Nvidia-backed Fireworks hits $17.5 billion valuation in latest funding round

Fireworks raised $1.5 billion in a Series D round that valued the Nvidia-backed startup at $17.5 billion after the company said it had reached $1 billion in annualized revenue. Fireworks also said its platform now processes more than 40 trillion AI tokens a day, a scale that positions the company as a major player in the infrastructure layer around artificial intelligence.

The valuation highlights a split in the AI market between firms building foundational models and the companies selling the tools to deploy, manage and scale them. Investors are rewarding businesses that sit close to inference, model hosting and enterprise rollout, especially when those businesses are tied to Nvidia’s ecosystem and the chips that underpin the boom. Fireworks is being priced less like a model lab and more like the kind of software and infrastructure company that can profit as enterprises move AI systems into production.

That bet comes as concern rises over the cost of running AI at scale. A June 29 Reuters report said soaring AI bills are reshaping how businesses choose models, and that dynamic favors cheaper hosting, lower-latency delivery and more efficient deployment tools. Fireworks is pitching itself directly into that need, with a platform built around cheaper model hosting and the claim that its systems can handle heavy enterprise workloads.

The new valuation is also a sharp jump from the $4 billion price tag Fireworks carried in 2025. Sacra estimated that the company reached $800 million in annualized revenue in May 2026, up from about $305 million at the end of 2025, and said the customer base grew from roughly 1,000 companies at the time of its Series B to more than 10,000 by October 2025. Fireworks said in its July 15 blog post that the Series D brought the company to $1 billion in ARR.

Fireworks’ LinkedIn profile says the company powers production workloads at Uber, DoorDash, Notion and Cursor, and says its systems deliver 15 times faster speed, four times lower latency and four times more concurrency than closed models. Those are company claims, but they help explain why investors are assigning such a high price to a startup built around deployment and inference rather than model creation itself. The size of the round suggests that, for now, capital is still flowing toward the infrastructure businesses closest to Nvidia’s hardware advantage.