The question is everywhere—whispered in boardrooms, debated on Twitter, and keeping investors up at night: Are we in an AI bubble?
If you work in tech, you’ve felt the tension. Record-breaking valuations sit alongside whispers of unsustainable burn rates. Revolutionary breakthroughs happen weekly, yet returns remain elusive for many companies. The contradictions are real, and they matter.
Here’s what the data shows, cut through expert perspectives from tech, AI, and finance.
The bubble evidence is hard to ignore
The numbers tell a concerning story. OpenAI commands a $500 billion valuation—double that of Salesforce, a company with decades of proven revenue. Meanwhile, startups like Mira Murati’s latest venture raised $2 billion before shipping a single product.
The economy looks shaky, too. Popular AI tools like Cursor sell below cost, burning through investor cash to capture market share. It’s a familiar playbook that historically ends with a reckoning.
Financial markets are already showing cracks. In days, NVIDIA dropped 3.5%, and Palantir fell 10%. When the so-called “Magnificent Seven” tech stocks wobble, the ripple effects extend far beyond Silicon Valley.
Financial analyst Craig McAskill says, “The pattern is familiar: hype, FOMO, cheap money… then collapse. The only difference is: this time everyone knows it.”
But this time, really might be different
The counterargument has substance. OpenAI’s revenue exploded from $2 billion to $12 billion annually. That’s not hype, money changing hands for real value.
The adoption story goes deeper than headline numbers suggest. While MIT studies claim “95% of firms see no ROI from AI,” they’re missing the shadow economy. Millions of employees quietly use ChatGPT, GitHub Copilot, and similar tools to work faster. This productivity boost doesn’t appear in corporate reports, but it’s reshaping how work gets done.
Then there’s the infrastructure angle. The $100+ billion flowing into data centres and semiconductors doesn’t evaporate when valuations crash. AI’s physical backbone will remain valuable, like the railroad tracks that outlasted the 1840s bubble or the fibre-optic cables that survived the dotcom bust.
As one AI researcher noted: “This isn’t tulips. Even if valuations fall, we’re left with the backbone of the next digital economy.”
The proximity illusion
Here’s the thing about living through revolutionary change: progress feels incremental when you’re inside it. Week by week, new AI releases seem like modest improvements. But step back and the trajectory is breathtaking.
In 2018, GPT-1 produced barely coherent gibberish. By 2025, GPT-5 demonstrates reasoning that borders on poetic. The leap is staggering, but daily exposure makes it feel routine. This proximity effect explains why many insiders feel like AI has plateaued—even as capabilities continue advancing exponentially.
What history teaches us
Bubbles follow predictable patterns. The railroad frenzy lasted four years before collapse. The dotcom boom ran five years from excitement to devastation.
AI entered the mainstream in November 2022 with ChatGPT’s launch. We’re roughly three years into the cycle. If history holds, we’re approaching the danger zone where bubbles typically peak.
The key signal to watch? Companies using stock instead of cash for acquisitions. That’s a classic late-cycle marker that suggests we’re running low on real money and high on paper wealth.
How to navigate what’s coming
Smart investors are already positioning for multiple scenarios. They’re backing infrastructure plays—data centers, chips, energy grids—that remain valuable regardless of valuation swings.
The Amazon strategy offers a blueprint: focus on solving genuine needs rather than chasing novelty. Amazon survived the dotcom crash because people actually wanted to buy books online. Adobe endured because designers needed better tools. The companies that last solve real problems, not theoretical ones.
For those with patience, crashes create opportunities. Google emerged from the 2002 wreckage to dominate search. Netflix traded at $3 per share in 2008 before revolutionising entertainment. The best technologies often become cheaply available after bubbles burst.
A tech investor shared this perspective: “The worst outcome is AI becomes essential, but lots of people lose money on the way there. That’s not the end of AI—it’s the end of bad bets.”
The uncomfortable truth
We can hold two realities simultaneously: AI might be in a financial bubble while simultaneously driving an irreversible technological revolution.
The internet didn’t disappear after the dotcom crash—it became more useful. Email, e-commerce, and search engines emerged stronger from the wreckage. Similarly, AI’s core capabilities won’t vanish if valuations correct.
The real question isn’t whether we’re in a bubble. It’s who will still matter when the dust settles.
Some companies are building sustainable businesses on genuine AI capabilities. Others are riding momentum with unsustainable economics. The market will eventually sort them out, as markets always do.
Smart money doesn’t ask if there’s a bubble. It asks who will survive what comes next.
