Fireworks hits $18B valuation as companies seek cheaper AI models
The AI inference startup's meteoric rise reflects a broader industry shift away from expensive proprietary models toward leaner, cost-effective alternatives.
Fireworks AI has reached a $17.5 billion valuation as enterprises increasingly hunt for cheaper alternatives to the massive proprietary models that have dominated the landscape. For context, the company was valued at $4 billion just months ago following a $250 million Series C round in October 2025.
The platform operates on a pay-per-token pricing model, deploying open-source and third-party models like DeepSeek and Qwen with low latency. CEO Lin Qiao has been vocal about the company’s approach: smaller, customized models that can slash token costs by 5 to 10 times for specialized applications.
The revenue numbers back up the thesis. Fireworks AI is projecting an annualized run rate of roughly $800 million by mid-2026, up from an estimated $305 million at the end of 2025.
From scrappy startup to $17.5B behemoth
Fireworks AI’s ascent has been rapid even by Silicon Valley standards. The company raised a $52 million Series B at a valuation of $552 million in 2024. By late 2025, total funding had surpassed $327 million, with backers including Sequoia Capital and NVIDIA.