Definition
Cloud platforms that host, fine-tune, and serve open-weight-models for developers and enterprises as an alternative to closed model APIs from openai, anthropic, and google.
Key Points
- Demand thesis: cheaper tokens + customization vs closed APIs
- fireworks-ai Jul 2026: 17.5B valuation
- Competitive set includes hyperscalers and neocloud GPU providers
- Upstream model releases (e.g. kimi-k3) expand addressable inference volume once weights ship