The phrase “AI bubble” keeps floating around in tech and financial circles. Venture capital is pouring billions into AI startups. New companies appear almost daily with “AI” somewhere in their pitch deck. And skeptics are increasingly asking a familiar question: Are we repeating the dot-com bubble?
But that framing misses the bigger story.
To call the current moment a bubble suggests that the underlying technology lacks real value and that investors and companies are chasing something fundamentally hollow, hoping to exit before the hype collapses. Yet the reality of AI, particularly the rapid advancement towards general intelligence, suggests something very different. What we’re seeing is less of a speculative bubble and more of a classic technology hype cycle layered on top of a genuine platform shift. The distinction matters.
Calling today’s surge in AI investment a bubble also ignores a key difference in how the technology’s adoption. In a true bubble, once the speculation ends, the technology itself fades away. But AI is already demonstrating durable utility across everyday workflows, from writing and research to customer support and coding.
The question isn’t whether AI works. It clearly does. The real question is which companies will build lasting businesses around it.
Is the Current AI Boom a Speculative Bubble?
Although the term AI bubble has become common, the current surge is a genuine platform shift rather than a hollow speculation. Unlike the dot-com era, AI is launching into a mature digital ecosystem with immediate real-world utility in coding, research and customer support. The bubble narrative ignores that AI directly affects the production of work, creating a permanent structural change in global productivity.
Why AI Is Different From The Dot-Com Era
If we look to the past, in the late 1990s, the internet was still largely a novelty for businesses and consumers. Many people weren’t yet online and didn't even have computers in their homes, and the infrastructure required to build scalable internet companies from broadband access to cloud computing was still in its infancy.
Today’s AI ecosystem, by contrast, is launching into a digital world that already exists. The cloud, smartphones and global connectivity have created a mature ecosystem where AI tools can be deployed instantly to millions of users. That means adoption is immediate. When a new AI model is released, developers integrate it into software products within days and businesses can begin experimenting with it almost instantly. That kind of rapid, real-world integration simply didn’t exist during the dot-com era.
History offers a useful guide here. The dot-com era is often remembered for its spectacular crash, but that narrative overlooks what happened next. Yes, thousands of companies failed and trillions in market value evaporated. But the internet itself didn’t disappear. Instead, it became the foundation of the modern economy. Amazon, Google and countless other companies emerged from the wreckage to define the next two decades of technology.
Something similar is happening now with AI. The hype surrounding AI startups may eventually cool. Some companies will prove overvalued. Others will fail to build sustainable products. But the underlying technology will remain in use and likely become even more deeply embedded in everyday life and business.
AI Changes How Work Gets Done
The scope of AI’s potential impact may be larger than many people realize. Unlike previous waves of innovation that focused primarily on access to information, AI directly affects the production of work itself. Language models and AI agents can draft documents, analyze data, write software and automate tasks. In the right hands, a variety of knowledge work fields are experiencing massive productivity boosts.
In previous technology cycles, companies often spent years trying to figure out what a new platform was actually good for. With AI, the use cases are emerging in real time. The necessity of writing code for software is dissipating rapidly: I personally don’t write code anymore. Customer support teams are deploying AI agents to handle routine inquiries. Marketing teams are generating campaign materials in minutes and, with a human in the loop, they’re going to market in days.
These may seem like small productivity improvements individually, but across entire organizations they compound into significant efficiency gains. When a technology consistently saves time, reduces cost or replaces repetitive labor while also delivering quality output its functionality is not hype. It’s real infrastructure.
The internet democratized access to information. Artificial intelligence changes the cost and speed of productivity. With the right institutional knowledge and powerful enough hardware, almost any hard problem can now be solved. This level of utility makes it difficult to argue that AI as a category is speculative.
Most People Don’t Know The Power Of AI
The biggest misconception about AI isn’t that it’s overhyped. It’s that most people are still benchmarking it against what it was 18 months ago.
The gap between AI’s actual capabilities and the public’s perception of them is widening fast. Most casual users have poked at a chatbot, gotten a mediocre answer and filed the technology away as a neat parlor trick. What they haven’t seen is what happens when people improve their prompting skills and understand the technology beyond search or vibe coding. This is when AI goes from being a feature added on as an afterthought to a core part of the way you think, build and compete. The people who have crossed that threshold aren’t talking about productivity improvements in percentages. They’re talking about multiples that have led them to operate in a fundamentally different reality.
I don’t write code anymore. Not because I can’t, but because I don't have to. Instead, I get to focus more on planning, creativity and customer adoption. Founders I back are running completely agentic development flows, doing legal reviews, testing go-to-market strategies and more, all with AI. In the past, this amount of work would have required entire departments of people.
That’s not hype. That’s a new operational reality that most executives, policymakers and even tech founders haven’t personally experienced yet.
This matters when evaluating the theoretical bubble narrative. Much of the skepticism comes from people whose reference point for AI is surface level. They haven’t seen what it looks like when a small team wields it at full capacity. When they do, the conversation shifts, not from skepticism to blind enthusiasm, but from asking if the technology is real to how fast they need to move to take full advantage.
The ceiling of what’s possible is being raised faster than the floor of public understanding can keep up with it. That’s not a bubble dynamic. That’s a capability gap, and it’s one of the most underappreciated forces in technology right now.
Real Disruption Will Be Structural, Not Financial
Those focused on an AI-driven financial collapse are likely looking at the wrong risk vector. The private market concentration of AI investment means that even a significant valuation correction would be absorbed largely within venture portfolios rather than rippling through public markets the way the dot-com crash did. Crypto demonstrated this dynamic — a dramatic boom and bust cycle that, despite real losses for individual investors, never threatened the structural integrity of the broader economy. AI’s capital stack is similar in its insulation.
That said, AI is already distorting how public markets price software, even without a wave of AI companies going public. Everyone has a hot take on what the future will look like, and that speculation is seeping into how investors value the entire software category. Volatility has increased not because of earnings or cash flow analysis, but because public market investors are essentially betting on what other investors think the future will be. It’s reflexive, narrative-driven pricing at scale, and it’s being applied bluntly to a software landscape that is anything but uniform.
Not all SaaS companies face the same AI exposure, and the market’s inability to separate them is a significant source of that volatility. I would say there are essentially three categories of SaaS providers worth distinguishing.
- First, seat-based software directly tied to work output; think tools where you pay per person to produce something. AI hits this type hardest because you no longer need seats to produce work.
- Second, seat-based software not directly tied to work output, where the displacement pressure is real but less immediate.
- Third, annual subscription software that functions as a system of record, deeply embedded in enterprise infrastructure and workflows, which ultimately still could charge seats as well. These are far more defensible. The switching costs are high, the integrations run deep and no one is vibe coding their way out of a core operational platform at a large enterprise.
The real question investors should be asking isn’t whether AI will eat software. It’s whether a given company’s value is tied to human headcount to something that exists independent of it.
The deeper problem is how disconnected much of the AI bubble analysis is from what is actually happening on the ground, particularly here in Silicon Valley. The investors in financial markets setting price targets and writing macro theses about AI’s impact often haven’t sat across from a three-person team building what used to take fifty.
AI is unlocking economies of scale and levels of productivity that are unprecedented. We can no longer view small teams as scrappy underdogs just doing the bare minimum to get by in the market. Instead, small teams, when taking advantage of AI properly, are supercharging their work to rival and surpass the legacy companies that are less capital efficient and not building with AI as a central part of their product and go-to-market strategies.
The gap between the ground truth of AI adoption and the story being priced into markets is significant. That disconnect is where real volatility lives, not in a bubble collapse, but in a market that is continuously recalibrating against a future it can’t quite see clearly yet.
