Arthur Hayes: AI's Debt, Insurance's Time Bomb, Bitcoin's Fuel

Original author: Arthur Hayes

Original compilation: Saoirse, Foresight News

Did you hear that? Those AI bigwigs suddenly grew a conscience and started worrying about the survival of humanity — because the product they've been building, an almost silicon-based deity, is about to be born. Actually, it's been close to being a silicon-based deity for a while now, but with the final breakthrough seemingly imminent, they suddenly began to reflect on the development path of artificial general intelligence (AGI).

(Note: Silicon-God, the super AGI that Silicon Valley says is about to be born. The author uses this term sarcastically, questioning whether this grand narrative is just fundraising and lobbying rhetoric.)

That's the narrative they're telling the public. But being naturally suspicious, I checked the calendar — it's almost the end of Q3, and Anthropic still isn't public. I can't help but wonder what their financials actually look like. Has the high-growth annualized revenue curve repeatedly mentioned in their press releases already slowed? They claim that after excluding all operating costs, the company is profitable. I can't wait to dig into the S-1 prospectus they're about to file, to figure out exactly how much it costs to serve each Token to users. Also, how many customers are actually profitable for them, and is that profitable customer base expanding or shrinking? Unfortunately, I won't get answers to any of these questions, and the reason is — safety first.

Readers can tell from my tone that I believe Anthropic, OpenAI, and SpaceX citing "safety first" as a reason to slow down AGI development is not out of concern for the welfare of ordinary humans, but rather stems from brutal economic reality: the market is unwilling to buy the AI products they're selling in sufficient quantities at current prices. More specifically, market demand for AI is robust, but what people want is Chinese pricing — at just one percent of the U.S. price.

When this "Chinese pricing" hits the self-important American AI practitioners, their first reaction is to shout: "But those Chinese products are low quality." And when Chinese models rapidly catch up in quality, they then cry: "China only built its own products by distilling our models." The market doesn't care why Chinese models are cheap; the market just wants the cheapest intelligence services. So these AI practitioners turn to lamenting: "We care about human safety, so we're pausing development." Sounds so noble... but this exaggerated "fake injury performance" worthy of a World Cup match will later come with demands: "But we still need to beat China, so the government should step in, introduce regulations, and continue funding this AGI race."

The reason these three leading U.S. AI labs using "safety first" as an excuse to slow AGI development is so critical to financial markets and global fiat liquidity is that these labs' compute demand underpins over $1 trillion in investment-grade debt, plus hundreds of billions in lower-rated loans.

This shows the proportion of purchase commitments OpenAI and Anthropic have made to the four major U.S. cloud providers relative to each provider's unfulfilled revenue orders, revealing the massive long-term orders the two major AI model companies bring to cloud providers.

These AI labs combined generate no profit whatsoever. Therefore, they need profitable tech companies like Nvidia, Broadcom, Google, and Microsoft as backers to provide off-balance-sheet guarantees for data center leases and chip procurement debt. Subsequent chip and hardware purchases depend on AI labs continuing to train the most cutting-edge large models — the ones that are "almost, almost, almost about to become a silicon-based deity" — while processing inference requests for customers. But if "safety first" becomes the new core principle, spending on training new models won't drop to zero, but it will certainly fall from current highs; companies will focus on improving the efficiency of converting electricity into intelligence, which means customers' compute spending will decrease. Essentially, safety first means destroying compute demand.

If AI capital expenditure were financed by operating cash flow, there'd be nothing to worry about. But the problem is, whether AI labs keep buying compute or not, this multi-trillion-dollar debt still exists. Default won't happen immediately, but once AI labs stop consuming compute at previously expected levels, the price of this debt will fall. The real core question: who bought this debt, and did they use leverage? The answer is obviously that these speculators used leverage to buy this junk debt. So who ultimately ends up holding the bag?

Millions of American insurance policyholders are indirectly betting on the AI story. And "safety first" will hurt them without any buffer. I didn't fully understand this scheme until Nick Nameth explained it very thoroughly on Substack. Next, I'll write it in simple language for my crypto audience. The conclusion is: if AI-related debt is repriced at fair market value, a large portion of the U.S. insurance industry is actually insolvent. This leads to the core proposition of investing in a global economy dominated by fractional reserve banking: the U.S. government faces a binary choice — either act as the buyer of last resort for compute in the name of national security, or print money to bail out deeply loss-making insurance companies.

Whichever path is chosen, we Bitcoin holders and crypto investors are the winners. If the government ignores market signals and insists on pouring money into developing this commercially unviable "silicon-based deity," it will need to print money to fund such unproductive spending, which will inevitably fuel more financial speculation and drive up Bitcoin's price. If the government chooses to bail out the insurance industry, it will print money to absorb bad AI debt, expanding the money supply and in turn driving up Bitcoin's price.

The rest of this article will break down this mechanism.

In the Name of China

Under the banner of national security, the U.S. government can find justification for almost anything. Consider how much destruction was caused after 9/11, when the U.S. dramatized threats to the public and launched the global war on terror. This time, the fabricated enemy is the Chinese, who bury themselves in mathematics, steal America's top technology, and sell products back to America at one percent of the price. To defeat China, state socialism must be pursued within the capitalist system.

AI bigwigs successfully persuaded Trump and his staff to ignore two realities: the market has already proven that the AI business is unprofitable, and bipartisan voters oppose building massive new data centers and demand compensation for data theft. Since China can provide affordable AI products, America's response is to invest more money to build that "almost, almost, almost, almost about to be born silicon-based deity." (I'll keep adding "almost," because we're only lacking faith, data theft, and taxpayer funding to reach AGI.)

War, economics, robotics — everything ultimately hinges on AGI. Therefore, the rest of the world must use AGI according to America's wishes. According to this narrative, America has the world's most inclusive and fair culture, and this silicon-based deity absolutely must not be controlled by a non-Judeo-Christian civilization, such as China. (Eye roll, and a big eye roll.) I have my preferred place to live, and others have theirs. Even if I believe the moral culture I identify with is superior to other countries', I wouldn't want to exhaust my wealth or even my life to impose those values on the entire world. You can believe American or Western culture is superior, but don't hand trillions of taxpayer dollars to Elon, Sam, and Dario. (The heads of the three leading American frontier AI companies: xAI, OpenAI, Anthropic)

The alleged superiority of the American system is that, most of the time, hundreds of millions of informed citizens determine the prices of goods and services through free markets, with the government staying out and letting market signals decide what to produce and how much. But now, because of national security, based on an assumption — that pouring in enormous amounts of money and feeding data to predict the next Token can create AGI — the market's signals are deemed wrong. So the U.S. government must increase investment to build the next generation of frontier models and use cultural superiority to suppress China. This is hubris. Remember what happened to Icarus when he flew too close to the sun?

Alright, enough grandstanding. Let's talk about the bailout plan.

"Safety first" means the three major U.S. AI labs' compute demand declines. The government can then step in and sign offtake agreements guaranteeing stable profits for the AI labs, similar to contracts the U.S. gives to certain defense and mining companies. The government will use this compute to advance AGI development. Finally, the government can lease its own models back to the AI labs, which would sell inference services to domestic and allied customers at extremely high prices.

This plan would give the government control over frontier models, for the Western world to use as it sees fit. The private AI lab model has a problem: these labs are global enterprises, sometimes selling model access to anyone worldwide for profit, which conflicts with the government's national security objectives. If Chinese labs partly rely on distilling American frontier models to make technological progress, direct government control over R&D could set China back months or even years in the AGI race. Even if this设想 holds, it will ultimately fail, just as trying to block advanced chip manufacturing equipment failed to stop China from producing cutting-edge chips. Information inherently wants to flow freely. In the internet age, information blockade is simply impossible. Even before the internet existed, after the U.S. successfully developed the atomic bomb, it couldn't prevent the Soviet Union from obtaining relevant intelligence. Those who believe AGI can be an exception don't understand human ingenuity and adaptability in the game of interests at the national level.

Funding this silicon-based deity requires issuing more debt. This plan is easy to sell because the monetary policymakers — Treasury Secretary Bessent and Fed Chair Warsh — both believe AI can boost productivity. They are convinced that by fully embracing AI, America can grow its way out of its massive debt burden, and this judgment isn't entirely wrong. In June 2026, U.S. nominal year-over-year GDP growth was 6.6%, while the effective federal funds rate was about 3.6%. Bessent keeps issuing short-term Treasury bills; the government earns 3% on this debt issuance, but savers bear the loss. If the fiscal deficit is kept below 3% (a big if), the debt-to-GDP ratio will decline. Economic growth is mainly driven by AI data center construction, and what supports all of this is AI labs' compute demand. So from a financing perspective, if issuing short-term Treasuries still yields 3%, the government acting as the buyer of last resort for compute is financially feasible.

There's no free lunch. The U.S. government runs chronic deficits and can only spend by borrowing. If Warsh cooperates, this is easy. But so far, the Fed under his leadership and the Treasury under Bessent are not marching in step.

For the first time since July 2023, U.S. monetary policy has raised rates: at last week's meeting, the Fed unanimously approved a 0.25% increase in the policy rate. The total money supply created by the Fed is no longer growing, and as of August 14, the RMP short-term Treasury purchase program has stopped.

If the government pushes this plan but the Fed doesn't lower funding costs or expand its balance sheet, massive debt issuance will push interest rates higher. Rising interest on mortgages, credit cards, and auto loans will only stoke voter anger toward AI-related policies. If Trump and Bessent can't secure the support of at least 7 FOMC members, this plan's feasibility will be greatly diminished.

The above is Fed policy from a traditional perspective, which easily makes people bearish on the market. But don't forget, there's another powerful money-printing machine: commercial banks. Both Warsh and Bessent advocate that the banking sector should take on the task of money creation. Since RMP purchases stopped on August 14, banks have created hundreds of billions of dollars in new money by expanding total assets, backed by relaxed liquidity regulatory constraints.

Beyond balance sheet expansion, after this 0.25% rate hike, banks' excess reserves held at the Fed will earn an additional $7.5 billion in interest annually. This money will be used for new loans and financial market speculation. Therefore, one cannot look only at the Fed raising rates and stopping balance sheet expansion; the impact of the commercial banking system must also be considered. Combined, the overall effect is still stimulative. That is to say, if the government is willing, the liquidity environment is sufficient to support new borrowing for investment in AI compute buildout.

But if the government doesn't step in to buy compute, the debt will be impaired. Those holding this debt with leverage will face a crisis. Below, let's dive deeper into the captive insurance scheme.

Captive Insurance

I had never studied the insurance industry before. Large private equity firms using captive insurance company assets to raise funds for investments didn't initially seem like a scam to me. But after digging deep into how this mechanism works, I found the ultimate victims.

Every credit bubble has a group of ultimate bag holders. Usually, it's ordinary retail money managed by trustees with impressive credentials. Trustees invest other people's money and earn double returns: collecting management fees, and selling their own assets to retail investors to inflate the price of their own holdings. This time, the victims are policyholders who bought American life insurance and annuity products. To understand this scheme, we first need to understand the life-extending tactics private equity bigwigs devised after the golden age of private equity ended.

After the 2008 global financial crisis, the private sector deleveraged, and the Fed cut rates to near zero. The classic private equity playbook: find mature companies with stable cash flow and almost no debt, add leverage, then extract dividends, and finally leave the broken company in the private markets. When public markets heat up, re-list the broken company, completing a cycle. In the low-rate era, this logic held perfectly. Ordinary people, burdened with negative-equity mortgages and struggling with monthly payments, had no capacity to increase consumption. Private equity bigwigs didn't want to expand production or offer better products; they just wanted to maintain existing cash flow, cut costs, and extract cash dividends for their own investors.

U.S. total credit as a share of GDP. Orange is private market credit, blue is government-related credit. After the 2008 financial crisis, private credit continued to decline while government credit continued to climb, reflecting the shift of debt structure from the private sector to the government sector.

Assets under management (AUM) of private equity and venture capital (PE&VC). Even through economic recessions, it has continued to expand since 2000, surpassing $15 trillion by 2025. Gray shading marks NBER-designated recession periods.

In the post-pandemic era, rising capital costs and diminishing marginal returns dealt a heavy blow to private equity fund returns. Because financing was cheap, to win a deal you had to offer higher valuations to acquire cash-flow assets. Private equity fund returns consequently declined. To complete the next round of fundraising, private equity bigwigs began looking for long-term capital pools that wouldn't care about short-term redemptions — enter insurance companies. Insurance companies sell life insurance and annuity policies; policyholders pay premiums, and the insurance company invests that money to earn returns, paying out policies decades later. This is exactly the perpetual capital pool private equity dreams of, which can be invested in private equity funds full of overvalued private companies and high-yield private credit.

So private equity bigwigs acquired insurance companies, appointed themselves as investment managers, and packaged junk assets to sell to unsuspecting policyholders. This is captive insurance.

The most absurd part of this arrangement is how captive insurance companies satisfy legally required capital buffers. Asset prices fluctuate, and regulators require insurance companies to set aside capital buffers to guarantee policy payouts. This gave rise to the reinsurance industry, which assumes the claims risk of primary insurers. Normally, the primary insurer and reinsurer are two independent institutions, and reinsurance prices risk at fair market value. But this rule doesn't work for the private equity insurance scam, whose core is finding a bag holder while private equity itself doesn't have to put in much of its own capital. So private equity-controlled insurance companies set up affiliated captive reinsurance entities. The parent company only needs to invest minimal own capital to obtain reinsurance coverage.

These are all regulated entities required to make regular public disclosures. If policyholders knew the insurance company was operating such a system, would they still buy policies? To conceal the scam, the American capital system sided with private equity. Regulations in some states like Vermont contradict national prudential regulatory standards. Primary insurers and affiliated reinsurers can privately establish reinsurance risk assets, set capital buffer sizes arbitrarily, and after state regulator approval, the relevant materials are sealed and kept confidential.

There are more details. Before continuing to break down this audacious scam, let me use a crypto analogy. Remember Terra Luna? Luna collapsed because USDT holders sold off the stablecoin, breaking the dollar peg. What if Do Kwon had the resources of those nearly seventy-year-old New York private equity bigwigs?

When USDT's price fell, Luna acquired an insurance company called Alameda Insurance. Luna used premium funds to buy USDT, trying to stabilize the peg. Alameda sold life insurance to Californians, holding billions in funds. Alameda couldn't directly buy altcoin stablecoins, but it could buy investment-grade corporate bonds. Luna bribed Moody's analysts to rate its own corporate bonds as investment grade. To attract buyers like Alameda, Luna offered interest 5% higher than the