The Magnificent Collapse of the Stock and Crypto Bubbles

The third bubble recession of the century will not be fun

There seems little doubt that the economy is once again being driven by asset bubbles, mostly in the stock market, but likely also with the crypto craze. Prices of these assets have grown out of line with any plausible story of economic fundamentals, which means that whenever any substantial number of investors take notice, there will be a large plunge as happened with the tech led stock bubble in the late 1990s and the housing bubble in the 00s. The fallout is not likely to be pretty.

The Magnificent Seven and the Stock Bubble

Business reporting has been filled with stories about the magnificent seven and the extraordinary run-up in their stock prices over the last five years. The run-up has been impressive, and people who got in two or three years ago have done very well, as was the case with the big tech stocks in the late 1990s. However, if we project out even modest future growth in stock prices, we get a story that most people would likely find implausible.

The table below shows the market capitalization of the Magnificent Seven stocks as of October 9^th, 2025, in the first column. The second column projects 2030 market capitalization, in 2025 dollars, assuming a 7 percent real annual return. That is approximately the average real return in the stock market over the last seventy years. That seems a modest assumption on expected returns given that most investors likely assume that there is considerable risk associated with these stocks at their current valuations, but I want to error on the low side.

In that scenario, the market capitalization of the Magnificent Seven would increase from $23.1 trillion today to $32.4 trillion in 2030. Suppose at that point earnings of the Mag-7 have increased to the point that the collective price-to-earnings ratio is a still high 20 to 1. That would put their total earnings in 2030 at $1,618 billion, as shown in the fourth column. That is up from a total of $440 billion in 2024.

We can compare this projection for Mag-7 earnings with economywide earnings, as projected by the Congressional Budget Office (CBO). This is projected to be $4068 billion in 2030. CBO projects that taxes will be $450 billion in 2030, leaving $3,618 billion in after-tax profits in that year. That is up from $3,266 billion in 2024.

However, if we subtract out the $1,618 billion profit figure for the Mag-7 for 2030, calculated above, that leaves profits of $2,000 billion for all the other companies in the economy. Adjusting for inflation, that would be equal to $1,762 billion in 2024 dollars, a decline of 37.7 percent from actual earnings in 2024. That means that the real profits of the non-Mag-7 companies would fall by more than a third over the next five years if the Mag-7 stocks grow at a modest pace and earnings rise to a point that would plausibly justify the 2030 stock price, and the CBO profit projections prove correct.

Any of these assumptions can prove wrong, but it’s worth asking what it would mean in each case. Starting with the CBO profit projections, CBO is never exactly on the mark, but its projections are generally a good starting point. There certainly is no evidence of a systematic bias in its projections that would lead us to expect that it has either over or underestimated profit growth.

The assumption on stock price growth seems very modest. My guess is that most people holding large amounts of Mag-7 stock expect better than a 7.0 percent real return. It seems very unlikely many people would hold these stocks expecting a much lower rate of return.

As far as the assumed 20 to 1 PE for 2030, these are all huge companies. This is a high ratio for what would be well-established companies at that point. It seems implausible that large well-established companies could sustain a PE much above 20 to 1. That would imply that investors would still be anticipating extraordinary future profit growth from these companies. Rapid profit growth is plausible from small companies and start-ups, it becomes highly implausible in companies that already are taking in historically large shares of corporate profits.

This analysis implies that either the Mag-7 stock prices are seriously over-valued or that we will see a massive shift of profits from the rest of the economy to the Mag-7 companies along lines we have never seen before. The over-valuation of the Mag-7 seems the more likely scenario.

Patterns in Bursting Bubbles

Having closely followed both the tech bubble in the late 1990s and the housing bubble in the 00s, I feel comfortable in saying that their collapse is unpredictable. The tech bubble began to unwind in March of 2000. A quarter-century later it would still be difficult to point to a specific event that led to its collapse.

The housing bubble peaked in the summer of 2006, at which point prices began to drift downward. The decline picked up speed in 2007, and went even faster in 2008, as mortgage credit began to dry up. As was the case with the stock bubble, there is nothing obvious that caused the turn in 2006, although lending practices had been getting gradually worse to the point where they were widely ridiculed by people in the industry.

The rate at which the economy is affected is also difficult to determine. The full impact of the collapse in the tech sector took around two years. Real investment in computers fell by more than 15.0 from 2000 to 2001 and then another 7.5 percentage points of the 2000 level from 2001 to 2002. Investment in communications equipment fell by almost 12 percent from 2000 to 2001 and then another 20.5 percentage points from 2001 to 2002. At the time, these drops were equal to around 0.5 percent of GDP.

The decline in spending from the collapse of the housing bubble was far more dramatic. Residential investment peaked at 6.7 percent of GDP in the second half of 2005. By the fourth quarter of 2007 it was down to 4.2 percent of GDP, a loss in annual demand that would be equivalent to $750 billion a year in today’s economy. It continued to fall in 2008 and 2009, eventually bottoming out at just 2.4 percent of GDP in the second half of 2010.

The hit to these sectors is only part of the story of a collapsing bubble. In the case of both the stock and housing bubble, people were spending based on the stock wealth the bubbles generated. The saving rate declined from just under 7.0 percent of disposable income in 1995 to just over 4.0 percent at the peak of the bubble in 1999 and 2000.

There was an even more dramatic story with the housing bubble with the saving rate bottoming out at less than 2.0 percent in 2005. After the bubble burst, consumption fell back, with the saving rate rising above 6.0 percent by 2010. This lost consumption is the equivalent of $900 billion a year in annual spending in the current economy.

The current saving rate is close to 5.0 percent, which suggests some impact of the AI bubble on consumption, but nowhere near as large as with the housing bubble. It is reasonable to expect that the saving rate will rise by around 1.0 percentage point in response to a collapse of the bubble, costing the economy around $220 billion in annual demand.

A collapse of the AI bubble is likely to also coincide with a collapse of the crypto bubble. This is not necessarily because the bubbles are directly related, but it is because the same people are doing the speculation. This is a similar story to the stock and housing bubble that drove the Japanese economy in the 1980s. At its peak in 1989, the value of the Japanese stock market exceeded the value of the U.S. market, even though the U.S. economy was more than twice as large. Japanese real estate was similarly inflated. The simultaneous collapse of the two bubbles led to a decade of stagnation.

Can AI Actually Make Sense of the Current Market Valuations?

I am far from an expert on the prospects for AI technology, but there are a couple of points I feel comfortable making. First the impact of AI on the economy is distinct from the profits of the AI companies. It is very possible that we will see substantial productivity gains from AI, but the companies directly involved in providing it are not big beneficiaries.

That is actually the story with the Internet economy. The providers of Internet, companies like Verizon and Comcast, have made plenty of money from the service, but they are not among the very biggest companies in the country. Their size is certainly not at all proportional to the impact that the Internet has had on our lives and the economy. In the same way, it is very possible that AI will substantially increase productivity in many areas, but Alphabet and Nvidia and the rest end up not being big winners from it.

The second point is that we are not alone in pursuing AI. Most obviously, China has been rapidly building up its capacity in AI. Whether or not it is ahead, it is clearly very much in the running. It seems that its leading companies are pursuing a fundamentally different strategy. Rather than abstractly trying to develop sophisticated systems, they have focused on developing AI that has practical applications.

The Chinese AI systems are also far more energy efficient than the U.S. systems. This was a result of necessity. They were largely cut off from the most advanced U.S. chips, which meant that they had to design systems that worked with somewhat less computing power. This means that they are cheaper and also use far less electricity.

This is an especially important point in a context where the U.S. seems to be facing limits in its ability to quickly increase electricity generation capacity. In addition to having more energy efficient system, China has much more plentiful and lower priced electricity. This is likely to be a huge advantage for the Chinese AI systems in the years ahead.

This doesn’t mean China will necessarily be the big winner in the AI race. It’s entirely possible that the market will be big enough for systems from both countries, as well as others also working with the technology. But the point is that even if AI does prove to be hugely important for the economy, it is not at all guaranteed the U.S. companies will be the main beneficiaries.

Is Crypto Crazy?

I have been a crypto skeptic from the beginning. While I realize I could have gotten very rich by putting $1,000 into Bitcoin 15 years ago, my view has not changed. We know crypto is good for illegal transactions. It is much easier for a kidnapper to demand $10 million in bitcoin as ransom than in hundred-dollar bills, but no one has yet developed a serious use case for crypto. In fact, the large run-up in its price is a huge argument against its potential use as a substitute currency, which was a claim made by its proponents initially. We want a currency to be stable in value, not gyrating erratically.

The newer stable coins, which are pegged to the dollar, or some other currency, can offer some advantages in speedier transactions, but this could also be done by creating a digital currency. Brazil has taken the lead here, with a digital currency embraced by more than 80 percent of its population. Its digital currency, the PIX, allows Brazilians to costlessly pay bills, receive paychecks and other deposits, and make other transactions virtually instantly.

This is a case, like the Social Security or Medicare systems, where the government can provide a universal benefit at far lower cost than the private sector. To prevent the private stable coin providers, including one from the Trump family, from having to worry about such competition here, Congress has prohibited the Federal Reserve from even studying the creation of a digital currency. This effectively is requiring people to throw money in the garbage to increase profits for the financial industry.

There is also the fundamental question of how we ensure that stable coins will maintain their peg to the dollar. This is the age-old problem of financial regulation. Financial institutions make more profit if they can minimize the amount of reserves they have to hold. They also benefit by being able to put reserves into more risky assets that offer higher returns.

That is the recipe for instability that has led to thousands of bank collapses over the years. Government regulation is supposed to be designed to prevent excessive risk-taking and ensure stability. It’s not clear why anyone would be confident in the Trump administration’s ability to regulate its friends’ and family’s crypto coins. The fact that crypto czar David Sachs biggest previous claim to fame was demanding a bailout of Silicon Valley Bank in 2023, when the bank discovered that bond prices could fall, does not inspire confidence.

In short, stable coins are an accident waiting to happen. Given the rapacious greed of the leading operators and the Trump administration’s commitment to lax oversight, the only question is when. Perhaps a stable coin blowout will be the event that leads to the collapse of the crypto, eliminating trillions of dollars of illusory wealth, but maybe it will take something else to restore reality.

Bubble Recessions Are Not Fun

The history has been that bubble recessions are harder to recover from than the more traditional recession brought on by the Fed raising interest rates. The main remedy for the more traditional recession is the Fed reverses course and lowers interest rates. This boosts demand from the interest sensitive sectors of the economy, most importantly cars and housing, which have pent-up demand from the period of high rates.

That remedy is not available when a recession is caused by a collapsing bubble. In addition to the Great Depression, we have the experience of the 2001 recession caused by the collapse of the stock bubble and the 2008-09 recession caused by the collapse of the housing bubble.

Contrary to the conventional view, that the recession was short and mild, the 2001 recession was actually quite bad from the perspective of the labor market. We didn’t get back the jobs lost in the recession for four full years. At the time, that was the longest period without job growth since the Great Depression. The severity of the Great Recession is generally recognized.

We do know how to get out of a recession caused by a burst bubble: spend money. But given the craziness of the Trump administration and deficit fears among Democrats and the public more generally, a downturn created by the collapse of the AI and crypto bubbles could be long lasting. The one thing that we can be certain about is that it is better if it happens sooner than have the bubble continue to inflate and create more destruction when it bursts.