Clem Delangue, CEO and co-founder of Hugging Face, believes that while we aren’t experiencing an AI bubble, there is currently a “large language model (LLM) bubble” that could soon burst. Speaking at an Axios event on Tuesday, Delangue, who leads the well-known AI platform and community, acknowledged that whether or not there’s a bubble is the “trillion-dollar question” right now. However, he doesn’t think the future of AI as a whole is threatened if the LLM bubble collapses.
Delangue argues that it’s the LLMs—such as those behind ChatGPT, Gemini, and similar chatbots—that are attracting disproportionate hype, and this focus may be temporary.
“My view is that we’re in an LLM bubble, and it could burst as soon as next year,” Delangue said. “But LLMs are only one part of AI. When you look at AI’s applications in fields like biology, chemistry, images, audio, and video, I think we’re just getting started, and the coming years will bring much more,” he added.
He pointed out that LLMs aren’t the answer to every problem, and that smaller, more targeted models will likely become more popular going forward.
“Right now, all the buzz, investment, and attention is centered on the idea that a single, massive model built with huge computing resources can address every need for every business and individual,” Delangue said. “But in reality, over the next months and years, we’ll see a range of models that are more tailored and specialized, each designed to tackle specific challenges.”
He gave the example of a chatbot for banking customers.
“You don’t need it to answer philosophical questions,” Delangue explained. “A smaller, more focused model can do the job more efficiently and at a lower cost, and it could even run on your own company’s infrastructure. That’s where I see AI heading.”
Delangue acknowledged that if the LLM bubble bursts, Hugging Face might feel some effects. Still, he emphasized that the AI sector is vast and already diversified, so even if LLMs are overvalued, it won’t dramatically affect the broader AI industry or his company.
He also mentioned that Hugging Face still has half of the $400 million it raised in reserve. This careful spending sets the company apart from many other AI firms, especially those focused on LLMs.
“In the AI world, that’s considered profitable, since other companies are burning through not just hundreds of millions, but billions of dollars,” he remarked.
In contrast, Hugging Face is opting for a more efficient use of capital.
“A lot of people seem to be moving quickly—maybe even acting out of fear—and are focused on short-term gains. Having worked in AI for 15 years, I’ve seen these cycles before,” Delangue said. “We’re taking those lessons to heart and aiming to build a company that’s sustainable and makes a lasting impact.”
