AI Companies Admit They’re Worried About a Bubble

LISBON, Portugal – Concerns regarding a potential bubble in the artificial intelligence sector are mounting, as expressed by top tech executives to CNBC. This sentiment highlights a growing unease within the industry concerning escalating valuations and the sustainability of the current AI boom.

The market has recently been grappling with the possibility that excessive capital is being funneled into AI, possibly overshadowing realistic revenue projections and concrete profits, which consequently casts doubt on inflated valuations. This apprehension is occurring despite continuous advancements in AI technology and its integration into various sectors.

Previously, warnings about potentially overvalued assets were primarily voiced by investors and financial leaders. Figures such as Goldman Sachs CEO David Solomon and Morgan Stanley’s Ted Pick have cautioned about potential market corrections as valuations of major tech firms reach unprecedented levels. However, these views of are now being supported by the very companies developing the technology.

Adding fuel to the fire, Michael Burry, renowned for “The Big Short,” has accused major AI infrastructure and cloud providers–commonly referred to as “hyperscalers”–of potentially underreporting depreciation expenses on semiconductors. Burry suggests that this accounting practice could lead to vastly overstated profits at companies like Oracle and Meta. His investment decisions further underscored this stance, with disclosed put options betting against Nvidia and Palantir. This position comes at time where these two players are being lauded as essential to the AI infrastructure.

CEOs actively involved in AI development echoed these concerns during interviews with CNBC at the Web Summit tech conference in Lisbon this week.

“I believe evaluations are quite inflated in certain cases, displaying signs of a bubble forming,” Jarek Kutylowski, CEO of the German AI firm DeepL, stated to CNBC on Tuesday. DeepL is the company behind some of the most sophisticated language models on the market today, taking on the likes of Google and Microsoft, and it’s core business translates directly to the future of AI.

Picsart CEO Hovhannes Avoyan concurred with Kutylowski’s assessment.

“We observe numerous AI companies raising capital at extremely high valuations despite lacking substantial revenue,” Avoyan stated. He further noted that this phenomenon is a “concern,” and highlights the issues when investment precedes innovation.

In Avoyan’s view, the market is assigning high values to smaller startups based on “hype and perceived potential” rather than actual sales, referring to companies receiving financial support despite minimal revenue generation. This trend is particularly evident in the realm of “vibe coding,” where AI is used for coding tasks without requiring in-depth technical knowledge.

Despite misgivings about valuations, the tech industry remains optimistic about AI’s long-term potential.

Lyft CEO David Risher acknowledges the potential impact of AI, while noting risks associated with the rapid growth.

“There is no question that we are in a financial bubble. It reflects the incredible, transformational nature of this technology. Everyone wants to be involved.”

Risher says there is a crucial distinction between the financial bubble and the industrial outlook.

“The data centers and AI model creation will have lasting value because they are transformational, simplifying and improving lives… However, the financial aspect carries considerable risk.”

Tech CEOs also shared their insights regarding the anticipated demand for AI from businesses in 2026, addressing a key area of interest for investors.

“There is substantial demand and interest in AI, with widespread recognition of its transformative potential for businesses. AI can significantly enhance efficiency across various operations,” Kutylowski stated.

However, Kutylowski pointed out that businesses are facing challenges in adopting AI, underscoring the need for further progress and development. “More advancements are needed before widespread adoption can be achieved, and every enterprise and organization fully understands and leverages AI.”

DeepL recently introduced a versatile “agent” designed to perform tasks on behalf of employees, alongside its AI translation tool.

Francois Chadwick, CFO of enterprise AI company Cohere, stated that “demand is definitely there” in on Tuesday.

Even with concerns overstretched valuations and substantial capex spending, the investment into artificial intelligence shows no signs of slowing down. Accel, a venture capital group, published a recent report estimating that the expansion of new AI data center capacity would reach 117 gigawatts by 2030, requiring approximately $4 trillion of capital expenditure over the next five years.

According to the Accel report, roughly $3.1 trillion in revenue would be needed to recoup the projected capex.

Already in 2025, announcements of deals worth billions have come from Nvidia and OpenAI as they look to invest in data center capacity to keep up with demand.

According to Accel partner Philippe Botteri, revenue will be spurred on by more AI models (requiring constant capacity to be trained), new AI services and the “agentic revolution in the enterprise.”

“Agentic” is the ability of AI to perform tasks automatically on behalf of users.

However, there are those who don’t think this spending is necessary.

Novo Capital Managing Partner Ben Harburg says the figures being discussed are being overblown.

“We hear crazy numbers about how much energy is going to be needed and how many chips are going to be needed, although I think there is probably more of a bubble brewing there than on the actual product front,” he told CNBC on Tuesday.

“I think we’re starting to realize that there’s been over exuberance around data centers. Even Sam [Altman] would privately admit that they need fewer chips than they originally set out, they need less capital, and even less energy.”