Stability AI, which funds the development of image-generating systems such as open source music and dance Diffusion and Stable Diffusion, today announced it has raised $101 million in a funding round led by Coteau and Lightspeed Venture Partners in partnership with O’Shaughnessy Ventures. LLC. According to a Bloomberg source, the tranche values the company at $1 billion post-money and comes as demand for AI-powered content production grows.
London- and San Francisco-based Stability AI is the brainchild of CEO Emad Mostak. Graduating from Oxford with a Masters in Mathematics and Computer Science, he worked as an analyst at various hedge funds before switching gears to more public-facing work. Mostak co-founded and bootstrapped Stability AI in 2020, citing a personal fascination with AI and a lack of “organization” in the open source AI community.
“No one has voting rights except our employees – no billionaires, big funds, governments or those who control the company or the communities we support. We are completely independent,” Mostak told TechCrunch in a previous interview. “We plan to use our computation to accelerate open-source, foundational AI.”
Stability AI has a cluster of more than 4,000 Nvidia A100 GPUs running on AWS, which it uses to train AI systems, including stable diffusion. It’s also costly to maintain — Business Insider reports that Stability AI’s operations and cloud costs exceed $50 million. But Mostak has repeatedly emphasized that the company’s R&D will enable it to train more advanced models.
The company’s work has not escaped controversy, it’s worth noting. Stable Diffusion’s open source release has been used to create offensive content such as graphic violence and obscene, non-consensual celebrity deepfakes. And Getty Images prohibits uploading content generated by systems like Constant Spread for fear of intellectual property disputes. (Stable Diffusion is trained on a data set that includes copyrighted works — and even private medical records.)
Stability AI is also the subject of a recent critical letter from US House Representative Anna G. Ishoo (D-CA) to the National Security Advisor (NSA) and the Office of Science and Technology Policy, in which she urged the NSA and OSTP. Address the release of “insecure AI models” that “do not moderate the content produced on their platforms”.
Stability AI has taken a very hands-off approach to moderation to date, including filtering tools with the open source Stable Diffusion package, but allowing users and companies to run the system as they see fit, as long as they follow company license terms. “One percent of people are obnoxious and weird, but that’s human,” Mostak said in a previous interview. “Of course, our belief that this technology will prevail, and the paternalistic and somewhat condescending attitude of many AI fans, has misled the community into believing it.”
Stability AI plans to make money by training “private” models for customers and acting as a common infrastructure layer. It also offers a platform and API, DreamStudio, through which its models can be accessed by individual users — Mostak told Bloomberg that DreamStudio has more than 1.5 million users, who have created more than 170 million images, and a steady spread of more than 10 million daily users. “On all channels.”
Meanwhile, the open source version of Stable Diffusion has been downloaded more than 200,000 times, according to a press release published by Stability AI this morning.
According to Mostak, the capital from the funding round will support the large-scale deployment of custom versions of Stable Spread for users and investment in more supercomputing power. It is also positioned to hire more people, with Mostak saying it expects to grow from 100 to 300 employees in the next year.
Stability has made several high-profile hires recently, bringing in AI research scientists From Google Brain and futurist and public speaker Daniel Jeffries.
Beyond stable expansion, Stability AI says it has other commercialized projects in the works, including AI models for generating audio, language, “3D” and even video. One of them is the aforementioned Dance Diffusion, a set of algorithms and tools that can generate clips of music by training them on hundreds of hours of existing songs.
Mr. Vishwanath, general partner at Coatue, said in a statement: “At Coatue, we believe open source AI technologies have the power to unlock human creativity and achieve greater good. Sustainability AI is a big idea that dreams beyond the immediate applications of AI. We are excited to be a part of Stability AI’s journey and look forward to seeing what the world creates with Stability AI’s technology.