The Coming AI Bubble: Lessons from the Dot-Com Crash
written by Belin Korukoglu ❂ 10 min read time
It is the year 2000. A young stockbroker can hear the champagne bottle popping in the office. It is a sunny March day. The sunbeams are reflected on his CRT screen. Still, he can conspicuously see the number. His portfolio shows gains of 300% in a year. All the little dot-com companies he invested in are now reaching up to billion-dollar valuations. Some of these don’t even have a product.
He can hear his colleagues in the background: “The internet will change everything.”
Fast forward two years later. The bubble finally burst. NASDAQ lost almost 80% of its value, bottoming out around 1,100 points.
The young stockbroker’s screen is all red. The companies he invested in? They are collapsing left and right taking billions of dollars with them as they evaporate. Only few survive which are now known in the whole world as Amazon, Google, eBay…
Are we going to see another bubble this time with AI?
Investors are fed with high expectations such as increased productivity, exponential ROIs, and the promise that every dollar poured into AI today will return tenfold tomorrow. Pro and counter arguments exist but first we need to define what an economic bubble is.
What Is a Bubble and How Does It Burst?
“An economic bubble is the commonly used term for an economic cycle that is characterized by a rapid expansion followed by a contraction, oftentimes in a dramatic fashion.” (Girdzijauskas et al., 2009)
A bubble bursts when the prices that have been overvalued collapse sharply back toward the fundamental values. Economic bubbles are mainly caused by speculative excess, cheap money, exhausted growth resources, and herd psychology. This results in economic crashes, bankruptcies of companies, economic crises, and a reallocation of resources with only a few survivors left standing.
Let’s go back to the famous Dot-Com bubble to see how it was formed.
How the Dot-Com Bubble Formed and Collapsed
Between the years 1990 and mid-2000, U.S. stock prices increased almost fivefold, with yearly growth jumping from around 10% in the early ’90s to more than 20% by the end of the decade. This might seem like a small number compared to today’s tech valuations and crypto/AI stocks, which can swing double digits within weeks or even days but in the 1990s, this was considered an extraordinary growth. It acted as a fuel to a wave of investor euphoria and convinced many that the internet had unlocked a “new economy” where old financial rules were out the door.
During that decade, market value grew even faster, first doubling in the early years and then tripling by the late 1990s before the peak. Other countries saw strong gains too, though not as dramatic. Then, the bubble finally burst. By 2003 U.S. global markets had lost about a third of their value. Looking back, many argued that the surge was driven less by real productivity and more by speculation and inflated hopes pinned on brand value, corporate organization, and IT infrastructure. This will be an important point in comparing the AI boom to the Dot-Com bubble.
Now, the question remains: Is AI gonna create the same bubble with even bigger valuations and consequences when the hype collides with reality?
The Rise of the AI Boom
The current AI boom really took off after the release of OpenAI’s ChatGPT in 2022, building on the so-called revolutionary transformer model introduced a few years earlier. The new wave was powered by high-performance GPUs and deep neural networks. Nvidia quickly became the Cisco and Intel of this era, with revenues and valuations rising as demand for computing power exploded. The AI was like a cookie monster: devouring energy, GPUs, and data at an ever-growing pace. Fueled by the promise of generative AI and large language models, the boom has drawn massive investment and global attention.
Warning Signs: Overvaluation, FOMO, and Bottlenecks
The challenge with today’s AI-driven market is that the top 10 companies in the S&P 500 appear even more overvalued than they were during the 1990s boom. Of course, higher valuations today are also partly natural. The economy is bigger, technology has advanced, and global capital is more abundant. The concern is not that valuations are high, but whether they are sustainable. If the current AI boom is a bubble and it bursts, it would result in an even greater massive scale collapse than the dot-com crash, wiping out trillions in market value. Furthermore, the current reports are troubling.
MIT’s GenAI Divide report shows a striking picture: only around 5% of enterprise AI projects manage to drive rapid revenue growth, while the other 95% stall out. Not because the models don’t work, but because companies can’t integrate them properly. Despite the hype and billions being spent, most big firms are struggling to turn generative AI into real financial impact. Startups tell us a different story. By locking onto a single pain point and building smart partnerships, some scale from zero to millions in a year. Large enterprises, on the other hand, often pour money into flashy sales and marketing tools, while ignoring the less glamorous but far more profitable things such as back-office automation.
Another pattern emerges from the dust: companies are almost twice as likely to succeed when they buy AI tools from specialized vendors instead of trying to build their own. This is especially true in complex, highly regulated sectors like finance and healthcare. Meanwhile, AI is already reshaping the workplace. Support and admin jobs are slowly disappearing, not through mass layoffs but because empty roles simply aren’t refilled. And employees everywhere are sneaking in tools like ChatGPT on their own, creating both huge demand and big risks for governance. At the frontier, the most advanced firms are already experimenting with “agentic AI”. Systems that can learn, adapt, and act on their own within set boundaries... It’s a glimpse of the next phase of enterprise AI adoption.
Tools like ChatGPT have already changed the way we work, but even OpenAI’s own CEO, Sam Altman, admits that AI is in a bubble. Some, like Joe Tsai, Ray Dalio, and Torsten Slok, warn that valuations may have already surpassed the levels of the 1990s. Others push back, arguing that the fundamentals in semiconductors and supply chains are strong enough to support continued growth. Yet the warning signs are there: bold claims from rivals like China’s DeepSeek, OpenAI’s rocky GPT-5 release, and the fact that OpenAI is still unprofitable despite hitting $20 billion in annual revenue. Even Altman has started to temper expectations, saying that the once-buzzing term “AGI” is losing its relevance.
All of these feels like the dot-com era all over again: hundreds of unicorns valued in the billions, many with little revenue. Investors are betting on future growth more than current profits, just as they did in the late ’90s. The “Magnificent Seven” tech giants drive most market gains, and their stretched valuations echo the bubble years.
Another factor to consider is the infrastructure bottleneck. Unlike the dot-com boom, today’s AI surge comes with massive hardware and energy costs. GPUs are scarce, electricity prices keep climbing, and scaling these models demands enormous data-center buildouts. Training costs are rising almost 2.5× every year, and McKinsey projects that by 2030 trillions will be spent on new data centers. In other words, AI is less a lean software play and more a capital-heavy infrastructure race.
Despite the soaring costs, no company wants to be left behind, so “AI-powered” has become the new buzzword. Over 40% of S&P 500 firms now drop AI into their earnings calls, often with little more than branding behind it. Startups with barely any revenue are landing billion-dollar buyouts, just like the dot-com days when adding “.com” to a name was enough to raise cash. The race to signal AI adoption is now driving valuations as much as real performance.
Beyond the Hype: What Happens If the Bubble Bursts?
The parallels are clear: sky-high valuations, shaky profits, infrastructure build-outs, and FOMO-driven branding. But unlike the dot-com years, today’s AI does have real deployments and growing demand. The risk is still a sharp correction, but with stronger fundamentals, the crash may be less catastrophic. If the AI bubble bursts, it may not look like the dot-com crash where thousands of companies disappeared overnight. More likely, we’d see consolidation: weaker startups fail, while a handful of giants absorb talent and IP. Valuations could reset sharply if revenue doesn’t keep pace with expectations, and public markets may punish firms that oversold their AI story.
The late ’90s taught investors and founders that hype can only carry you so far. Many internet firms back then burned cash without clear models, and AI risks repeating this mistake. The lesson: prioritize sustainable revenue, clear business use cases, and realistic timelines over buzzwords.
Despite the bubble talk, AI’s long-term role looks far stronger than many dot-com bets ever were. Even if valuations cool, AI will likely remain embedded in enterprise workflows, healthcare, finance, and consumer tech. As with the internet, adoption will outlast the hype cycle. The real winners may not be the flashy unicorns but the firms that quietly turn AI into productivity gains at scale.
REFERENCES:
Girdzijauskas et al. (2009) — “Formation of Economic Bubbles: Causes and Possible Preventions”,Technological and Economic Development of Economy
National Bureau of Economic Research — “Rational Bubbles: Theoretical Foundations” (PDF)
Apollo Academy — “The AI Bubble Today Is Bigger Than the IT Bubble in the 1990s”
Fortune — “MIT Report: 95% of Generative AI Pilots at Companies Are Failing”
CNBC — “OpenAI’s Sam Altman Warns the AI Market Is in a Bubble”
CKGSB Knowledge — “From Dot-Com to DeepSeek: A 25-Year Tech Bubble Comparison for the AI Era”
Forbes — “Why AI Stocks Are Giving Some Investors Dot-Com Déjà Vu”
McKinsey & Company — “The Cost of Compute: A $7 Trillion Race to Scale Data Centers”
Reuters Breakingviews — “AI Boom Is Infrastructure Masquerading as Software”
HP Workforce Experience — “AI FOMO: How Fear of Missing Out Shapes Workplace Adoption”
Bloomberg — “Three Big Differences Between the AI and Dot-Com Bubbles”
Barron’s — “The Dot-Com Bubble Burst: Is AI Next?”
Reuters — “Is Today’s AI Boom Bigger Than the Dot-Com Bubble?”
Business Insider — “Is AI the Next Internet Bubble? Analysts Warn of Tech Stock Parallels”
Air Street Press — “AI Isn’t the Dot-Com Bubble”
Forbes — “Balancing Artificial Intelligence Goals with Shareholder Expectations”
