Fireworks AI raised a $1.5 billion Series D at a $17.5 billion valuation, led by Atreides Management, Index Ventures and TCV, with Nvidia and existing investor Lightspeed Venture Partners also participating, according to CNBC and SiliconANGLE reporting published July 16. The company says it now serves 40 trillion AI tokens per day and has surpassed $1 billion in annualized revenue, five times its total from a year earlier.
Fireworks operates in the AI inference and model-serving layer -- helping companies like Uber, Shopify and Doximity deploy, customize and run AI models in production -- a category that's grown in importance as enterprises shift focus from which frontier model to use toward how cheaply and reliably they can actually serve that model at scale to real users.
“What to watch next: whether Fireworks' revenue growth rate holds through the back half of 2026 as more enterprises build in-house inference capability.”
The round places Fireworks alongside Together AI and Baseten as the best-capitalized independent players in AI inference infrastructure, a layer increasingly seen as structurally advantaged the same way TSMC's foundry position is: profitable regardless of which specific foundation model ultimately wins the most enterprise adoption, since inference providers serve models from multiple labs rather than betting on one.
Nvidia's participation is notable less for the dollar amount and more for the pattern -- the chipmaker has now backed multiple companies across the AI software stack that drive incremental demand for its own hardware without directly competing with its core GPU business, a strategic-investment playbook it's run consistently through 2026.
The bear case: inference-layer margins are under constant pressure from both open-source serving alternatives and hyperscalers building comparable capabilities in-house, and a $17.5 billion valuation on $1 billion of revenue implies the market is pricing continued hypergrowth rather than steady-state margins. What to watch next: whether Fireworks' revenue growth rate holds through the back half of 2026 as more enterprises build in-house inference capability.