The AI hype bubble might parallel the dot-com era bust

Similarities between dot-com and AI hype

The dot-com era was characterized by a world-changing technology that "broke out of the lab" with the commercialization of the internet via the World Wide Web, Phelan said. This required massive new infrastructure, and a gold rush ensued.

"People went crazy and invested a lot of money. There was a massive build-out of infrastructure," he said.

The AI hype bubble involves another world-changing technology that broke out with the availability of generative AI, which also requires a massive infrastructure build-out.

"Another gold rush ensued as everybody's fear of missing out pumps in a lot of money," Phelan said.

Another gold rush ensued as everybody's fear of missing out pumps in a lot of money.

The commonality between the two hype cycles is the capital surge, he said. The 1990s saw massive amounts of venture capital funding go to internet startups with high stock valuations but little actual revenue.

A similar investment craze is happening now with the AI hype bubble, where any company that has a business plan with AI in it is getting funding regardless of whether they have a proven business model.

"Even worse now than in the 1990s phase, the ratio of people losing money on the services they offer is quite a bit more now," Phelan said. "FOMO is definitely driving investments. It's a stake your claim time and get your claim as big as possible."

Differences between the eras

While there's reason to believe that the AI hype bubble will burst as the dot-com bubble did, Phelan explained that there are differences between the two eras.

One big mistake that caused the downfall of the dot-coms was that a backbone was built, but not the last mile of the infrastructure, leading to a slowed demand for backbone infrastructure. There is a very robust internet now, which means that AI startups have much more runway than the dot-coms had, he said.

Broadband was available in the dot-com era, but it hadn't rolled out widely, and many people were still on dial-up so there wasn't the demand for dot-com services.

"Venture funding dried up and software startups that weren't related to the build out and had minimal value propositions folded," Phelan said. "However, the building of the internet, Web 2.0 technologies and everything that went with that set the stage for what we're doing now."

Warning signs for the AI hype bubble

We are in the midst of the AI hype era right now, so it may be too early to know if the same issues are going to cause a similar collapse to the dot-com era. But there are warning signs.

For example, in the dot-com era, a major cause was what Phelan called "red money," wherein the value of the services ran far below their costs. It's estimated now that OpenAI spends $3.50 for every dollar of revenue that it brings in.

"That could lead to the same basic thing of what happened in the dot-com era -- they run out of money," Phelan said.

Another issue is circular funding in the AI space right now, with examples such as Nvidia investing heavily in OpenAI and OpenAI using that funding to buy and use Nvidia's chips, Phelan said.

"If you follow the investment circles right now, it's very confusing," he said. "Everybody is investing in everybody else, to the point where it seems that somebody pulls the wrong Jenga log, the whole thing is going to collapse."

The energy required to run AI could also be a significant factor in slowing the AI momentum. The lack of power needed for AI computing may not burst the bubble, but it could slow things down, Phelan said. Many AI companies may think that they can recoup any losses by driving up the volume of services they sell, but energy limitation could also lead to volume limitations.

Finally, there's a general anti-AI feeling that could play into the AI hype era that largely didn't exist in the dot-com era.

"There were people that were against dot-coms and so on, but I don't think a good communications network has ever been the villain in a sci-fi story," Phelan said.

Mitigating factors for the AI hype era

The AI hype bubble might burst as the dot-com bubble did, but it also might not happen -- at least in the same way -- due to several mitigating factors, Phelan said.

In the dot-com era most of the venture capital money went to startups that were fragile and didn't have the business to be sustainable. However, the money in the AI era is going to companies that are already fairly well-established or the money is coming from entities that are much more established than the dot-com financiers.

"They have deep pockets that can ride out a lot more than the venture capitalists who were funding the dot-com era bubble point," Phelan said.

Another mitigating factor is that the infrastructure is far more established for the AI market than it was for the dot-com era. 

"There are no dark GPUs, if you can build it, they come at this point in time," Phelan said. "There's no last mile to build out, the internet is already there to serve all of these, we can get the traffic of the users to there."

Heading for an AI crash?

It's clear that AI hype bubble is heading for a crash or an adjustment. There certainly will be some kind of adjustment, Phelan said. How big is unknown, but with the number of companies investing in building LLMs, there's going to be sorting out.

"There will be businesses going out, there will be some startups that are put out of business by the bigger companies who turn their startups into features of their platform," he said. "There's a lot of that going on now, where various business plans get ruined by the latest announcement from OpenAI or Anthropic."

There are many AI products and tools available now, but many of today's vendors will likely face a reckoning, according to Dominick Gray, president of Center City Communications, a telecommunications MSP based in Southfield, Mich.

Center City Communications is in the early stages of implementing AI, and Gray attended IT Expo to get some first-hand information on what's available. However, he does not expect that all the vendors who crowded the exhibition floor will be in it for the long term.

"A lot of vendors have high hopes and great intentions, but some won't make it because the funding will run out or the technology will pass them by just as they're getting out the door," he said. "We're trying to pay attention to the trends versus just focusing on a specific company."

The difference is in transformation

There are significant differences between the dot-com era and the current AI era, particularly in what the expected results are, according to Mark Beccue, principal analyst for artificial intelligence at Omdia, a division of Informa TechTarget.

The dot-com era was about moving companies toward digitization, while the AI era is about building efficiencies that fundamentally transform them, he said. The dot-com era did open new opportunities, but it didn't affect all companies as the AI era will.

Another significant difference between the dot-com era and today's AI environment is that major enterprise vendors have been investing quietly but heavily in embedding AI into their products, Beccue said.

These companies -- most prominently Adobe, IBM, Salesforce, SAP and ServiceNow -- are starting to see real returns for their AI investments.  

"None of these companies are by nature an AI vendor, but they've been the quiet, pragmatic ones that built AI under the hood," Beccue said. "When GenAI reared its head, all of those companies were ready because they already knew the fundamentals and had the culture in place."

Jim O'Donnell is a news director for TechTarget, where he covers IT strategy and enterprise ESG.