Hugging Face CEO: We Are in an ‘LLM Bubble,’ Not an AI Bubble
Clem Delangue, co-founder and CEO of Hugging Face, stated his view that the technology industry is currently in an "LLM bubble," not an overall "AI bubble." At an event hosted by Axios, he agreed that the topic of this bubble is today's trillion-dollar question. However, he emphasized that even if this LLM bubble were to pop—which is predicted to potentially happen next year—the future of AI technology itself is not at risk. According to him, current attention and funding are excessively concentrated on LLMs like those powering ChatGPT and Gemini, even though this is just a small fraction of AI's potential.
Delangue emphasized that LLMs are merely a subset of AI, while AI applications in other fields such as biology, chemistry, image, audio, and video processing are just beginning and will expand rapidly in the coming years. He criticized the current trend where many parties focus on the idea of building one giant model with massive compute to solve all problems. According to him, the reality of the future will shift towards a "multiplicity of models" that are more customized and specialized to solve specific, distinct problems.
As an example, he described the use case of a banking chatbot for customers. Delangue explained that such a system does not need a giant model that understands philosophy or the "meaning of life." Instead, companies can use smaller, specialized models that are much cheaper, faster, and can be run directly on the company's own enterprise infrastructure. This efficiency and specialization are what he believes to be the true future of AI implementation, rather than merely chasing all-knowing but resource-intensive models.
Although admitting that the bursting of the "LLM bubble" could impact Hugging Face to some extent, Delangue believes the AI industry is sufficiently diversified so that the impact will not destroy his business or the AI field at large. Hugging Face itself takes a different, more cautious approach compared to its competitors; they still hold half of the $400 million in funds they have raised. Delangue refers to this strategy as a capital-efficient approach, unlike other companies that are currently "panicking" and spending billions of dollars chasing short-term trends.
Regarding Hugging Face, the company is a leading open-source collaboration and community platform, often referred to as the "GitHub of the AI world." Hugging Face provides a space for developers, data scientists, and machine learning engineers to build, train, and deploy AI models. The platform hosts hundreds of thousands of pre-trained models and datasets accessible to the public, enabling the democratization of AI technology so it can be used for various applications ranging from Natural Language Processing (NLP) to computer vision without the need to build models from scratch.