Money Stuff — Someone Has to Take the AI Risk

Author: Matt Levine (Bloomberg Money Stuff) URL: https://www.bloomberg.com/opinion/articles/2026-10-20/someone-has-to-take-the-ai-risk One-line: The AI data-center buildout has created a massive pile of correlated credit risk that the entire financial system is simultaneously long, making insurance against it expensive, scarce, and structurally unresolvable.

Key claims

• The AI data-center loan boom is almost entirely systematic risk, not idiosyncratic — if the AI boom disappoints, every data-center loan in every portfolio goes bad at once, regardless of who the specific borrower or tenant is.
• A Brookings paper estimates the data-center pipeline will require approximately $10.3 trillion in investment over 2025–2032, needing roughly $3.7 trillion in annual revenue by 2032 (at a 10% unlevered return) to justify the build.
• Five years ago, five-year credit default swaps on Oracle cost roughly 0.33% of principal per year; they now cost approximately 3.3% per year — a 10× increase driven by demand for AI infrastructure hedges.
• Weekly CDS trading volume on Oracle, SpaceX, Alphabet, Amazon, Meta, Broadcom, and Nvidia surged from under $2 billion in July 2026 to more than $9 billion in September 2026, per Morgan Stanley Research.
• KKR estimates that, with tech firms projected to pour nearly $8 trillion into AI infrastructure by 2030, a fifth of the investment-grade bond index could end up exposed to AI risks.
• 41 AI-related companies already account for nearly 49% of the S&P 500’s combined market value; AI-linked firms have climbed $27 trillion in value over the past three years, equivalent to 36% of the entire U.S. stock market.
• The supply problem is symmetrical: every would-be insurer (insurance companies, banks, hedge funds) is already long AI data-center risk through its own portfolio, so selling CDS means concentrating that risk further — the one entity that could plausibly absorb it is the U.S. government, which OpenAI has reportedly floated.
• The newsletter also covers two other substantial topics: insurance companies rolling “short-term” loans to related parties indefinitely (the Mark Walter / Delaware Life Insurance saga) to exploit lower capital-charge rules; and Polymarket CEO Shayne Coplan facing pressure from ICE and Founders Fund to mature his governance as Polymarket pursues a regulated U.S. exchange model.

Notable quotes

Investors are racing to buy credit default swaps to protect their portfolios in the event the AI boom unravels, sending trading volumes surging for an instrument that shot to prominence during the global financial crisis.

More than $9bn worth of CDS linked to bonds issued by Oracle, SpaceX, Alphabet, Amazon, Meta, Broadcom and Nvidia was traded on average each week in September, up from less than $2bn in July, according to Morgan Stanley Research.

“Clearly when you’re in a market where the risk is very concentrated, what you want to buy is CDS,” said Jim Caron, chief investment officer of the portfolio solutions group at Morgan Stanley Investment Management, adding that these swaps were also a “cheap hedge” for investors.

This risk is basically all systematic, not idiosyncratic. You might lend money to one particular data-center owner… but it doesn’t matter so much as long as there’s a real data center filled with real AI chips. If AI takes off, revenue is high and chips are in demand, but your data center’s tenant falls behind and goes bankrupt, that’s fine: The owner will lease the data center to some other AI company and never miss a payment.

Now, the world is enormously and increasingly long AI data-center risk, so lots of people want insurance against AI data-center risk, but who would want to sell it? If you call up an insurance company and ask for $1 billion of insurance against AI data-center risk, half of that insurance company’s investment portfolio is probably in AI data-center bonds. It doesn’t want more AI data-center risk.

Eventually, if there’s going to be a $10 trillion AI buildout, it seems unavoidable that we will all own some of that risk in our retirement accounts, one way or another.

The rolling strategy was used to get lower capital charges for assets that might not be able to secure an investment-grade credit rating, or any rating at all… For volatile long-term investments, such as stock or unrated loans, capital rules can require insurers to set aside 30% to guard against potential losses. But for certain short-term investments, the charge can be as low as 0.39%, no matter how risky the asset.

How it connects

• Matt Levine — author; Bloomberg’s Money Stuff newsletter
• Artificial Intelligence — the underlying technology driving the buildout
• Data Centers — the physical assets at the center of the credit risk
• Capital Allocation — how institutional investors are weighing concentrated AI exposure
• Private Equity — covered in the evergreen continuation vehicles section; the shift from transactional to permanent ownership model
• Blackstone — cited as “leaning into AI every way it can,” financing data centers
• Microsoft — named as a relatively safe AI-linked company whose CDS spread still hit record highs
• Nvidia — one of the seven companies whose CDS volume surged to $9B/week
• Amazon — included in the CDS surge list
• Meta (Company) — included in the CDS surge list
• Alphabet — included in the CDS surge list
• Broadcom — included in the CDS surge list
• CoreWeave — named as a riskier neocloud whose CDS spread climbed alongside Microsoft’s
• Kalshi — covered in the Polymarket section; the rival prediction market that bet on regulatory legitimacy and won