Analysis
Fireworks closed a $1.505 billion Series D at a $17.5 billion valuation, led by Atreides Management, Index Ventures and TCV, with Evantic Capital, Lightspeed Venture Partners, Nvidia, 20VC, Bessemer Venture Partners and Menlo Ventures also participating. The company says its annualized revenue run rate has crossed $1 billion, up roughly 5x year-over-year since its last round.
Fireworks' pitch is that most enterprises don't want a general-purpose foundation model, they want a smaller, specialized model trained on their own data and served cheaply and reliably at scale -- the company says more than 95% of the 40 trillion tokens it now serves daily come from exactly that kind of fine-tuned deployment, not off-the-shelf API calls. That positioning puts Fireworks closer to an inference-and-fine-tuning platform than a model lab, competing most directly with Together AI, which raised $800 million at an $8.3 billion post-money valuation in its own Series C weeks earlier, and SambaNova, valued at $11 billion in a January raise.
“What to watch next: whether Fireworks' revenue growth holds through a chip-market downturn that could pressure the compute costs underlying its margins.”
At $17.5 billion, Fireworks now sits meaningfully above both comparables, a premium the company is justifying with revenue growth and the claim that specialized-model serving carries better unit economics than general-purpose API resale. Nvidia's participation is notable beyond the capital -- it signals the chipmaker wants direct exposure to the inference layer, not just GPU sales, as more enterprise AI spend shifts from training to serving.
For founders in the inference-infrastructure space, Fireworks' valuation trajectory is the clearest signal yet that investors are willing to pay a premium for companies that can prove they're capturing durable, specialized workloads rather than commoditized API pass-through. What to watch next: whether Fireworks' revenue growth holds through a chip-market downturn that could pressure the compute costs underlying its margins.
