Arthur Hayes claims that the American insurance industry is technically insolvent due to debts related to artificial intelligence (AI). According to expert accountants, the internal reinsurance system has recorded $1.54 trillion in active liabilities with $657 billion in surplus capital. If this reinsurance is excluded, 29 out of the 30 leading American insurers are effectively insolvent.
Only three reinsurance companies in Vermont hold just 3.7% of the assets needed to cover their liabilities. Hayes notes that until a wave of revaluation of AI center debts occurs, nothing will change. Insurers already control a growing share of the market for AI center debts, which threatens their solvency if AI companies stop purchasing computing power. Private credit funds have already restricted investor withdrawals once this year – the same warning sign as the growing tension in the private credit market now faced by insurers. Government guarantees are supposed to support insolvent insurers, but payouts are limited to $250–300 thousand per policy. These funds are financed by the remaining insurers, many of whom use the same reinsurance model. This means that retirees receiving annuities from these insurers could suffer real losses if the reinsurance is as thin as described.
Essentially, it's a domino chain, and the question is what will happen if the first tile falls. Insurers have used accounting tricks to make their reserves appear stronger than they actually are. If AI companies slow down their purchases of computing power, the loans behind these data centers will be downgraded, and insurers holding this debt will suddenly require more real money than they have. Steve Eisman, one of the real investors depicted in the movie "The Big Short," discussed this research in a podcast, calling it a "slow-moving scandal that could one day lead to a great financial crisis." This leaves two possible scenarios: regulators will either allow insurers to go bankrupt, accepting the consequences, or the government will intervene to prevent it.
Related articles
