As big tech companies increasingly spend billions on AI initiatives, speculation about an “AI bubble” has dominated the country’s economic discourse.
The phrase “AI bubble” denotes recent concerns about tech companies investing hundreds of billions of dollars in AI, despite consumers only spending $12 billion on AI technology yearly. Pitt professors shared their views on whether massive investments in AI will bring returns for the country and for Pittsburgh.
An economic bubble occurs when there is a rise in the market value of a certain product due to hopeful speculation about its future, which results in a crash after investors withdraw their money. The current massive AI investment also draws concern due to what some see as “circular investment” among big tech companies, which occurs when companies in the same field heavily invest in each other.
Morgan Frank, an assistant professor in the School of Computing and Information, said he thinks the AI bubble could burst, considering OpenAI has invested over a trillion dollars in infrastructure to build data centers. Frank noted that the company’s investment is twice as big as the federal government’s Bipartisan Infrastructure Law, a $1.2 trillion investment in repairing and modernizing the country’s infrastructure.
“OpenAI is spending more than the federal government,” Frank said. “It’s an insane amount of money, and I don’t know how they’re doing that, but they see payoffs.”
Brad Messner, an assistant professor in the School of Business, said investors are betting on AI because there’s currently a competitive race for companies to develop the best and most lucrative AI technology.
“People are overinvesting at what is technically an inflated valuation because right now we’re in a race,” Messner said. “We’re in that bucket right now of asking, ‘Who are going to be the winners?’”
As for AI’s prevalence in Pittsburgh, Eric Donald, an assistant professor of economics, said he believes the city is a “relatively small player” in the field of AI and will most likely not overinvest heavily like other regions such as Silicon Valley.
“Pittsburgh strikes me as [being] in a pretty good spot of not being left behind, but not being so far in the lead that you get vast overinvestment either,” Donald said. “We’re not a top AI hub — not like Silicon Valley. It just seems like we’re attracting more investment relative to our size than other old Rust Belt cities.”
Frank said Pittsburgh is not a huge AI hub because startups in the city often relocate elsewhere. For example, Zach Lipton, a professor in the School of Computing at Carnegie Mellon University, cofounded an AI-health startup in Pittsburgh which raised hundreds of millions of dollars in venture capital and shifted much of its corporate functions to California.
“Pittsburgh’s not very good at hanging on to tech startups,” Frank said. “We have great universities — CMU and Pitt — and there are a lot of startups that spin out of that ecosystem, spend a couple of years in Pittsburgh and then as soon as they have some initial success, they get sent somewhere else.”
Messner said he believes Pittsburgh has become a significant market for AI, noting that many large tech firms are planning data centers around the region. However, he said, data centers using an abundance of natural resources, such as water, could add to the concerns about a potential AI bubble.
“One thing causing some people to pull back is the fear that we’re moving so fast and haven’t stabilized how we’re handling a lot of these resources,” Messner said. “We were gung-ho for years now with trying to push these systems, and now, we’re realizing this is actually more of an infrastructure that we need to look at.”
Donald said it’s common for a new technology — like AI — to emerge without having the full infrastructure for it built yet. He referenced how factories in the 1900s took a long time to incorporate electric engines, though the technology was introduced decades before. Donald said he believes society’s adaptation to AI will somewhat reflect this process.
“We have this very large technological innovation that we don’t really know how to use properly,” Donald said. “To use [AI] most effectively, I think we would have to substantially reorient a lot of our production processes and work relationships, which is going to take a lot of time.”
After this happens, Donald said the “productivity gains from that technology take time to realize,” and will then match the current massive investments.
“We’ll then use the infrastructure — the data centers that we’re building now — and the faster you expect this diffusion, the more justified would be the financial investment now,” Donald said.
However, Messner said he believes he has not seen substantial returns for companies who invest in AI technology. Companies who have replaced human staff with AI technology have slowly rehired their workers because they likely “did not quite get what they were hoping for,” according to Messner.
“People are using these tools, but are their tools really showing improvement in valuation, operations and efficiency?” Messner said. “There’s promises of it, but we’re not seeing it just yet.”
Messner said AI companies are posing a “big ask and promise” by claiming the infrastructure will stabilize the AI platforms and drastically increase its quality and value. He said investors may soon realize there’s “no plan on how they’re getting [their money] back.”
“Investors aren’t going to keep throwing money in if there’s no potential for return,’” Messner said. “Yes, we have a 20 bucks-a-month subscription to the higher versions of these platforms, but that’s not going to push the needle enough to justify the investment.”
Messner added he feels the current immense AI investment may be an “investment black hole.”
“We really need to be looking at what financial model is going to work, and nobody has that yet,” Messner said. “More and more investors are now starting to ask, ‘How long do I have to truly wait? And, is it better now when I can get at least a decent portion of that fund back?’”
In the grand scheme of things, Donald said his chief concern with current AI investment is the potential for overall instability in financial markets — which could “slow down the entire economy.”
“The concern here is that the stock market is so concentrated in technology firms that if they start to go down, that will generate shockwaves that create financial instability,” Donald said. “This is what could cause a recession. People become scared, they start spending less money and investing less. Then, there’s less demand for labor, which could reduce salaries.”
If overall financial markets remain stable, Donald said the average American wouldn’t be highly affected and would also benefit from the new AI technologies that come out of the investment.
“An optimistic story of an AI bubble is that private investors are convinced to put out a lot of money into things that are socially useful, and private return to those investments is not necessarily what they were hoping for,” Donald said. “Then, a bunch of rich people lose money, but society has a bunch of great new ideas and infrastructure as a result.”