Blackstone-owned QTS Data Centers is looking to raise $4.6 billion to fund its AI data center build-out, beginning with the sale of a 10-year investment-grade green bond.

Reporting by Bloomberg revealed that the company had held investor calls for the funding. Initial discussions put the bond at a yield of a comparable 10-year US Treasury bond, plus an additional 1.625 percentage points.

The bond will be issued by two QTS subsidiaries associated with the construction of two data centers on a campus in Fayetteville, Georgia.

Moody’s Ratings gave the offering a first-time credit score of Baa2, meaning it is considered subject to moderate credit risk. A Baa2 rating is two notches higher than a “junk,” non-investment grade rating.

Moody’s said, among other uses, proceeds would be used for refinancing credit facilities tied to the campus project.

JPMorgan Chase, Wells Fargo, SMBC Nikko Securities, and Goldman Sachs are managing the sale.

Founded in 2003, QTS was acquired by Blackstone for $10 billion in 2010. The company has multiple data center sites worldwide, either in operation or under construction, including in the US, UK, the Netherlands, and Spain.

Blackstone, through QTS and other means, has been aggressively pursuing AI data center development in recent years, with concrete plans to become one of the biggest investors in AI infrastructure.

Since the acquisition, QTS has seen its leased capacity grow by 14 times, with Blackstone’s ownership of QTS now accounting for 20.4 percent of the real estate value of Blackstone Real Estate Income Trust.

In February, QTS was reportedly seeking a more than $2 billion Commercial Mortgage-Backed Securities note to refinance an existing $1.36 billion debt associated with three data center campuses across Atlanta, Illinois, and Virginia.

And just last month, reports emerged that Blackstone is pursuing plans to launch a publicly traded AI data center acquisition company, with potential investment from sovereign wealth funds. The company will invest in pre-built and leased data center assets.

Some estimates put the required investment for the projected AI buildout at more than $3 trillion, and investment firms and asset managers worldwide are working on considerable investments to achieve this.

However, persistent fears of data center overbuild and weak returns on AI investments have led to discussion around investor risk. While hyperscalers view AI underinvestment as an existential threat, prompting massive capital expenditure and the promise of huge returns, the large upfront capital investment required to meet demand is putting pressure on credit metrics.

Moody’s warned that, given upfront investment requirements for AI data centers, it could be between 12 and 24 months between initial spending and revenue generation, and higher capital intensity and debt levels could lead to a “reassessment of creditworthiness” if profit growth fails to materialize.