If last week Trillion Dollar Slide The reason why the massive tech stocks looked familiar is because we've been here before – the last time the hype about innovation quickly penetrated economic reality.
Because the markets collapse in consequence Investor discomfort As artificial intelligence (AI) corporations' valuations rise, commentators are asking the identical query they did back then The dotcom crash 25 years ago.
Can technology really defy basic economics?
I discussed this query in my inaugural lecture on the University of Otago in August 2000, when web stocks collapsed and a whole bunch of dot-coms failed.
I argued on the time that many Internet corporations were “naked” because their business models were visible to everyone. They spent huge sums of cash to draw customers with no credible prospect of profit.
A generation later, the identical logic is driving the AI boom.
Different metrics, same story
In 2000, the Internet promised to revolutionize commerce, with success measured in “eyeballs” and “clicks” fairly than profits. Today, these indicators have turn out to be “processed tokens” and “model queries.”
The language could have modified, but the assumption that size robotically results in profit has not.
Just after we heard that the Internet would eliminate the middlemen and eliminate traditional middlemen like retailers and brokers, there have been guarantees that AI will eliminate cognitive work.
Both have encouraged investors to overlook losses within the pursuit of long-term dominance.
At the peak of the dot-com craze, corporations like Online retailer eToys spent lots of money on marketing to draw customers. Today, AI developers are investing billions in computing power, data and energy – and yet remain unprofitable.
Nvidia's multi-trillion-dollar valuation, OpenAI's continued losses despite rising revenues, and the flood of enterprise capital funding for AI startups are all harking back to the 1999 bubble.
Then, as now, spending is confused with investment.
What the dot-com crash must have taught us
In 2000, I suggested that Internet corporations construct market-based assets like brand equity, customer relationships, and data that would only create real value in the event that they produced loyal, profitable customers.
The problem was that investors viewed spending as evidence of growth and marketing as a business model in its own right.
The AI economy repeats this pattern.
Datasets, model architectures, and user ecosystems are treated as assets, even in the event that they don’t yet generate positive returns.
Their value comes from the assumption that monetization will eventually meet up with costs. The logic stays the identical; only the story has modified.
The dot-com boom was driven by fragile startups fueled by enterprise capital and public enthusiasm.
Today’s AI surge is led by powerful incumbents like Microsoft, Google, Amazon and Nvidia, which may suffer years of losses of their quest for dominance. This reduces systemic risk but concentrates market power.
Getty Images
Where the cash goes has also modified. Internet corporations used to spend money on promoting. AI corporations are looking forward to computing power and data.
Spending has shifted from the marketing agency to the information center, however the query stays: are they creating real value or simply the illusion of progress?
AI also goes deeper than the web. The web has modified the best way we communicate and shop, but AI is shaping the best way we expect, learn and make decisions.
Should a crash occur, it could undermine public trust within the technology itself and slow innovation for years. Relatively low real rates of interest and abundant capital have also helped fuel this current wave of technology investment.
Similar to the boom within the late Nineties, when favorable monetary policy helped fuel an increase in technology valuations, this cycle shows how the macrofinancial backdrop can reinforce technology optimism.
The return of the intangible mania
Despite these differences, the evaluation pattern is familiar. Investors once more rate potential more vital than performance.
In 2000, analysts based their valuations by counting the users that an organization might in the future monetize. In 2025, they model “inference need” and “data advantage.” Both are assumptions about an imaginary future.
Narratives have turn out to be capital as markets reward conviction over evidence. The danger just isn’t technical failure but economic distortion when storytelling trumps ability to pay.
Even profitable corporations can go into decline.
In 2000, leading corporations like Yahoo! and eBay lost most of their market value when the bubble burst, despite their long-term survival. The same thing could occur to today's AI giants.
There are two lessons left. First, scalability without profitability just isn’t a business model. Exponential growth can deepen losses fairly than reduce them.
Each additional AI query has an actual computational cost related to it, so growth only matters if it results in sustainable margins.
Second, intangible assets must create measurable value: marketing, data and algorithms are only assets in the event that they generate lasting money flow or clear social profit.
For policymakers, the implication is obvious: fund AI projects that deliver tangible productivity or social advantages, fairly than simply fueling the hype.
While AI will change the best way we work and think, it cannot remove the connection between cost, value and customer needs. Lasting value comes from delivering real advantages to people.
The query now is whether or not AI's actual productivity gains will ultimately justify today's valuations, because the Internet finally did after a painful correction.
