With open-weight models rapidly closing the performance gap, enterprises are increasingly adopting a multi-model strategy, using frontier models for complex tasks while relying on open-weight models for high-volume inference. This shift is creating demand for platforms that can efficiently deploy, serve, fine-tune, and manage AI models across different clouds and GPU environments while reducing infrastructure costs and avoiding vendor lock-in.

This edition highlights three startups benefiting from this trend:

• Fireworks AI – AI inference platform optimized for serving and fine-tuning open-weight models at scale.
• Baseten – Enterprise AI deployment platform enabling production-grade inference, model serving, and optimization.
• Together AI – Full-stack AI cloud offering model APIs, inference infrastructure, and GPU cloud services for enterprise AI workloads.