Is the AI bubble about to burst?
Silicon Valley has seen manias before. From the dotcom frenzy to the crypto craze, technology's promise has often outpaced its profits.
Now, artificial intelligence sits at the centre of another investment whirlwind. The question is no longer whether AI is revolutionary, but whether its current valuations can survive a reality check.
Over the past couple of years, the world's largest technology companies have poured record sums into AI infrastructure. Microsoft, Amazon, Google, and Meta together spent over $350 billion on capital expenditure this year, with forecasts pointing to $400 billion next year.
The top CEOs wanted the investors to spend big with even bigger promises. Joe Fath of Eclipse VC put it, there is "a push and pull between those companies and investors," with many now asking if the returns will ever match the enthusiasm.
The anxiety is visible across markets. Shares in AI-linked firms such as Nvidia, Palantir, and AMD have stumbled. Nvidia, once the poster child of AI's rise, saw a multi-trillion-dollar valuation dip as investors wondered how much higher it could go.
Michael Burry, the investor who famously predicted the 2008 crash, recently bet against both Nvidia and Palantir, warning of overvaluation.
A recent MIT study found that 95% of companies investing in generative AI have yet to see measurable returns. Sam Altman, the chief executive of OpenAI, admitted that "many parts of AI are kind of bubble-y right now." Amazon founder Jeff Bezos has voiced similar concerns, saying that when excitement peaks, "every experiment gets funded," even when it should not.
But if AI is a bubble, it may not be one that bursts in a single dramatic pop. Economists suggest it could deflate slowly, squeezing weaker players while leaving giants standing taller. Pierre-Olivier Gourinchas, chief economist at the IMF, told Reuters that, unlike the dotcom bust, the AI boom is "not financed by debt," which makes a systemic crash less likely. Investors may lose, but the economy will not necessarily collapse.
Still, the scale of investment is hard to ignore. OpenAI is reported to be seeking $1.5 trillion for computing capacity, far exceeding its current revenue streams. Nvidia is not only supplying chips but also investing in AI companies that depend on its hardware.
Analysts have likened this web of "circular deals" to the 1990s internet bubble, where hype fed upon itself until it could no longer sustain the weight.
The World Economic Forum's president, Børge Brende, recently warned that AI could be one of three major bubbles threatening the global economy, alongside crypto and debt. The Bank of England issued a similar caution, saying that "the risk of a sharp market correction has increased."
Not everyone is convinced of doom. Some investors, like Eric Schiffer of the Los Angeles-based investment firm Patriarch Organisation, argued that "adoption may be low right now, but that's not a forward indicator."
He believes that as AI tools become a regular part of offices and industries, profits will increase. Products like Microsoft's Copilot and Google's AI assistants are slowly becoming integrated into daily work. They automate repetitive tasks and improve efficiency.
The story of AI resembles previous technological leaps: vast early investment, speculation, and uneven rewards.
During the dotcom era, companies like Cisco never regained their bubble-era peaks, even after decades of growth. While others, like Amazon and Apple, transformed into world-defining firms.
Economist Brent Goldfarb, who studies technological bubbles, argued that AI fits every pattern: uncertainty, hype, novice investors, and powerful narratives. He calls it an "eight out of eight" on the bubble scale. "Uncertainty has not gone down," he said, noting that no one yet knows how AI will be integrated profitably across industries.
Perhaps the AI boom is both a bubble and a breakthrough. The technology may last, but the values placed on it may not.
When everything calms down, the winners might not be the most vocal startups or the trillion-dollar giants. Instead, they could be the companies quietly creating the tools and systems that make AI genuinely useful. Until then, fear of missing out will likely influence the market.