Business IEA Says Data Center Power Demand Could Double to 1,000 TWh by 2030, With AI Supplying Three Quarters of the Growth Chatbots ChatGPT Now Transcribes as You Talk: OpenAI Removes the 10-Minute Cap and Streams Words Live on Every Plan Business Ex-Alibaba Executive's Everful Raises $48.5 Million to Put AI Agents in Charge of Sourcing From 30,000 Factories AI Agents Google Built a Mirror Dimension of the Internet to Train Its Chrome AI Agents Without Breaking the Real Web Productivity Canva Finally Adds an Off Switch for Its Generative AI Features, and a Filter to Hide AI-Made Templates Business Anthropic Begins Trading Today: 980 Million Shares at $83, a $1.5 Trillion Valuation and the Largest IPO on Record News Google Books Blue Origin New Glenn to Launch TPU-Carrying AI Satellites in Early 2029 Security Hackers Used the AI Penetration Tool ARTEX to Disrupt European Commission Systems for Over Ten Days Business IEA Says Data Center Power Demand Could Double to 1,000 TWh by 2030, With AI Supplying Three Quarters of the Growth Chatbots ChatGPT Now Transcribes as You Talk: OpenAI Removes the 10-Minute Cap and Streams Words Live on Every Plan Business Ex-Alibaba Executive's Everful Raises $48.5 Million to Put AI Agents in Charge of Sourcing From 30,000 Factories AI Agents Google Built a Mirror Dimension of the Internet to Train Its Chrome AI Agents Without Breaking the Real Web Productivity Canva Finally Adds an Off Switch for Its Generative AI Features, and a Filter to Hide AI-Made Templates Business Anthropic Begins Trading Today: 980 Million Shares at $83, a $1.5 Trillion Valuation and the Largest IPO on Record News Google Books Blue Origin New Glenn to Launch TPU-Carrying AI Satellites in Early 2029 Security Hackers Used the AI Penetration Tool ARTEX to Disrupt European Commission Systems for Over Ten Days

US Life Insurers Hold $85.2 Billion of AI Debt, Up 24 Percent in Six Months, and Moody’s Sees $185 Billion

US Life Insurers Hold $85.2 Billion of AI Debt, Up 24 Percent in Six Months, and Moody’s Sees $185 Billion

US life insurers held 85.2 billion dollars of AI-related debt as of June 30, up 24 percent from six months earlier and equal to 2.1 percent of their total debt, according to Moody’s Ratings. The agency expects holdings could reach 185 billion dollars by the end of next year. Insurers supplied 3.4 billion dollars of the 27 billion dollar financing behind Meta Hyperion data center.

US life insurers held 85.2 billion dollars of debt tied to artificial intelligence as of June 30, up 24 percent from six months earlier, according to Moody's Ratings. The holdings equal 2.1 percent of the industry's total debt, and Moody's expects life insurers could carry as much as 185 billion dollars of AI exposure by the end of next year. The New York Times reported the figures on November 16.

The structure of these deals is what makes the number worth watching. In most cases the technology company is not technically the borrower. Advisers create a financing vehicle that borrows against chips or a data-center shell and leases the assets back to the operator, which keeps the official debt load off the tech company's balance sheet. When Elon Musk needed to finance a large purchase of AI chips this year, the vehicle approach was used and 2.4 billion dollars of that debt was sold to Athene, the annuity and life insurance manager owned by Apollo, according to the Moody's report. An Apollo spokeswoman said Athene finances what she called the safest and most defensible parts of the AI ecosystem.

Meta's Hyperion campus in Richland Parish, Louisiana, shows how the money flows. Meta holds a 20 percent stake; the investment firm Blue Owl owns the rest. S&P Global rated the project's 27 billion dollars of debt one notch below Meta's corporate rating, citing guarantees that included a 28 billion dollar lease commitment over 20 years. That rating let insurers buy the paper, and Moody's says they funded 3.4 billion dollars of it. Meta has pushed back on how the guarantee is characterized: a company spokesman said Meta has not guaranteed repayment of the bonds.

The gap between a lease obligation and a debt guarantee is the whole argument. Ares Management, which runs 691 billion dollars and lends in the same market, wrote in a client newsletter that structural complexity and financial engineering are sometimes used not to improve risk but to hide it, describing transactions that look safer than they are. Ratings agencies assign investment-grade-adjacent marks based on the corporate parent's credit, while the legal claim sits with a project entity that may have fewer obligations if construction is delayed.

Exposure is also concentrated. A handful of companies, including Nvidia, Meta, Alphabet, SpaceX, Anthropic and OpenAI, provide the guarantees behind most of these deals, and tech companies frequently retain ways to walk away when projects slip. Barclays estimates that life insurers owned about 15 percent of AI data center debt as of June 30. Andrew Granato, an assistant professor at the University of Texas at Austin School of Law who studies the overlap between private equity and insurers, put the shift plainly: the life insurance industry is now in effect backing large swaths of the AI industry.

The regulatory question is capacity rather than rules. State insurance regulators supervise solvency and asset quality, and the deals arriving on their desks are structured through special purpose vehicles with layered leases, backstops and parent guarantees. Rating agency treatment does much of the work that would otherwise fall to examiners, and Ares and others have argued publicly that the ratings can flatter the underlying risk.

Why this matters beyond insurance: annuities and life policies are long-duration liabilities, which makes insurers natural buyers of long-duration paper, and AI infrastructure has become one of the largest sources of it. That is a genuine maturity match. It also means the AI buildout now depends partly on retirement savings, and a single delayed data center or a renegotiated lease is absorbed by policyholder balance sheets rather than by a technology company's cash flow. If the credit cycle turns, the failure mode is not a startup running out of money. It is a solvency question at institutions that sold guarantees decades long, priced on the assumption that the guarantee behind them was as good as the parent's name.

Comments (0)

Log in to join the discussion

Log In

No comments yet