Fireworks AI is in early talks to raise new funding at a $30 billion to $40 billion valuation, Startup Fortune reported, citing Business Insider’s October 20 report. The proposed investment amount wasn’t disclosed, and the transaction hasn’t been finalized. Fireworks provides infrastructure that software teams use to customize open AI models and run them in production.
A developer can choose an open model and use a serverless API to prototype. Fireworks’ training tools support supervised and reinforcement fine-tuning, and teams can deploy the resulting model on the same platform before moving production workloads to dedicated GPUs with autoscaling. Fireworks supplies the training and inference infrastructure; the customer builds the application that uses it.
The company charges for serverless inference based on input, output and cached tokens, while on-demand deployments are billed by GPU time. Fireworks identifies Cursor’s coding models and Harvey’s legal AI as products built with its platform. That usage-based model ties Fireworks’ revenue to the computing consumed by customers rather than a fixed software-seat count.
In July, Fireworks closed a $1.505 billion Series D at a $17.5 billion valuation. The company said its annualized revenue run rate had surpassed $1 billion and that the platform served more than 40 trillion tokens daily. It attributed more than 95% of that usage to models specialized on customers’ proprietary data. Those operating figures are company-reported.
The reported $30 billion to $40 billion range would represent an approximately 71% to 129% increase from the July valuation. That is a substantial repricing after only about three months. The Startup Fortune report doesn’t state the proposed investment amount or identify a lead investor, so the terms remain much less developed than the headline valuation range.