AI boom: is the next financial bubble coming?

Artificial intelligence (AI) has become the new gold rush as we enter the fifth industrial revolution. Investors worldwide are betting that AI will redefine every industry, from healthcare and finance to logistics and entertainment. Since late 2022, the combined market value of major AI-related tech companies has more than doubled, pushing markets to historical highs. But as expectations grew faster than profits, some economists warned that this boom was beginning to resemble speculation more than innovation, concerns that markets started to price during late 2025. Could the excitement around AI be setting the stage for the next financial correction?

Since early 2023, breakthroughs in AI, especially generative models like ChatGPT, Copilot, and Gemini, have reshaped investor expectations, turning a once-niche research field into a core driver of future profits and industry transformation.

Major tech players are doubling down on AI globally. Nvidia, once best known for graphics chips, has become the backbone of AI infrastructure, with data center demand soaring and a valuation reaching $4.55 trillion as of January 2026. Microsoft’s investment in AI and Azure cloud has pushed its market cap to around $3.5 trillion after peaking above $4 trillion in early 2025, while Amazon is developing AI infrastructure and custom chips via AWS. Even outside the U.S., Chinese and European companies are accelerating AI deployment, signaling a truly global AI race.

Investors seem to believe AI will touch every sector. In 2025 Q3, AI companies accounted for around 46% of global venture funding, and 29% went exclusively to Anthropic. For many, the promise of transformative AI justifies paying premium valuations, even before profits are realized.

A financial bubble occurs when an asset’s price climbs far beyond its real economic value, driven by hype rather than fundamentals. Investors buy not because something is worth more, but because they expect someone else to pay more later. History provides examples: the dot-com bubble in the late 1990s, the housing bubble in the mid-2000s, and more recently, surges in cryptocurrencies and electric vehicle stocks. 

Major institutions have raised red flags. The IMF and the Bank of England warn that AI-focused tech firms are “heading toward levels seen during the internet boom over 25 years ago”. Some AI-related companies are priced as if extraordinary growth is already guaranteed. For example, Palantir trades at roughly 393 times earnings and 111 times sales, still far above tech-sector averages, though down significantly from its 2025 peak. (P/E 180, P/S 11).

Private players like OpenAI are valued at hundreds of billions despite uncertain profits. Even OpenAI CEO Sam Altman has described current valuations as “insane.” Investors are buying “pure-play” AI stocks even when revenue is limited, and companies often highlight AI more in branding than in actual performance.

Some market participants now argue that the long-anticipated AI bubble has already partially burst. Jan van Eck, CEO of asset manager VanEck, has described the sharp sell-offs of late 2025, particularly in leveraged AI infrastructure plays, speculative compute providers, and related crypto assets, as a “healthy correction” rather than a collapse. In his words, the market “took out its own trash,” removing excess speculation while leaving structurally strong AI leaders largely intact.

Yet some analysts argue the market may still be rational in certain areas. AI-driven healthcare solutions or energy optimization technologies show real revenue potential. The debate is not whether AI will matter, but how much hype is priced today.

As AI valuations began correcting in late 2025, outcomes started to resemble a mild-to-moderate adjustment rather than a systemic collapse. In a soft landing, investors adjust expectations, stock prices fall gradually, and companies continue AI development with minor disruptions. 

A hard landing, by contrast, would involve abrupt valuation resets, heightened volatility, and a broader loss of confidence across technology and financial markets. Such a shock could spill over into pension funds, mutual funds, and household wealth, while also putting pressure on tech employment and innovation funding. Even in that scenario, the underlying economic value of AI would likely support a longer-term recovery, though the transition could prove both disruptive and uneven.

At the consumer level, the impact of an AI investment slowdown would likely be less dramatic but more diffuse. Rather than an immediate disappearance of services, consumers would likely see slower deployment cycles, higher prices for premium AI tools, and fewer experimental features reaching the market. In sectors such as healthcare or logistics, this would translate not into technological reversal, but into delayed productivity gains and postponed efficiency improvements. 

In that sense, the AI boom does not resemble a classic speculative bubble destined to burst uniformly across the market. Rather, it looks like a period of extreme over-enthusiasm concentrated in specific segments, followed by a necessary phase of selection. Parts of the market have already been corrected, not because AI failed, but because expectations temporarily outran feasible business models.

The real test is therefore no longer valuation, but execution. The next phase of the AI cycle will not be driven by hype, but by operating margins, scalable revenues, and defensible business models. The central question is not whether AI was overhyped, but which firms can convert extraordinary technological potential into sustainable profits, and which will discover that technological relevance does not automatically translate into financial success.