Fireworks AI Raises $1.5B and Valued at $17.5B Amid Inference Infrastructure Boom

US-based Fireworks AI raised $1.505 billion in a Series D funding round led by Atreides Management, Index Ventures, and TCV, with participation from Nvidia and Bessemer, at a total valuation of $17.5 billion, marking the largest funding deal for inference infrastructure platforms since the onset of the AI wave.

Fireworks AI crossed the $1 billion threshold in annualized revenue in July 2026, making the 17.5x revenue multiple high but justifiable in the infrastructure market. Its rival in this sector, Together AI, raised $800 million at an $8.3 billion valuation last July, forming a duo that dominates the enterprise inference market.

This funding reflects a fundamental shift in investment philosophy: after training large AI models dominated investor focus, serving these models and processing daily user requests has become the real commercial opportunity, generating recurring revenues and high margins similar to cloud software markets.

The main risk remains the threat from cloud giants like Amazon, Microsoft, and Google, which own proprietary inference platforms, as analysts believe they can offer the service within their cloud packages in 18 to 24 months, which could narrow the margins of independent companies and completely redraw the competitive landscape.

What do these terms mean?

Inference Infrastructure (AI Inference): The technical infrastructure that powers AI models and responds to user queries in real time. Just as training a model is like building a car engine, inference is running it reliably and fast every day.

Venture Capital: Investment funds that inject money into high-risk startups in exchange for an equity stake, in the hope of realizing massive returns upon a sale or initial public offering (IPO) on the stock market.

Revenue Multiple: The ratio of a company's valuation to its annual revenue. A 17.5x multiple means investors are paying $17.5 for every $1 the company earns, betting on massive future growth.