Welcome, Hugging Face!
We are thrilled to announce our investment in Hugging Face. Hugging Face has built the largest and most important AI open source community in the world.
- Founder: Clément Delangue, Julien Chaumond, Thomas Wolf
- Sector: Artificial Intelligence
- Location: Brooklyn, NY
At Salesforce Ventures, we strongly believe open source will shape the future of artificial intelligence. Open source enables collaboration, transparency, accountability, innovation, and trust. As AI becomes increasingly critical across industries, an open ecosystem will be key to ensuring AI technology develops responsibly and ethically.
That’s why today, we’re excited to announce we’re leading the Series D round in Hugging Face, the leading open source platform for data science and machine learning (ML). Hugging Face acts as the central hub connecting AI developers, researchers, and enthusiasts—like a GitHub for AI.
Founded by Clément Delangue, Julien Chaumond, and Thomas Wolf in 2016, Hugging Face began as an AI chatbot. The founders soon realized the true potential was not the chatbot itself, but the natural language processing (NLP) models powering it. We decided to back this talented founding team because of their creative vision, technical expertise, and proven ability to build passionate communities around their products. In particular, Clem’s business acumen, Julien’s engineering skills, and Thomas’ research background form a powerful combination capable of advancing the field of AI.
In 2018 and 2019, Hugging Face garnered fame and a burgeoning open source following as it built a Pytorch replica of the popular Transformer model which was widely adopted, and created the Transformers library that provides easy access to thousands of pre-trained models, including state-of-the-art NLP models. Since then, Hugging Face has expanded into other categories beyond NLP, such as computer vision and speech, and built the world’s largest open source community for AI with over 2 million users interacting with its models, datasets, and other resources.
Open source has been the driving force behind much of the rapid progress in AI over the past decade. Nearly all major ML frameworks like TensorFlow and PyTorch are open source. Similarly, breakthrough models like LLaMA, BERT, and Stable Diffusion are open source. By transparently sharing ideas and innovations, researchers can build upon each other’s work to advance the field faster than any one organization could alone. Open source also promotes transparency and trust in AI systems and allows researchers worldwide to inspect, audit, and improve them. This transparency is critical as AI is deployed into sensitive domains like healthcare and finance. As we hear time and again from enterprises, they want to adopt a hybrid strategy (or approach) that combines the benefits of open and closed-source models.
We believe the future of AI will be driven by a thriving open ecosystem of researchers, developers, enterprises, and enthusiasts collaborating together. Hugging Face sits at the center of the open AI movement and has become the most important community for ML researchers and practitioners.
The Hugging Face Hub brings together over 1 million repositories, including models, datasets and app demos, in one central platform. This makes it easy for anyone to find and access the latest innovations in AI, collaborate with others, and build on top of existing work. The Hub contains essentially all major open source AI models and is frequently the first destination for researchers to release their work – for instance, the much talked about LLaMA 2 model from Meta, Falcon, Vicuna and even Salesforce research team’s BLIP model – making Hugging Face a one-stop shop for the ML community. It promotes a virtuous cycle where the platform facilitates extremely rapid innovation cycles. This open ecosystem approach has been critical to the meteoric progress in AI over the past decade.
Beyond accelerating research, Hugging Face also provides the tools enterprises need to operationalize AI. This includes inference endpoints to deploy models into production, AutoTrain for scalable, low-code model training, Spaces to easily build and share demos and ML apps, and private Hugging Face Hubs for internal collaboration and model hosting. Hugging Face supports the entire ML workflow from research to deployment, enabling organizations to go from prototype to production seamlessly. This is another vital reason for our investment in Hugging Face – given this platform is already taking up so much of ML developers and researchers’ mindshare, it is the best place to capture the end-to-end ML workflow of its users.
The combination of its massive open community and enterprise-ready tools for deploying AI makes Hugging Face a fundamental part of any organization’s ML strategy. For both researchers and enterprises, Hugging Face is the gateway to the future of AI. This has led to fast-growing enterprise team adoption, an increasing number of whom have started to pay for the various ML tools on the company’s platform. We were deeply impressed by the speed of commercial growth (one of the fastest we have seen in recent history) and at the continuous amazing feedback we received through customer conversations. Many of these customers are not just the next AI native startup or big tech company who already has the ML engineering talent, but established enterprises from traditional industries such as healthcare, financial services and automotive. This speaks to how profoundly impactful the Hugging Face platform is to companies’ journey to AI adoption and the excitement from customers further solidified our belief that Hugging Face will become a generational company.
We believe open source AI, and Hugging Face’s role in catalyzing it, will fundamentally transform industries. We’re thrilled to partner with Clem, Julien, Thomas, and the entire Hugging Face team on their mission to democratize AI. Together, we’ll work to maximize the positive impacts of open AI across industries and society.
Welcome to the Salesforce Ventures family, Hugging Face!
Related Articles
