Does the 2000 Dot-Com Bubble Foreshadow the 2026 AI Bubble?

The Analysis: Does 2026 AI Look Like the 2000 Dot-Coms?

In March 2000, the Nasdaq peaked at 5,049 points. Companies with no profit were worth billions. Pets.com (online pet food delivery) raised $82 million in its IPO with a sock puppet mascot — and vanished nine months later. Webvan (home grocery delivery) was spending $30 million per week before collapsing.

Then came the crash. In two and a half years, the Nasdaq lost 78%. Over $5 trillion in market capitalisation evaporated. Tens of thousands of jobs disappeared overnight.

In 2026, artificial intelligence is generating the same frenzied enthusiasm. Companies that have never turned a profit are valued at tens of billions. Words like "revolution", "disruption", and "paradigm shift" fill every press release. Markets are massively overweight in the tech sector.

The legitimate question: is history repeating itself?

The Historical Fact: When Amazon Lost 97%

There is one story from the dot-com era that still shocks people when they hear it for the first time.

Amazon — today one of the world's largest companies — lost 97% of its stock market value between 1999 and 2001. The share price fell from $300 to $7. Respected analysts predicted imminent bankruptcy. Jeff Bezos himself watched friends and family ask him whether he would "survive this."

Amazon survived. From that $7 floor, the stock has since multiplied many times over.

But for every Amazon, there were ten Pets.coms. For every Google (founded in 1998, profitable by 2001), there were hundreds of start-ups burning cash at full speed chasing the mirage of the "attention economy." Cisco, the networking giant of 2000, has never recovered its all-time high of $80.

The Anecdote: "Eyeballs" and Conflicts of Interest

In the late 1990s, a new metric had invaded valuations: "eyeballs." Instead of looking at revenues, margins, or cash flows, analysts counted how many users visited a website.

The more eyes on your page, the more you were worth. It didn't matter if those eyes were spending nothing.

Goldman Sachs, Merrill Lynch, and other major banks produced research notes with price targets based on 10-year growth projections… for companies that had never had a profitable quarter. IPO fees flowed freely. Conflicts of interest were massive: banks were recommending the purchase of shares they had themselves brought to market.

The SEC later investigated several of these practices.

The lesson: when a new metric replaces fundamentals, beware. In 2000 it was "eyeballs." In 2026, it might be "compute hours" or "API call volume" — metrics that look like growth without necessarily translating into profitability.

The Concept: The Hype Cycle and the Natural Selection of Technologies

Every major technology has followed the same three-act arc, documented by consulting firm Gartner as the Hype Cycle:

• The Peak of Inflated Expectations: everyone believes the technology will immediately revolutionise everything. Valuations spiral. Investors fear missing the opportunity of a lifetime.

• The Trough of Disillusionment: the technology doesn't deliver on all its short-term promises. Bankruptcies pile up. The media shifts from "revolution" to "crash." Valuations collapse even for solid players.

• The Plateau of Productivity: survivors have built real moats (network effects, data, integration). The technology delivers on its promises — but on a 10-to-15-year horizon, not 18 months.

The internet followed this cycle exactly. Generative AI almost certainly will too.

The difference between 2000 and 2026: AI is already generating real revenues. Microsoft, Alphabet, and Nvidia are posting actual profits. It is not the same bubble as 2000. But bubbles can coexist with real technologies. The underlying gold can be genuine while the gold rush remains speculative mania.

The Axone lesson: the real question is not "is AI a bubble?" but "at what valuation is it reasonable to hold exposure?" Amazon survived — but its 1999 shareholders waited 10 years to recover their initial stake. The right asset at the wrong price is still the wrong investment.

At Axone, the Macro · Technique · Mindset method always comes back to separating the quality of a technology from the quality of its price. One can be excellent, the other disastrous. They are not the same analysis — and that is precisely why reading cycles, not just charts, changes everything.