What Would Happen to Bitcoin if the AI Bubble Bursts?

• Henrik Zeberg warns about overinvestment in AI infrastructure.
• AI companies are spending more on infrastructure than they generate in revenue.

When a child takes a piece of wire shaped like a hoop and plays to make soap bubbles, the fun lies in seeing how big they can grow before they burst. In financial markets, the opposite occurs: the more they inflate, the bigger the burst can be, along with the economic losses.

Now, a bubble does not necessarily mean that the technology behind it is useless. It can form when expectations about its future and the money invested grow much faster than the revenues capable of justifying that spending. As long as capital continues to flow in, that gap can be sustained. The problem arises when returns disappoint or financing becomes more difficult: sales begin, valuations fall, and the most fragile projects are exposed.

Henrik Zeberg, a financial market analyst, believes that this dynamic is already taking shape around artificial intelligence (AI). In an analysis published on October 1, the economist compares the current cycle with the great infrastructure booms of railroads, electricity, and telecommunications.

His starting point is important: he does not question the usefulness of AI. On the contrary. For him, "the technology is right and productivity is real."
Henrik Zeberg argues that a crisis of gigantic proportions is brewing. Source: The Julia La Roche Show.

What he questions is the price that markets are paying today to build a future whose returns could take much longer to arrive. According to his analysis, investment in AI infrastructure between 2025 and 2032 would reach about $10.3 trillion, averaging 3.63% of the United States' gross domestic product per year. Zeberg argues that no private investment cycle of that size has ended without a significant financial correction.

And this is where Bitcoin (BTC) enters the equation. Although Zeberg does not mention it in his thesis, it is a fact that a collapse of AI-linked companies would spread to the stock market, credit, and liquidity. In that scenario, the currency created by Satoshi Nakamoto could face a shock originating outside its own ecosystem.

There is no relationship that determines that a fall in tech stocks necessarily drags BTC down. In fact, Glassnode showed that for much of the last 30 days, Bitcoin outperformed the S&P 500. However, that advantage weakened at the end of September: between the 22nd and the 29th, BTC fell 2.9%, compared to 1.2% for the U.S. index.

The question is, then, what would happen if the correction were much deeper than a bad week in the stock market.

AI Needs More and More Money

Zeberg's thesis is based on the gap between the infrastructure being built and the revenues needed to make it profitable.

Microsoft, Alphabet, Amazon, Meta, and Oracle went from investing a combined $97 billion in 2020 to over $400 billion in 2025. By 2026, projected spending exceeds $800 billion. UBS also estimates around $4.1 trillion in investment from major cloud providers between 2026 and 2028.

Zeberg himself summarizes the problem this way: "Capital spending does not wait for revenues. It anticipates them."

The data helps to understand the statement. Amazon, Alphabet, and Microsoft would allocate in 2026 a capital expenditure amount equivalent to approximately 102% of their revenues from cloud services. At the same time, the investment of the five major providers would exceed for the first time their combined operating cash flow.

FactSet reaches a similar conclusion. The firm projects over $690 billion in infrastructure investment for their respective fiscal years of 2026 and nearly $800 billion if the calendar year is considered. As a result, free cash flow could approach zero or enter negative territory for all except Alphabet and Microsoft.

The firm explains the mismatch by pointing out that "AI costs are paid upfront, while returns are expected over a longer term."

The problem does not end there. Expansion increasingly depends on external financing. According to FactSet, new debt represented only 9% of infrastructure investment in 2024. By mid-2026, that proportion had climbed to 32%.
The joint investment in infrastructure by Alphabet, Amazon, Microsoft, Meta, and Oracle could rise from $379 billion in 2025 to $915 billion in 2028. Source: FactSet.

Zeberg observes in this dynamic one of the classic traits of bubbles: first, a functioning technology appears, then expectations arrive, followed by money, and finally the construction of capacity above actual demand.

His conclusion leaves little room for ambiguity: "This is a bubble."

Revenues Still Need to Catch Up to Spending

Another way to observe the imbalance is to calculate how much money AI should generate to justify all that infrastructure.

Bain & Company estimates that the industry would need to generate around $6 trillion in annual revenues by 2031. The currently known business and consumer applications could contribute between $1.2 and $1.8 trillion. The rest would have to come from activities that do not yet exist or remain in early stages.

David Crawford, technology head at Bain, warns that the infrastructure "is being built well ahead of the demand curve."

That does not mean that demand does not exist. Revenues from companies like OpenAI and Anthropic have been growing at extraordinary rates. Zeberg's point is that even that growth is still lagging behind the capital already committed.

According to his calculations, the difference between the revenues generated by the leading labs and those necessary to compensate the invested capital would rise from $165 billion in 2024 to about $568 billion in 2026. Even assuming that the revenues of the leading labs continue to double each year, the gap could approach $1.2 trillion by 2028.

How Would This Problem Affect Bitcoin?

If investment in AI began to correct itself, the first impact would likely appear in technology companies and their sources of financing. A drop in stocks could lead to reduced risk exposure, sales aimed at covering losses in other positions, and increased demand for cash.

Bitcoin would be exposed here. Firstly, because the digital asset is increasingly integrated into the traditional financial system. It is enough to mention that there are currently exchange-traded funds (ETFs), greater institutional participation, and public companies and even governments (United States, El Salvador, and Bhutan) that hold bitcoin on their balance sheets. Strategy is the company with the most BTC in its treasury. Source: Bitcoin Treasuries.

If a widespread exit from risk assets were to affect BTC, a second front could open: corporate treasuries.

Bitcoin already went through a test of this kind during 2026. From the peak of $126,000 in October 2025 to the $63,408 recorded in July, the price fell nearly 50%. During that period, many public companies with bitcoin treasury strategies traded below their net asset value, as reported by CriptoNoticias. Bitcoin price over the last 12 months. Source: TradingView.

Companies Could Become Another Pressure Channel

Strategy is the most visible case. At the end of the second quarter, it held about 846,000 BTC and acknowledged in its regulatory documents that its bitcoin are "less liquid than our cash and cash equivalents."

Between June 29 and July 5, Strategy sold 3,588 BTC for about $216 million. The company itself explained to the U.S. Securities and Exchange Commission (SEC) that the money was used to meet its financial obligations, such as paying dividends on its preferred shares and replenishing its dollar reserves.

This does not imply that any drop in BTC will lead to insolvencies. The financial structures of these companies are different, and many have sufficient reserves to withstand prolonged periods of volatility.

The risk appears when three elements coincide: bitcoin drop, difficulties in obtaining new financing, and obligations that require cash.

Zeberg finds a historical parallel in previous investment cycles. When leverage (the use of debt to finance investments and increase exposure) stops driving growth and starts to operate in the opposite direction, he summarizes the process in three words: "Selling begets selling."

The mechanism applied to bitcoin would be simple. A drop in traditional markets could pressure its price and reduce the value of corporate treasuries. If any company needed to sell part of its reserves to obtain liquidity, it would add new supply to a market that would already be under pressure.

In that scenario, bitcoin could go from receiving an external shock to contributing to amplify it. For its price, this would imply additional downward pressure: corporate sales could deepen the initial drop, accelerate liquidations, and increase volatility.

A company with the scale and financial margin of Strategy would, in principle, have more tools to withstand such a shock. The question is what would happen with smaller companies, with less access to financing or whose strategy depends more on the value of their bitcoin.

For them, a deep drop could quickly turn an accounting loss into a liquidity problem. If they needed cash when bitcoin is trading below its average purchase price, they would have to part with some of their reserves and crystallize losses. The more companies that find themselves in that situation at the same time, the greater the likelihood that those sales would add pressure on the price.

This remains a hypothesis, not a prediction. The AI bubble that Zeberg suggests may not burst in the terms he anticipates, and a technological correction does not guarantee a proportional drop in bitcoin.

What has changed compared to previous cycles is the number of connections. The expansion of AI increasingly depends on credit; bitcoin is becoming more integrated with traditional financial products; and public companies hold hundreds of thousands of BTC on their balance sheets. Tags: Bitcoin (BTC) Artificial Intelligence (AI) Latest Prices and Trading