{"city":"Redwood City","name":"Fireworks AI","slug":"fireworks-ai","facts":[{"date":"2026-09-09","type":"revenue_band","value":"100-500m","source":"TechCrunch"},{"date":"2026-08-01","type":"funding_round","value":"Series D at approximately $17.5 billion valuation","source":"The Business Perspective, Value Add Pulse"},{"date":"2026-07-08","type":"product_launch","value":"NVIDIA Nemotron 3 Ultra became available through six hosted AI platforms","source":"NVIDIA blogs"},{"date":"2026-05-27","type":"deployment","value":"Open source equivalent of a shared S3 bucket with content hashing, existing Hugging Face token permission, and native composition with Spaces, datasets, and models","source":"Hugging Face"},{"date":"2026-05-14","type":"funding_round","value":"Series C at a $4 billion valuation","source":"The AI Insider"},{"date":"2025-12-03","type":"deployment","value":"Fireworks AI deployed Kimi K2 on NVIDIA B200 to reach the top of the Artificial Analysis leaderboard","source":"NVIDIA blogs"}],"peers":[{"name":"Baseten","rank":94,"slug":"baseten","score":"80"},{"name":"OpenRouter","rank":96,"slug":"openrouter","score":"76"}],"stage":null,"state":"California","aliases":["Fireworks"],"country":"United States","domains":[{"why":"Fireworks AI runs a hosted platform for deploying and serving foundation models such as Mistral and Kimi K2, selling inference infrastructure and APIs.","list":"AI Infra 100","rank":95,"tier":"member","layer":"Compute software","total":80,"domain":"AI Core","scores":{"gate_ne":["technology","firepower"],"leverage":8,"position":6,"progress":10,"firepower":10,"commercial":8,"deployment":8,"production":10,"technology":6},"edition":"November 2026","segment":"Inference & serving clouds"}],"founded":null,"signals":[{"date":"2026-09-26","header":"Fireworks AI considering new funding round","outlet":"36Kr newsflashes"},{"date":"2026-09-09","header":"Harvey launched its in-house Harvey Tenet legal AI model.","outlet":"TechCrunch"},{"date":"2026-08-31","header":"Fireworks Nexus provides a model-abstraction layer for enterprise-specific intelligence.","outlet":"a16z"},{"date":"2026-08-28","header":"Fireworks processes 40 trillion tokens per day.","outlet":"TechCrunch"},{"date":"2026-08-11","header":"Classmethod reduced costs by 27% in initial internal testing with NeMo Switchyard.","outlet":"NVIDIA blogs"},{"date":"2026-07-30","header":"Moonshot AI’s Kimi K3 became available to enterprise customers through Fireworks AI on Microsoft Foundry.","outlet":"TechNode"}],"website":"fireworks.ai","category":"AI Infra · Compute software","categories":[{"name":"Inference & serving clouds","domain":"AI Core","parent":"AI Infra · Compute software"}],"last_round":null,"updated_at":"2026-10-25T23:17:49.578217+00:00","facts_total":17,"positioning":"Inference and fine-tuning platform for open models, the most highly valued of the independent inference clouds after its 2026 round; runs on rented GPU capacity, no owned sites.","sub_category":"Inference & serving clouds","signals_total":23,"url":"https://www.frontiertechintelligence.com/companies/fireworks-ai","title":"Fireworks AI · ai core, Redwood City, California, United States · scores, facts and signals · Frontier Tech Intelligence","description":"Fireworks AI runs a hosted platform for deploying and serving foundation models such as Mistral and Kimi K2, selling inference infrastructure and APIs. AI Infra 100 rank 95 (score 80) 23 signals and 17 facts with sources.","publisher":"Frontier Tech Intelligence (Future Frontiers, a trade name of Essential Cognition Inc.)","note":"Assessment by Frontier Tech Intelligence; facts and signals carry their sources."}