San Francisco startup Runware, which aims to speed up generative AI, has announced $50 million Series A funding.
London-based Dawn Capital led the investment and included Comcast Ventures, SpeedInvest, Insight Partners and a16z Speedrun as other participants.
This round brings the funding raised by Runware to date to $66 million, and will be used to continue building what co-founder Flaviu Radulescu described as the company’s core mission: “An API for all AI.”
Runware was founded in 2023 in an effort to overcome the frustrations caused by the relatively slow speed of early generative AI tools, with the goal of making high-performance AI accessible to everyone.
In the two years since, it claims to have driven over 10 billion outputs for over 200,000 developers and over 300 million end users – all run through the same API.
Explaining the logic, Radulescu Said In a blog post: “Most teams hit the same roadblocks when trying to ship AI at scale: fragmented access to AI models, slow inference that breaks the user experience, and costs that rise faster than adoption.”
The company claims that by blending custom AI inference hardware with its optimized software stack, its Sonic Inference Engine can deliver “ten times lower price and faster performance” than traditional data center deployments.
It does this by aggregating approximately 300 AI model classes and hundreds of thousands of model variants into Runware’s API, which can be integrated into developer apps, allowing images, video, and audio to be generated through a single interface. The open source model can also be accommodated, with speeds claimed to be up to 40% faster over rival platforms.
Runware’s customers include Wix, Together.AI, ImagineArt, Quora, OpenArt, Freepik, and Higgsfield AI, and it says the new funding will be used to expand the platform.
According to Radulescu’s blog, the specific goal is “to make every AI model available to every developer through a single API… (and) that includes deploying all of the more than two million new AI models from Hugging Face by the end of 2026.”
Runware will also use the investment to increase its roll-out of affordable “inference pods” – efficient infrastructure that avoids the need to use data centres, which can be costly and take years to build.
