Key Takeaways
- • American Banker columnist Paul Vigna argues the biggest risk in the AI credit boom is revenue: the capital markets are directing trillions of dollars into AI infrastructure, and the industry may not generate enough income to service that debt.
- • Insurance companies hold about $1.2 trillion in AI-related exposure, according to the National Association of Insurance Commissioners (NAIC), channeling risk from private credit into traditional credit markets.
- • Gartner estimates data centers will need to generate $8.5 trillion in revenue, but the consultancy projects no more than $3.6 trillion, implying large write-downs, defaults and possible bankruptcies.
- • Banks' exposure to AI lending is unquantified, and pending rule changes could expand their participation further.
- • Anthropic generated $16 billion in revenue in the first half of 2026, up from $1 billion a year earlier, yet reported an $81 billion net loss.
Summary
American Banker columnist Paul Vigna contends that the central risk in the artificial intelligence credit boom is whether the sector can generate revenue sufficient to repay its debt. He describes the current investment cycle as rivaling any boom in U.S. history, with capital markets committing trillions of dollars to AI buildout. Regulators have begun to recognize the problem, according to a prior American Banker report by Kate Berry, and the Securities and Exchange Commission has told firms to "maintain vigor" around valuations, a directive tied to SEC Chair Paul Atkins' push to open investing beyond accredited investors.
Vigna argues that banks should have limited direct exposure to this lending because post-2008 reforms were designed to restrict risky bets, and private credit grew to fill that gap. Risk has nonetheless leaked into traditional credit markets, particularly through insurance companies that private credit firms have acquired or affiliated with as a source of funding. According to the NAIC, insurers hold about $1.2 trillion tied to AI. Banking-sector exposure remains unknown, a gap Vigna notes is itself a problem, and proposed rule changes could widen bank participation. Transparency is also limited in private credit; the Federal Reserve has acknowledged it lacks full visibility into the market.
On revenue, Vigna cites Anthropic's first-half 2026 revenue of $16 billion, up from $1 billion the prior year, which still coincided with an $81 billion net loss. He notes Anthropic is one example rather than the whole industry. Citing Gartner, he writes that data centers would need $8.5 trillion in revenue, while the firm projects no more than $3.6 trillion, a shortfall he expects could produce write-downs, defaults, consolidations and bankruptcies. Vigna declines to forecast near-term outcomes, saying the central question for the next few years is whether the industry can generate enough income to repay its debt.
Vigna argues that banks should have limited direct exposure to this lending because post-2008 reforms were designed to restrict risky bets, and private credit grew to fill that gap. Risk has nonetheless leaked into traditional credit markets, particularly through insurance companies that private credit firms have acquired or affiliated with as a source of funding. According to the NAIC, insurers hold about $1.2 trillion tied to AI. Banking-sector exposure remains unknown, a gap Vigna notes is itself a problem, and proposed rule changes could widen bank participation. Transparency is also limited in private credit; the Federal Reserve has acknowledged it lacks full visibility into the market.
On revenue, Vigna cites Anthropic's first-half 2026 revenue of $16 billion, up from $1 billion the prior year, which still coincided with an $81 billion net loss. He notes Anthropic is one example rather than the whole industry. Citing Gartner, he writes that data centers would need $8.5 trillion in revenue, while the firm projects no more than $3.6 trillion, a shortfall he expects could produce write-downs, defaults, consolidations and bankruptcies. Vigna declines to forecast near-term outcomes, saying the central question for the next few years is whether the industry can generate enough income to repay its debt.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.