Hugging Face co-founder and CEO Clem Delangue stated that we are not in an AI bubble but rather in an “LLM bubble”—one that could burst soon. Speaking at an Axios event on Tuesday, the entrepreneur behind the prominent AI platform and community acknowledged that the bubble question is today’s “trillion-dollar issue,” yet he emphasized that even if it bursts, the future of AI itself remains secure.
Delangue argued that excessive attention is being placed on large language models (LLMs)—the kind powering ChatGPT, Gemini, and other chatbots—and that this focus may not be sustainable.
“I believe we’re in an LLM bubble, and it might pop as early as next year,” Delangue explained. “But LLMs are just one subset of AI. When it comes to applying AI in biology, chemistry, images, audio, and video, I think we’re still in the very early stages, and we’ll see far more progress in the coming years,” he noted.
He stressed that LLMs aren’t the optimal solution for every problem and predicted a shift toward smaller, more specialized models in the near future.
“All the attention, all the focus, all the funding has converged on the idea that building one massive model with enormous compute can solve every problem for every company and every person,” Delangue said. “But reality will show that over the next few months and years, we’ll see more customized, domain-specific models tackling distinct challenges.”
As an example, he cited customer service chatbots used by banks.
“You don’t need it to explain the meaning of life, right? You can deploy a smaller, specialized model that’s cheaper, faster, and even runs on a company’s own infrastructure. That’s where I see the future of AI heading,” Delangue pointed out.
The Hugging Face founder admitted that a collapse of the LLM bubble could impact his company to some degree, but he underscored that the AI industry has become so vast and diversified that an overvaluation in one segment—like LLMs—won’t significantly harm the broader field or his business.
He also highlighted that Hugging Face still holds half of the $400 million it raised in reserve—a capital-efficient strategy that contrasts sharply with the spending habits of many other AI firms, particularly those focused on LLMs.
“By AI standards, that’s called profitability—because other companies aren’t spending hundreds of millions; they’re clearly spending tens of billions,” he remarked.
In contrast, Hugging Face has adopted a more fiscally disciplined approach.
“I think many players right now are rushing—even panicking—and taking a very short-term view. I’ve been in AI for 15 years, so I’ve seen several cycles,” Delangue added. “We’re learning from them and striving to build a long-term, sustainable, and impactful company for the world.”
