The platform for specialized intelligence — build, tune, and scale AI on open models
HQ: San Mateo, CA, United States | Founded: 2022 | Employees: ~200 (targeting 600 by EOY 2026) | Stage: Late Stage VC (Series D) | Website: https://fireworks.ai
Fireworks AI is an AI inference and model-serving platform headquartered in San Mateo, California, founded in October 2022 by seven engineers from Meta's PyTorch team and Google's Vertex AI. The company provides enterprises with a cloud platform to run, fine-tune, and deploy open-source large language models, vision models, and multimodal models with low latency and optimized cost. Its proprietary FireAttention inference engine delivers up to 4x higher throughput and industry-leading latency compared to open-source alternatives like vLLM.
The platform offers three deployment modes: serverless (pay-per-token, zero setup, no cold starts), on-demand (dedicated GPUs with autoscaling), and reserved capacity (guaranteed capacity with BYOC options). Fireworks serves 250+ models across text, vision, embedding, and multimodal formats, including day-zero support for new model launches. The company's customers span AI-native startups and Fortune 500 enterprises, including Uber, Shopify, Notion, Cursor, DoorDash, Samsung, GitLab, MongoDB, Elastic, and Vercel.
Fireworks has achieved remarkable growth, reaching $1B in annualized revenue by July 2026 — up 5x year-over-year — and processing over 40 trillion tokens per day. More than 95% of those tokens come from models specialized on customers' proprietary data rather than stock open models. The company raised a $1.505 billion Series D at a $17.5 billion valuation in July 2026, led by Atreides Management, Index Ventures, and TCV, with participation from NVIDIA and Lightspeed. Fireworks plans to grow headcount to 600 by the end of 2026 and deepen its partnerships with Microsoft Azure and NVIDIA.
Fireworks AI is the leading managed inference platform for open-weight models, founded by the team that built PyTorch at Meta. The company has achieved explosive growth — surpassing $1B in annualized revenue by July 2026 (up 5x YoY) and reaching a $17.5B valuation — by enabling enterprises to build, fine-tune, and serve specialized AI models at production scale. With 95% of served tokens coming from customer-customized models, Fireworks has proven that the value in open AI is not the weights but the specialization and serving infrastructure wrapped around them.
Browse: Home | Blog | About | All Startups A-Z | View full profile on StartupWiki