Fireworks AI, a Redwood City-based provider of training and inference infrastructure for open-source AI models, has raised $1.505 billion in Series D funding led by Atreides Management, Index Ventures, and TCV. The platform optimizes open models for production use with specialized fine-tuning and high-throughput inference. The capital will expand engineering teams, global compute capacity, and partnerships with Microsoft and NVIDIA.
Inference Platforms Scale Rapidly
The timing aligns with surging demand for open-model infrastructure. Together AI raised $800 million at an $8.3 billion valuation in July 2026, while Baseten secured $300 million Series E at $5 billion in January 2026. Fireworks differentiates through PyTorch-level optimizations and compound AI support for agentic workflows.
Enterprises Seek Model Ownership
Companies face high costs and limited customization when relying on closed frontier APIs. Over 95 percent of tokens processed by Fireworks come from models specialized on proprietary customer data. This shift addresses the gap where generic models fail to capture domain-specific knowledge.
Co-Designed Training and Inference
Fireworks built a disaggregated inference engine delivering up to 4x higher throughput than competitors, with seamless transitions from fine-tuning to production. Customers report concrete gains: Notion cut latency from 2 seconds to 350 milliseconds, while Quora achieved 3x faster responses.
"General intelligence will be abundant. Specialized intelligence will be the moat." — Lin Qiao, CEO of Fireworks AI
Elite Investors Signal Conviction
Atreides Management, Index Ventures, and TCV led the round, with participation from NVIDIA, Lightspeed Venture Partners, Bessemer Venture Partners, and Menlo Ventures. TCV published a dedicated thesis on the "Inference Era," and Index highlighted Fireworks alongside its top AI bets like Anthropic. The syndicate validates the specialized intelligence thesis.
AI Inference Market Expands
The AI inference market reached $103.7 billion in 2025 and is projected to hit $312.6 billion by 2034. Fireworks processes over 40 trillion tokens daily at $1 billion ARR, up 5x year-over-year. Competitors like Together AI and Baseten have also raised hundreds of millions as capital flows into the inference layer.
Specialized Models Drive Adoption
Fireworks powers 10,000-plus companies including Cursor, Uber, Notion, DoorDash, and Shopify. Its serverless and on-demand tiers map to customer scale, while reserved capacity offers newest hardware access first. The platform supports over 400 models with day-zero support for new releases.