Economic Bubbles: Lessons From The Dot-Com Era

Day 3 of 3 for what the Dot Com Bubble can teach us about the AI one. There's a joke or a rule of thumb - depending on who you ask - in commercial real estate that says a new building needs to go bankrupt twice before it turns a profit. The person who builds it loses money, the next guy does, then finally the third person to pick it up for a fraction of the cost can actually get a return out of it. Something similar happened in the Dot Com era. All the money going into infrastructure built out fiber optic networks and Internet backbone well in excess of what anyone could figure out a business case for. One interesting story was 360networks. They had a method for laying cable from rail cars along rail right of way. They hit a $13B market cap on $234M revenue and filed for bankruptcy 7 months later. Companies like that built the modern Internet infrastructure then sold it for pennies on the dollar. This stuff got so cheap that I used to watch Boise State football games when I was in college in the early 2000's. A Boise company streamed the video from the Jumbotron along with the radio announcers... for free. It was insanity. The explosion of e-commerce, video streaming, and telecom from 2001-2011 was built on the carcass of the Dot Com Boom. People keep ranting about how there can never be enough revenue from AI to pay for the tens of billions of dollars being spent on Nvidia chips and data centers. I just laugh. First, you never know what people are going to be able to charge for. I attached a figure from a recent WSJ article showing rocket launches in Florida. (Note that this graphic is terrible because the 2024 line is only half the year so it is still increasing exponentially.) I attended the ISDC private space conference in 2006 when it was just weirdos and talks on the space elevator. I heard a solid talk about how the Space Shuttle proved reusability didn't make sense because the launch volume was so far below where it broke even to do it. And when Musk started, we all KNEW there wasn't enough launch demand to pencil out, either. But look at that chart. Supply does sometimes create its own demand. Even if the AI data centers getting built today don't make money, that's not a bad thing for the rest of us that aren't paying for them. Because we'll have the chance to buy up that capacity for pennies on the dollar. Just like JibJab, College Humor, Strong Bad, and a startup called YouTube couldn't have afforded to build the infrastructure they needed to operate, so will there be a zillion new ideas built on buying Nvidia chips by the pound. The current AI moment itself was started using excess capacity left over from the crypto mining mania. Don't worry that you can't imagine how anyone will ever turn a profit after buying all those Nvidia cards. Maybe they won't. But the price of AI compute is going to drop fast. Anything that's marginal today will be able to turn a profit after the bust. That's the biggest lesson from 2000.

How the GenAI Bubble is and is not like the Dotcom Bubble In late 1996, Fed chairman Alan Greenspan warned that the frenzied investment in any startup that had a .com in its name was “irrational exuberance.” But excitement was such that the warning, even coming from such a creditable source, was ignored. Between 1995 and March 2000, investments in the NASDAQ rose by 800% - But the bubble bust and by October 2002 all those gains were lost. When ChatGPT was released in November 2022 people were dazzled by its ability to generate text that appeared indistinguishable from human writing. It appeared a great advance in AI had been achieved and a new AI millennium was proclaimed. Intelligent machines would replace expensive human workers resulting in unprecedented productivity, boosting corporate profits sky high. Since then, we have seen a rally that has added almost $16 trillion to the S&P 500. But today creditable sources like Jim Covello, Head of Global Equity Research at Goldman Sachs, are warning that we are again in an investment bubble. He said: “Most technology transitions in history, particularly the ones that have been transformational, have seen us replace very expensive solutions with very cheap solutions, potentially replacing jobs with tremendously costly technology is basically the polar opposite.” That LLMs are vastly expensive, and from an energy standpoint even unsustainable, is now obvious. But can they deliver on the touted productivity gains? If not, it is not for lack of trying. Research by The Upwork Research Institute reveals that 39% of C-suite leaders are mandating the use of genAI tools, with an additional 46% encouraging their use. But the results are not encouraging, the same study found that nearly half (47%) of employees using genAI say they have no idea how to achieve the productivity gains their employers expect, and 77% say these tools have actually decreased their productivity and added to their workload. The Internet (eventually) actually did “replace very expensive solutions with very cheap solutions” but the dot-com startup investment bubble was irrational because there were no barriers to entry. After the bust, most failed right out of the gate and it would take decades for the real winners to emerge. LLMs startups also have no barriers to entry, the technology will always be vastly expensive, and in the end, it just doesn’t deliver. When this bubble bursts it could be ugly indeed. There may be no long-term winners, at least not big ones, this time around.

The current wave of artificial intelligence (AI) innovation brings to mind the lessons of the dot-com bubble. Back then, hype and capital fueled a flood of startups, but many chased ideas without clear value or sustainable business models. Take Pets.com, which gained massive attention but collapsed under the weight of unproven demand and unsustainable spending. Similarly, Webvan promised to disrupt grocery delivery but failed due to over-expansion and poor operational execution. In contrast, companies like Amazon and eBay thrived because they solved real problems and built models that could scale. Today, the AI space is showing similar signs. Many startups are focusing on use cases that lack clear value or trying to solve problems that don’t truly exist. While the potential of AI is undeniable, the rush to innovate often sacrifices practicality and impact. Innovation for its own sake is not enough. If AI startups are to avoid the fate of many dot-com casualties, they need to focus on creating solutions that address genuine, pressing needs. The winners in this era will be those who build sustainable models, deliver measurable value, and stay grounded in the realities of their markets. Let’s remember: technology isn’t about looking futuristic—it’s about making a meaningful difference. Are we learning from the past, or are we destined to repeat it? #artificialintelligence #technology #startups #strategy #entrepreneurship