solid contracts with hyperscalers averaging 10 to 12 years,” Salgame said.
AI-related data-centre spending could reach US$2.9 trillion between 2025 and 2028, with roughly half requiring external financing, which is driving debt issuance, according to Morgan Stanley. CoreWeave Inc, one of the leading data centre operators in the US, plans capital spending of US$30 billion to US$35 billion this year, and has pushed its long-term borrowings to more than US$14 billion as it builds capacity for customers like Microsoft Corp and Meta Platforms Inc.
Lenders are willing to provide billions because revenue is often assured for a decade or longer to investment-grade tenants like Microsoft, Alphabet Inc’s Google, or Amazon.com Inc, providing a steady cash flow to service the debt.
See also: Meta hikes fees for advertisers to cover Europe’s digital taxes
Those three companies together with Meta have forecast capital expenditures that will reach about US$650 billion this year, a tidal wave of capital as they seek a competitive advantage in AI’s future. What’s not clear yet is whether they can generate enough revenue from corporate customers and consumers to make those investments profitable.
Adding to such concerns is the circular nature of deals in the AI economy. Nvidia Corp, the leading provider of AI accelerators, has invested in customers such as OpenAI, for example. That’s prompted sceptics to question whether such agreements are artificially inflating the demand for AI chips and other gear.
PDG’s Salgame said that such concerns have not affected his company’s ability to land financing.
PDG is securing more than US$4 billion in loans by linking them to individual data centre assets in Mumbai, Jakarta, Johor and Tokyo, while about US$750 million will be borrowed by PDG’s parent company.
The holding-company loan is an expansion of the US$400 million debt raised last year, with the full US$750 million converted into a sustainability-linked loan, the company said.
Uploaded by Tham Yek Lee
