Fireworks AI, a startup that provides infrastructure for running customized AI models, has raised a $1.505 billion Series D round at a $17.5 billion valuation, the company announced on July 15.
A Big Round for AI Infrastructure
The round was led by Atreides Management, Index Ventures, and TCV, with additional participation from Evantic Capital, Lightspeed Venture Partners, Nvidia, 20VC, Bessemer Venture Partners, and Menlo Ventures, according to Fireworks’ announcement. The company said its annualized revenue run rate has topped $1 billion and that its platform now serves more than 40 trillion tokens a day.
Betting on “Specialized” Models
Fireworks built its business around what it calls “specialized intelligence”: rather than renting access to a single general-purpose chatbot, companies fine-tune smaller models on their own data and workflows, then run them on Fireworks’ infrastructure. The company says more than 95% of the tokens it processes daily now come from these customized models rather than off-the-shelf frontier systems. It cites coding tool Cursor and legal AI platform Harvey among the companies running specialized models on its platform, alongside enterprise customers such as Uber, Shopify, and GitLab.
Riding the Inference Boom
The raise adds to a wave of funding for companies that run — rather than train — AI models, as demand for inference compute has grown alongside the spread of AI agents and coding assistants. Fireworks said the new capital will go toward expanding its compute infrastructure and engineering team to keep pace with that demand.
The Series D more than doubles Fireworks’ valuation from its previous funding round, though the company did not disclose that prior figure in its announcement. It also did not give a timeline for further fundraising or a public listing.