ARR Leaderboard
Fireworks AI turned serving open-source models into a $1B+ annualized revenue run rate — up 5x in a year — and closed a $1.5B Series D at $17.5B to prove inference is its own market. Every reported ARR, funding and token-volume number is collected below, each traced back to a public source.
$1B+
Annualized revenue run rate, Jul 2026
5x
Year-over-year growth
40T+
Tokens served per day
$17.5B
Valuation, Series D Jul 2026
Annualized revenue run rate in USD millions, Series C through the Series D announcement.
$0 → $280M ARR
~3 years
2022 founding → Series C
~$280M → $1B+ ARR
~1 year
Series C → Jul 2026
Valuation $4B → $17.5B
~1 year
Series C → Series D
Token volume growth
10T → 40T+ / day
Series C → Jul 2026
2022
$0M ARR
Founded by Lin Qiao and ex-PyTorch colleagues from Meta to make open-model inference production-grade. — Company
2025
$280M ARR
$250M Series C at a $4B valuation, with ~$280M ARR and 10T+ tokens served per day. — Sacra / press
May 2026
$800M ARR
Sacra estimated ARR near $800M, up from ~$305M at the end of 2025. — Sacra
July 16, 2026
$1B ARR
Surpassed $1B in annualized revenue run rate — up ~5x year over year — serving more than 40 trillion tokens per day. — Fireworks / CNBC
July 2026
$1B ARR
$1.505B Series D at a $17.5B valuation, led by Atreides Management, Index Ventures and TCV with NVIDIA participating — a ~4.4x valuation jump in about a year. — Fireworks / CNBC
Annualized run rate
$1B+
Fireworks / CNBC, Jul 2026
Growth
~5x YoY
CNBC, Jul 2026
Tokens served
40T+ / day
Fireworks, Jul 2026
Valuation
$17.5B
Series D, Jul 2026
Series D
$1.505B
Fireworks, Jul 2026
Backers
NVIDIA, Atreides, Index, TCV
Fireworks
Founded
2022, Redwood City
Company
CEO
Lin Qiao
Company
Fireworks runs an inference cloud: per-token serverless APIs for open and custom models, plus dedicated capacity and fine-tuning for enterprises that want optimized private deployments. The wedge is cost and speed — serving open models on Fireworks' optimized stack undercuts closed frontier-model APIs, and the volume shows it: from 10T+ tokens per day at the Series C to more than 40 trillion per day at $1B+ annualized revenue in July 2026.
The next act is 'specialized intelligence' — custom enterprise models built on Fireworks' infrastructure, with planned distribution through Microsoft Foundry and Azure. The capital follows the revenue curve: a $250M Series C at $4B (~$280M ARR) was followed by a $1.505B Series D at $17.5B led by Atreides Management, Index Ventures and TCV, with NVIDIA staying on — roughly a 4.4x valuation jump in about a year as run rate grew ~5x.
Revenue Mix
FAQ
Clink bills by the token — checkout, usage metering and merchant-of-record tax handling for inference clouds serving trillions of daily tokens.
Bill Token Volume→