Clément Delangue

ESCP identifies Delangue as a 2012 Master in Management graduate. His own professional biography describes Moodstocks, a company applying machine learning to computer vision, as his first startup experience. In 2025, ESCP and Hugging Face announced a partnership giving the school’s community access to platform resources and learning materials.[2][8]

In September 2026, NVIDIA announced an agreement to acquire Hugging Face. Its announcement said the platform would retain its brand and continue supporting models, clouds and computing platforms from across the industry.[9]

Sharing models through a common library

Delangue is a co-author of the Transformers paper. The project brings different pretrained language-model architectures behind a common programming interface and connects them with a collection of community models. That design helps practitioners reuse trained systems without rebuilding every architecture’s loading and execution code. Paper co-authorship establishes a documented contribution; it does not mean he personally wrote the whole library or created the models hosted on the platform.[3]

Extending the platform to robotics

Delangue co-authored Hugging Face’s announcement of its Pollen Robotics acquisition. In a separate signed essay, he described open AI robotics as a combination of models, datasets and hardware that people can inspect and build upon. He also suggested a future store of reusable robot skills. The distinction matters: the essay presents his view and an idea for further development, while the acquisition announcement documents a company action. Neither establishes that he designed Pollen’s robots or authored the LeRobot library.[1][5]

Access as part of oversight

In his September 2023 written statement to the AI Insight Forum, Delangue argued that openness can improve the ability of researchers, regulators and civil society to inspect AI. He described a spectrum of access rather than a single unrestricted-release model, and acknowledged that all systems can be misused. His proposed approach combines transparency with controls such as community moderation and access gates. The statement is useful because it connects the company’s openness argument to concrete practices: documenting models and datasets, evaluating harms and making evidence available to more people.[4]

Competition and specialized models

In a July 2026 a16z interview, Delangue discussed regulation, local AI and routing tasks among specialized models. The episode frames competition as central to his preferred direction for AI development. It is a useful spoken introduction for someone deciding whether to follow him: his subject is the organization of the AI ecosystem as much as the capabilities of any single model. The episode page identifies the subjects discussed; its claims about industry outcomes should be understood as his argument rather than a settled forecast.[6]