Fireworks AIRegister for a premium account to gain access to Sterling AI.
Get StartedThings you can ask Sterling:
Fireworks AIProduction inference platform for open-source AI; 4 to 9x faster than vLLM on high-traffic apps like Cursor and Notion
Annualized Revenue (May 2026)
~$800M
per Sacra; up from ~$305M (end-2025)
Valuation
$4B closed / $15B in talks
Oct 2025 Series C closed; $15B reported in talks (May 2026)
Inference Speed-up vs vLLM
4 to 9x
FireOptimizer kernel-level optimizations; cornerstone of competitive moat
Founder Pedigree
Lin Qiao (ex-Meta PyTorch lead)
Co-founded PyTorch at Meta; deep low-level systems credibility
Total raised: ~$330M+ · Latest valuation: $4B (Oct 2025 closed); ~$15B reported in talks (May 2026)
| Round | Date | Amount | Valuation | Key Investors |
|---|---|---|---|---|
| Seed | 2022 | Undisclosed | - | BenchmarkY Combinator |
| Series A | 2023 | $25M | - | Benchmark (lead)Y CombinatorDatabricks VenturesSnowflake Ventures |
| Series B | Jul 2024 | $52M | $552M | Sequoia (lead)BenchmarkDatabricks VenturesSnowflake Ventures+1 more |
| Series C | Oct 2025 | $250M | $4B | Lightspeed (co-lead)Index Ventures (co-lead)EvanticSequoia |
| New Round (reported) | May 2026 (reported) | In talks | $15B (reported) | Index Ventures (co-lead, reported); round not closed |
| Product | Type | Status |
|---|---|---|
| Fireworks Inference | High-throughput open-source model serving | GA |
| FireOptimizer | Custom kernel + speculative-decoding optimizations | GA |
| FireFunction | Function-calling model + framework | GA |
| F1 | Fireworks-developed reasoning model | GA (Sep 2025) |
| Fireworks Fine-Tuning + LoRA | Custom model adaptation services | GA |
| Compound AI Systems | Multi-model orchestration framework for agentic workflows | GA |
Cursor's Tab + completion latency depends on Fireworks; one of the highest-RPS AI products in the industry.
Notion AI built on Fireworks for low-latency LLM inference across collaborative documents.
High-traffic consumer + workflow AI products served via Fireworks.
Cross-data-cloud strategic positioning; Fireworks integrates with both data platforms.
Co-led the Oct 2025 Series C at a $4B valuation.
Lin Qiao (former Director of Engineering at Meta; PyTorch leader) and team start Fireworks
FireOptimizer (custom kernel-level optimizations) ships; 4 to 9x throughput improvements over open-source inference servers
Sequoia leads $52M; Fireworks is the inference layer behind multiple high-traffic AI products
Function-calling model + compound-AI system frameworks released; targets agentic workflows
Two of the highest-traffic AI products in the industry standardize on Fireworks for inference
Fireworks's first own-developed reasoning model; combines compound-AI + speculative decoding
$250M co-led by Lightspeed and Index Ventures
Reportedly in talks at a $15B valuation, nearly 4x the October mark; ARR reportedly ~$800M run-rate
Inference platform specialized in low-latency serving of open-weights models with proprietary optimizations (FireOptimizer, FireAttention). Customer mix includes AI-native startups and enterprises. Series B raised $52M at $552M valuation in 2024. Fireworks AI was founded in 2022, is headquartered in Redwood City, CA and employs ~80.
Fireworks AI has raised ~$330M+ in total at a $4B (Oct 2025 closed); ~$15B reported in talks (May 2026) valuation, based on the funding rounds Sterling tracks.
Sterling groups Fireworks AI in the AI Software sector, alongside CoreWeave (CRWV), Nebius Group (NBIS), Applied Digital (APLD), IREN (Iris Energy), Together AI (private) and Modal (private).
If you are researching Fireworks AI, also look at CoreWeave (CRWV), Nebius Group (NBIS) and Applied Digital (APLD). You can compare all of them side by side or browse the full AI Software sector on Sterling.
