Editor's Note: The AI boom is transitioning from a technological narrative to a financial reality check.
Over the past year, the discussion around AI in the market has focused more on model capabilities, computing power gap, and application prospects. However, this article reminds us that what truly needs to be calculated is the capital return behind this boom. Mega-scale cloud service providers such as Microsoft, Alphabet, Amazon, Meta, and Oracle are investing hundreds of billions of dollars in AI data centers. According to current analyst expectations regarding revenue and capital expenditures, except for Amazon, the implied return on investment for most companies may be negative.
This means that the similarity between the AI bubble and the previous dot-com bubble is not just market sentiment, but a high degree of capital expenditure tied to macroeconomic growth and stock price expectations. The author points out that in the past four quarters, 93% of the U.S. GDP growth can be explained by tech investments; once cloud providers reduce investments in data centers, chips, and infrastructure, not only will companies in the supply chain such as Nvidia, TSMC, and ASML be impacted, but the U.S. economy itself may come under rapid pressure.
Of greater concern is that if AI companies like OpenAI and Anthropic go public at a market sentiment peak, it may not just be a fundraising event but a risk transfer: early-stage capital and existing shareholders will transfer the uncertainty in the AI narrative to retail investors, pension funds, and other investors willing to continue buying into the growth story.
The core issue of this article is not whether AI has a future, but who will foot the bill for this expensive infrastructure race once the marketing hype subsides.
Below is the original article:
In December 1996, then-Fed Chairman Alan Greenspan described the prosperity of technology, media, and telecom stocks at the time as showing signs of an 'irrational exuberance.' Nearly 30 years later, we can make a similar judgment about today's AI boom.
However, despite the similarities between the current tech boom and the previous dot-com bubble, there is an important difference: a part of today's boom is already much larger in scale than the TMT bubble of yesteryears. By 2025, U.S. corporate investment in IT equipment and software will be close to $15 trillion. At the peak of the TMT bubble, this number was $466 billion, which even after adjusting for inflation, is only $829 billion.
In fact, the current growth of the U.S. economy is almost entirely driven by the tech boom. Based on my calculations, 93% of the U.S. GDP growth in the past four quarters can be attributed to tech investment. Even at the peak of the TMT bubble, this proportion only barely reached 60%.
Major language model developers such as OpenAI and Anthropic are preparing for a blockbuster IPO later this year to capitalize on investor optimism about their growth prospects. At the same time, mega-cloud providers like Microsoft, Alphabet, Amazon, Meta, and Oracle plan to invest hundreds of billions of dollars in data centers over the next five years to provide computing power for running these models.
The crux of the matter lies here: the math behind the AI boom is starting to get tricky. For these mega-cloud providers, I have collected analysts' consensus expectations for their capital expenditure and revenue from 2025 to 2030.
During these five years, their capital investment is expected to grow at an annual rate of 20%, a pace unseen in this industry before. Meanwhile, revenue is forecasted to grow by 15% annually. If we make an extremely bold assumption – that these companies have no costs whatsoever – then the additional revenue can be seen as the profit they derive from their AI data center investments. However, even under this extremely optimistic assumption, the implied return on investment that I calculated is highly negative for all companies except Amazon.
These numbers indicate that if mega-cloud providers continue on their current trajectory, the AI boom will evolve into one of the largest shareholder value destruction events in history. However, they still have two ways out.
The first way out is for the rise of AI to bring in revenue for these companies far beyond current expectations. But this itself poses a mathematical challenge. Assuming these mega-cloud providers aim for a 10% return on investment, they would need to find an additional $20 trillion to $50 trillion in revenue sources annually. For a group of companies currently with a total annual revenue of only $1.5 trillion, this is nearly an impossible task.
The second way out is that the investments originally planned for data centers, chips, and other areas will ultimately not materialize. The reasons could be that stock investors are starting to become more cautious about the industry, or data center debt financing is becoming more difficult.
So, what would happen if these companies announced cuts to some of their investment plans?
From NVIDIA to ASML, Samsung, and TSMC, the stock prices of the largest companies on every continent are built on these investment plans and the resulting demand expectations.
Don't forget, the current growth of the U.S. GDP is actually entirely driven by continuously increasing tech spending. If this spending starts to decline, the U.S. economy will quickly fall into a recession. Even a slight decline in tech investment, for example, a 4% to 6% decrease, would have such an effect. Historically, similar magnitude investment declines have occurred after a modest tech boom in the 1960s and during the 2009 recession.
This mild correction in investment spending is likely to push the U.S., UK, and European stock markets into a new bear market. A reenactment of the early 21st-century tech stock crash is a real risk; back then, the market fell 50% or even more within the first year.
The next question is: when might we see announcements from these hyperscale cloud providers similar to scaled-back investment plans?
I believe such a scenario is unlikely to happen in 2026. Companies like OpenAI, Anthropic, etc., will still strive to maintain market hype at least until their respective IPOs are completed, which may continue to support this boom in the short term. But what happens after that? The "impossible math" faced by hyperscale cloud providers will not change, and the marketing hype may eventually fade. In the end, reality will set in.
It may not be in 2026, but it could happen in 2027 or 2028. After all, Greenspan talked about "irrational exuberance" in December 1996, and the bubble truly burst three years later in 2000.
From this perspective, the IPOs of these AI companies are likely just a massive investment risk transfer: risk moving from current owners to retail investors, pension funds, and other investors willing to pay for the narrative.
Welcome to join the official BlockBeats community:
Telegram Subscription Group: https://t.me/theblockbeats
Telegram Discussion Group: https://t.me/BlockBeats_App
Official Twitter Account: https://twitter.com/BlockBeatsAsia
