AI companies admit they are concerned about bubbles

LISBON, Portugal — Top tech executives told CNBC they fear a bubble is forming in the artificial intelligence space, highlighting growing anxiety within the industry over soaring valuations.

In recent weeks, the market has floated the idea that too much capital is being poured into the AI ​​boom, clouding the outlook for revenue and actual profits and casting doubt on lofty valuations.

So far, warnings about inflated valuations have mostly come from investors and financial leaders. Goldman Sachs’ David Solomon and Morgan Stanley’s Ted Pick are warning of a possible correction as valuations for some big tech companies reach historic highs.

This concern was crystallized by prominent “Big Short” investor Michael Varley, who this week accused major AI infrastructure and cloud providers, or “hyperscalers,” of underestimating chip depreciation costs. Mr. Burley is a member of Oracle and Meta May be highly exaggerated. He recently revealed a put option betting on: Nvidia and Palantir.

But CEOs of companies that are developing AI themselves expressed concerns in interviews with CNBC this week at Web Summit, a tech conference in Lisbon.

“I think valuations are pretty overstated here and there. I think we’re on the verge of a bubble,” Jarek Kutilovski, CEO of German AI company DeepL, told CNBC on Tuesday.

Picsart CEO Hovhannes Avoyan echoed similar sentiments.

“We’re seeing a lot of AI companies get huge valuations without any revenue,” Avoyan told CNBC on Tuesday, adding that was “concerning.”

He noted that the market values ​​small start-ups with “some noise and ambiance revenue” and those that are supported even if their sales are minimal.

Vibe’s revenue is a play on the term “Vibe coding,” which refers to coding using AI without requiring deep technical expertise.

The technology industry remains bullish on the long-term potential of AI, despite concerns about valuations.

Lyft CEO David Risher said there is reason to be optimistic given the potential impact of AI, but he also acknowledges the risks.

“Let’s be clear: we’re totally in the middle of a financial bubble. There’s no question about it, right? Because this is an incredibly innovative technology. We don’t want anyone to be left behind.”

Risher went on to argue that there is a difference between a financial bubble and the outlook for an industry.

“The data centers and all the modeling, all that stuff is going to be around for a very long time because it’s transformational. It makes people’s lives easier, it makes people’s lives better… On the other hand, you know, financially, it’s a little bit risky right now.”

Tech CEOs also discussed the outlook for AI demand from companies in 2026, as investors look for some clues about what AI demand will look like.

“I think there’s a lot of demand, there’s a lot of interest. I think everyone understands that AI can bring magical things to business, and… we can all operate at another level when it comes to efficiency,” Kutilovsky said.

Still, companies are “struggling to adopt” AI. “We’re going to make further progress, but like all companies and organizations, I don’t think we’ll ever get to the point where we can say we’ve completely solved it,” Kutilovsky said.

DeepL’s core product is an AI translation tool, but it recently launched a more general-purpose “agent” designed to perform tasks on behalf of employees.

Francois Chadwick, chief financial officer at Cohere, which also focuses on enterprise AI, told CNBC on Tuesday that “the demand is definitely there.”

Despite concerns about inflated valuations and large capital expenditures, investment in artificial intelligence does not seem to be slowing down. New AI data center capacity is expected to reach 117 gigawatts by 2030, representing about $4 trillion worth of capital investment over the next five years, according to a report released this week by venture capital group Accel.

The company would need about $3.1 trillion worth of revenue to repay that capital investment, according to the Accel report.

Already this year, a number of deals worth billions of dollars have been announced by the likes of Nvidia and OpenAI, which are looking to develop data center capacity around the world to meet demand.

Philippe Bottelli, a partner at Accel, says there are three factors driving its returns. More powerful AI models that require training capabilities, the availability of new AI services, and an “agent revolution in the enterprise.”

“Agent” is a term often used to describe a type of AI tool that can automatically perform tasks on your behalf.

But not everyone believes the huge expense is necessary.

Ben Harburg, managing partner at Novo Capital, said the numbers big tech companies are discussing about future investments may be exaggerated.

“You hear all these crazy headline numbers about how much energy it’s going to take, how many chips it’s going to take, and so on, but again, I think there’s probably more of a bubble going on there than maybe on the front end or the front end of the actual product,” Harburg told CNBC on Tuesday.

“I think we’re starting to realize that there was probably too much excitement around data centers. I think even Sam (Altman) will personally admit that it requires fewer chips than originally planned, it requires less capital than originally planned, it requires less energy than originally planned.”