The equity market has rarely been this confident and divided. Major indices continue to hit record highs, mostly driven by a small group of AI-exposed megacaps.
Capital is pouring into artificial intelligence at a historic pace, valuations are expanding ahead of cash flows, and infrastructure spending is reaching levels previously reserved for national-scale projects.
To some, this looks like the early innings of a productivity revolution. To others, it has all the hallmarks of a classic bubble.
Even experts have polarized views. On one end of the spectrum, skeptics like Gary Marcus warn that expectations have raced far ahead of the technology’s actual capabilities, while valuation-focused voices such as Aswath Damodaran caution that parts of the current AI boom look a lot like the dot-com bubble.
On the other end, market strategists like Josh Brown and Tom Lee argue that investors are still underestimating the scale of the opportunity, comparing AI to past platform shifts whose true economic impact only became clear years later.
To skeptics, the current AI boom looks less like a breakthrough and more like a familiar speculative cycle, where expectations are outpacing execution and fundamentals.
For the bears, the capital behaviour surrounding the AI hype feels like a rerun of the dot-com era. Then, as now, a revolutionary technology collided with speculative enthusiasm, massive capital inflows, and expectations that didn’t match the practical timelines.
During the dot-com boom, the market was hyped by what the Internet era was promising. A small group of dominant companies captured most of the market gains, while hundreds of other smaller companies tried to ride the same wave. After brutal attrition, roughly 50–60% of venture-backed internet companies ultimately failed or disappeared within a few years of the crash.
Today’s AI landscape looks similarly crowded to the dot-com boom. An approximate number of 300 AI unicorns have emerged at unprecedented speed, supported by automation, open-source models, and cheap access to cloud infrastructure.
Bears argue that many AI companies are structurally redundant because they all rely on the same generative AI models (e.g., ChatGPT, Claude, DALL-E) and on the same infrastructure expansion promised by hyperscalers. The market becomes oversaturated, with many startups offering products that essentially do the same thing with a layer of better user experience on top.
Moreover, the solutions they build are not backed by strong research. Over 40% of AI startups are expected to fail by 2027 because they target pain points that are not real needs, or they fail to capture a critical mass of customers in a way that can sustain healthy growth.
And that’s not all. There are additional concerns about their monetization models. Many AI companies fail to attract the right audience, and consequently, they are attracting non-paying or non-returning customers. In these cases, current investments are keeping them alive, but when capital becomes more selective, a culling may be inevitable.
While AI has been a hot topic on a global scale for some time now, some corporations are not impressed. A recent study conducted by ISG found that while AI adoption doubled in 2025 compared to 2024, two-thirds of the projects are not yet in production, and only about one in four initiatives meets revenue impact expectations.
Similarly, a 2025 MIT report shows that 95% of enterprises investing in gen AI have produced zero returns.
Another concern centers around Baumol’s cost disease. This economic principle states that productivity in the service sector tends to fall behind, keeping costs actively high.
