Nearly two hundred US tech startups have formed the Little Tech Association alliance and jointly written to the Trump administration and Commerce Secretary Howard Lutnick, opposing a ban on Chinese open-weight AI models. The backdrop is that recent Chinese models are closing the performance gap with Anthropic Fable 5 and OpenAI GPT-5.6 Sol, and within the US government there is now talk of a ban. The startups argue that a ban would directly weaken the small and mid-sized US teams that rely on open weights to iterate quickly, and it would not stop the global spread of open weights anyway. The real lead isn't restricting what competitors can use — it's whether the US itself can keep shipping world-class open-weight models and ensure domestic teams can freely tap into the best open-source foundation anywhere in the world. This joint letter sends a clear signal: at a moment when the frontier gap is shrinking, the real moat of US AI competitiveness is open ecosystems and innovation density, not export controls. If regulation forces a hard cut, the first ones hurt are the largest group — the startups that depend most on open source. The letter also echoes recent moves like Anthropic removing the Claude Code anti-distillation backdoor — AI governance is shifting from "confrontational lockdown" to "transparent collaboration".