AI burning cash starts raising prices! Meta refinances $12 billion, borrowing costs are clearly rising

As tech giants continue their fervent investment in AI infrastructure, the bond market is flagging higher prices for this cash-burning race. Meta’s latest $12 billion data center financing shows that risk premiums on AI-related debt are climbing significantly, with borrowing costs rising sharply compared to the previous deal nine months ago.

On Friday, according to the UK’s Financial Times citing sources, this bond—led by BlackRock and intended for a nearly 1 gigawatt data center project in El Paso, Texas—had a yield discussion in preliminary stages surpassing 7%, with some investors demanding a risk premium about 0.4 percentage points higher than Meta’s “Hyperion” data center deal completed last October. Sources added that pricing negotiations are still in early stages, with issuance possibly launching formally as soon as next Monday and final terms potentially changing.

The rise in borrowing costs reflects growing caution in the bond market about AI financing risks. At the same time, AI-related stocks have recently faced massive sell-offs, intensifying investor doubts over whether the boom in this sector can last, with sentiment in both markets resonating.

Borrowing Costs Rise as Market Reprices AI Risk

The cost burden of this financing has substantial significance in the bond market. One credit investor focused on investment-grade debt noted: When you issue tens of billions of dollars in bonds, even a 0.1 percentage point rise in cost means tens of millions in extra interest expenses each year—a very notable impact in the high-grade market.

For reference, Meta’s previous “Hyperion” project bonds saw a record $27 billion issuance in October last year, while bonds linked to that project, issued via a special purpose vehicle called “Beignet Investor”, dropped to about 96 cents on the dollar as of Thursday.

Sources pointed out that the increased borrowing costs directly reflect lenders’ cautious attitude toward their growing AI exposure—such caution is intensifying after tech giants triggered a borrowing wave in recent months.

Special Purpose Vehicle Structure Keeps Tech Firms’ Balance Sheets “Clean”

This financing follows the same structural design as Meta’s previous deal. The bond will be issued via a special purpose vehicle named “Sopaipilla Investor”—the name is derived from a popular South American fried pastry, in the same vein as the Louisiana dessert-named “Beignet Investor” from the previous deal. Sopaipilla will hold 80% of the Texas project’s equity and Meta the remaining 20%.

S&P analyst Viviane Gosselin said the deal is “almost a carbon copy of the previous transaction.”

Raising debt via project entities rather than the main company has become the mainstream way for tech companies to fund the AI arms race while keeping their balance sheets tidy. Last month, Anthropic completed $35 billion in financing through a deal backed by GPU leasing and Broadcom guarantees.

Robust Structural Design, but Specific Risk Clauses Exist

Structurally, the Sopaipilla-issued bonds mature in 2048, and are backed by Meta’s 20-year lease payments starting in 2028. Meta holds an option to renew every four years, up to four times; early exit would require hefty penalties, providing strong protection to lenders. Meta also assumes construction risk, covering any costs exceeding 105% of the initial budget.

However, the structure is not without limitations: the bonds are not directly pledged against physical assets. S&P reported that if the project suffers a major accident delaying completion by more than 18 months, Meta can terminate the lease without any penalty.

On ratings, S&P grades these bonds A+, one notch below Meta’s AA- corporate rating. Gosselin commented: “From our perspective, this is a very robust structure.” Fitch and KBRA rate the deal AA-, the same as Meta’s corporate rating.

Risk Warning and DisclaimerThe market has risks; investment requires caution. This article does not constitute personal investment advice, nor does it consider the unique investment goals, financial situation, or needs of individual users. Users should consider whether any views, opinions, or conclusions in this article suit their particular circumstances. Investment based on this article is at your own risk.