To understand the nature of the current crisis, Dalio suggests drawing a hard line between the concepts of wealth and money. Wealth is the capitalization of assets, often existing only on paper. A startup can have a $1 billion valuation after raising a funding round of just $50 million. These numbers get reported and increase the virtual wealth of founders and investors, but they cannot be spent directly.
Instead, money is the real means of payment, the liquidity needed to service debts, pay taxes, cover losses, or return funds to investors when positions are closed.
Deflating a bubble is the process of converting wealth into money. It is triggered when holders of paper assets urgently need real cash to meet their financial obligations. Especially when “wealth” does not generate “money” directly through profits.
What’s the problem?
The AI sector’s main problem by the summer of 2026 is the solid gap between capital expenditure and real revenue. This year, tech giants – Microsoft, Alphabet, Meta, Amazon – planned to spend about $650 billion on buying chips, servers and building data centers.
Venture funds like Sequoia Capital estimated a year ago that in order to recoup infrastructure investments of such a scale, the AI industry should generate over $1 trillion in revenue from end products. The reality has turned out to be an order of magnitude more modest. The corporate sector is extremely reluctant to integrate language models because of the risks of data leaks and algorithm hallucinations. Revenue from the sale of AI subscriptions and specialized software is in the tens, but far from the hundreds of billions of dollars. Companies have built digital infrastructure whose products have yet to find a mass solvent buyer.
Energy barrier
The problems began not only because of weak demand, but also because of physical constraints. Generating grids require increasing amounts of base-load power generation year on year. This is where geopolitics interfered with technological plans.
The blockage of the Strait of Hormuz in the spring of 2026 triggered a global surge in all types of energy. The cost of oil has consolidated above $100 a barrel, and natural gas prices have soared in Europe and Asia. Although the U.S. has its own natural gas base, the U.S. power grid was unprepared for the simultaneous rise in resource costs and the surge in consumption from new mega data centers. Operating costs for cooling and powering server racks have risen 30-40% over 2024 business plans, making many data centers unprofitable to operate.
In parallel, technology companies have faced resistance from the public, who are not happy about rising energy bills. Approval of new construction sites for data centers in the U.S. states of Virginia, Texas and Ohio is being blocked by local municipalities. The appetite of servers leads to a shortage of capacity in the local networks and the growth of tariffs for the population. In addition, cooling systems require millions of gallons of fresh water. Against the backdrop of climate change, the withdrawal of water resources provokes open protests by farmers and local residents. The industry is facing a “not in my backyard” effect that has physically halted the expansion of AI infrastructure.
The deflation of the bubble will not affect the survival of bigtech per se, but the valuation multiples of the stock price to earnings ratio. The market has built a premium into the stock prices of these companies for the exponential growth of AI. Removing this premium means a 30-40% drop in the capitalization of the technology sector. In absolute terms, compressing the Nasdaq 100 index by a third would evaporate $7-$8 trillion of virtual wealth. This would be a severe stock market crash, extremely painful for retail investors and pension funds, but, unlike 2008, it is unlikely to trigger a classic banking crisis and a series of defaults in the financial sector.
Cooling system
For the United States, the cooling of the AI boom poses the threat of a hard landing for the economy. Growth in the United States since 2024 has held on two pillars: the wealth effect from the stock market and massive capital investment by corporations in infrastructure.
If the stock market corrects by 30%, consumers feel poorer and sharply cut back on spending. Retail sales go into negative territory. At the same time, technology giants, having recorded losses from AI, go into austerity mode. Construction of new data centers will be frozen, and contracts for equipment, concrete, cables and cooling systems will be canceled. Programs for hiring engineers will also be curtailed.
Against the backdrop of Middle East inflation (expensive gasoline) and high Fed interest rates, stopping the investment cycle of technology companies deprives the U.S. economy of its last growth driver. Stagflation ensues: the economy stalls due to a lack of investment, while prices remain high due to disrupted global energy logistics. In the extreme case, the economy could fall into a long – up to a year – recession, especially adding a sharp decline in migration to the equation (which means not only a reduction in the labor force, but also in consumption, the collapse of the part of the economy that is focused on serving millions of migrants). But even a drop in growth rates to 0-0.5% would be extremely painful.
Further everywhere
The global economy will feel the blow through microelectronics supply chains. A drop in orders for new gas pedals from Nvidia and AMD will instantly hit TSMC’s production plans in Taiwan. But the main victims will be memory chip makers.
South Korea (Samsung, Hynix), whose economy has already suffered from the energy crisis, will face a whiplash effect. Giant stocks of HBM memory, produced under the expected data center boom, will be unclaimed. Chip prices will plummet, export earnings of Asian economies will plummet, triggering a recession in the Asia-Pacific region. From there, the crisis will go down the chain to every country in the world. With inflation still high, the arsenal of central banks will be limited, threatening to further exacerbate the crisis. Finally, both geopolitical and domestic political situation in many countries is very unstable, which may prompt politicians and regulators to take erroneous steps. In general, the prospect of a new global recession looks, although not a “base” scenario, quite probable.
Ray Dalio’s predictions indicate the end of the stage of irrational optimism about the AI sector. The technology will certainly remain with us, but its financial prospects need to be reassessed. Artificial intelligence will cease to be a magic word that attracts free trillions for a healthy living. After the shake-up, neural networks and the industries that serve them will become a “mature” sector where every dollar invested must have a clear return-on-investment rationale. Adaptation of the global economy to this reality will take the next few years, at the same time flushing out the excess capital accumulated in recent years.
Prepared by Bloomberg TV
