Ray Dalio

Dalio’s central warning is that transformative AI can still be a terrible investment at the wrong price. Previous breakthroughs—from household electricity to radio, cars, and airplanes—attracted capital because their potential was real, yet “everything has its price.” Investors must distinguish good technology from good investment.

The bubble mechanism begins when paper wealth grows much faster than available cash. A company can raise $50 at a $1 billion valuation, instantly creating nominal wealth, but “wealth is not money”: spending requires owners to sell assets or borrow. Taxes, IPO liquidity, frightened holders, or debt obligations can therefore force selling long before AI’s ultimate value becomes knowable.

Dalio’s bubble indicator puts current conditions in the same order of magnitude as 2000 and 1929. He characterizes 2008 as a smaller, more debt-intensive housing bubble; today’s danger is unusually concentrated across GDP, wealth, equities, bonds, trade, employment, and geopolitics. His portfolio conclusion is blunt: “Don’t do everything AI,” whether that means going all-in or taking a leveraged short.

The sovereign-debt backdrop is the “plaque” that could turn an AI correction into an economic heart attack. Dalio’s approximate arithmetic has Washington spending $7 billion against $5 billion of revenue, carrying debt equal to roughly six times receipts, paying about $1 trillion of interest, and refinancing approximately $10 trillion annually. Meanwhile, foreigners buy about one-third of U.S. debt, and demand from China, Japan, and others is weakening.

A productivity miracle only solves the debt problem if it becomes taxable income quickly enough. Thompson’s bull case is a 10-fold productivity era powered by “geniuses in the data center”; Dalio concedes that sufficient income growth would work, but insists investors trace the whole chain from productivity to wages or profits, government revenue, distribution, and timing. Competition may instead make services cheaper without transferring the productivity gain into enough income.

Dalio’s fiscal prescription is to reduce the deficit below 3% of GDP before the ship “hits the rocks.” He sketches a politically neutral mix—perhaps raising taxes and cutting spending by roughly 4%–5% each—while ensuring AI succeeds and benefits most citizens. Government ownership of AI equity does not itself solve the cash problem: “I’m trying to buy a banana,” he jokes, and a small Anthropic stake is not a dollar.

China’s advantage may lie less in winning the frontier-model contest than in deploying AI as infrastructure. Dalio says Chinese leaders treat AI like electricity, water, roads, or railways, then integrate cheaper systems into an industrial base representing 32% of global industrial production—more than the United States, Germany, and Japan combined. That deployment supports productivity, export competition, economic warfare, and military power.

Dalio sees AI as a powerful productivity tailwind that is “virtually certain”—though he says even that is debatable—to replace many workers, but not as an automatic cure for America’s fiscal problem. Diamandis fears cheaper capital will fund robots and agents rather than jobs, breaking the familiar chain from employment to wages, consumption and debt service. Dalio’s answer is distributional: higher output per hour is good, but the resulting wealth divide could intensify civil conflict unless society manages how people deal with each other.

The investment question is whether AI arrives fast enough to overpower simultaneous debt, political, geopolitical, climate and demographic headwinds. Dalio expects a revolution larger than the shift from rulers and graph paper to spreadsheets and computerized decision-making, yet refuses to quantify its timing or magnitude, saying he does not know whether it will be 1.2, 1.25 or 1.5 times as large. The 1920s combined record patent activity with the 1929 crash: “We cannot simply say that these innovations will quickly and in time create such a productivity miracle that the other forces don’t matter.”

U.S. markets look to Dalio roughly like 1998: late-cycle, expensive and euphoric. He says that analogy implies about 1.5 years of good times, but tells entrepreneurs not to assume that, over one to five years, there will be no economic downturn, bear market or political disappointment. His valuation rule is blunt: investors can be “much better off buying bad companies at good prices than good companies at bad prices.”

The sovereign-debt risk comes from too many Treasury bonds meeting insufficient demand while interest costs compound. If private buyers retreat, rates rise and worsen debt service; if the Federal Reserve buys the bonds, it prints money and depreciates the currency. Dalio calls credit the economy’s blood and accumulated debt its plaque: once borrowing is required merely to pay interest, the system approaches a “debt death spiral.”

Dalio’s proposed escape is a politically negotiated “3% solution,” not confidence that DOGE alone can fix the budget. With prior Trump tax cuts extended, he estimates the deficit at about 7.5% of GDP and argues it must be brought toward 3% through some mix of spending restraint, tax revenue—not necessarily higher tax rates—and lower interest expense. On DOGE’s effects, his answer is deliberately limited: “I honestly don’t know,” because each cut produces second-order reactions.

Dalio prefers gold to Bitcoin while placing both inside a 10%-15% “anti-money” allocation intended to preserve purchasing power. Bitcoin is monitorable, taxable and vulnerable to government regulation, while physical gold is “the only asset that you can have that’s not somebody else’s liability” and remains a reserve asset when states distrust one another. He owns some Bitcoin, but cannot explain its price movements as readily as gold’s and therefore holds “much more gold.”

Dalio says the U.S. and China are already at war—subversively and technologically, if not yet militarily—and neither governments nor hyperscalers can afford to lose the technology race. He sees the U.S. as advanced in chips, though by how much is uncertain, while China performs better in applications and practical usage; Diamandis adds that chip restrictions force China to build capabilities internally and become more efficient, as he sees with Huawei and DeepSeek. Dalio expects near-term “pushing the edge” without necessarily going over it, while warning that the Treasury market and budget fight could leave investors “not as happy a year from now as we think.”

Dalio’s framework places the U.S. near the dangerous end of a roughly 80-year debt cycle, with $36.4 trillion of federal debt against $29.1 trillion of GDP. The cycle runs from sound money through a debt bubble, a top, deleveraging, and a restart. Debt that generates sufficient income is healthy; debt issued to service existing obligations becomes “plaque in the arteries.” A debt crisis can be addressed through austerity, restructuring, taxation, or central-bank monetization, the last repaying creditors in cheaper money.

The critical market warning is long-term yields rising while the central bank cuts short-term rates, signaling that private buyers no longer want the debt at prevailing prices. Gold and Bitcoin rising, sterling weakening as UK yields rise, and central banks and sovereign wealth funds reducing bond exposure all fit that pattern. Dalio’s U.S. long-term debt-risk gauge is at 100%—its highest historical reading, not a 100% crisis probability—while the near-term gauge remains 0%.

Dalio’s “3% solution” is to cut the expected federal deficit from roughly 7.5% of GDP to 3%, an adjustment of about $900 billion annually, while the economy is still strong. A credible fiscal reduction would itself lower market interest rates; a 100-basis-point decline would materially reduce interest expense. Delay makes the adjustment nonlinear: “The faster you cut, the less you have to cut.”

Investors must measure returns in purchasing power, because nominally rising assets can conceal severe real losses. Dalio says equity prices have sometimes fallen 60%-70% in inflation-adjusted terms, while Friedberg notes that 1966-1984 produced a negative real return. The book’s portfolio guidance emphasizes 10-15 genuinely uncorrelated positions. Dalio owns some Bitcoin but “not nearly as much as gold,” and favors stores of wealth that are international, mobile, relatively private, secure, and comparatively difficult to tax or confiscate.

AI can drive a major productivity transformation and is a strategic war “that no country can lose,” but that does not make today’s expensive technology leaders automatic winners. Dalio sees risk in Nvidia and the hyperscalers, with more opportunity potentially accruing to businesses implementing AI and creating applications. Dalio says China owns 33% of global manufactured goods, more than the United States, Germany, and Japan combined, and could pair inexpensive chips with robotics and manufactured products; however, “a great company that gets expensive is much worse than a bad company that’s really cheap.”

AI productivity is unlikely to arrive soon enough to solve the immediate debt imbalance and may initially add job losses and public-support demands before its gains arrive. Profits, capital gains, deregulation, tariffs, and efficiency might improve revenue, but Dalio rejects making the fiscal plan a “crapshoot.” The legislative window is short—the first 100 days, followed by roughly two years to the midterms—while the distribution of AI’s gains will be intensely political.

Debt stress is converging with internal polarization, U.S.-China rivalry, technological disruption, military spending, and climate costs. Dalio expects greater state-federal fragmentation and a world increasingly governed by “might is right,” though he does not predict a hot civil or military war as inevitable. Within the coming decade, he thinks there will be a “hellacious” period when problems intensify while the cooperation needed to solve them weakens.