ChinaTalk Why it Sucks to Work in AI in China + Open Source with Kevin Xu
68 snips
Mar 12, 2026 
Kevin Xu, writer and analyst behind the Interconnected newsletter, explores why building AI in China is tougher: constrained compute, limited upside, and political pressure. He traces China’s open source history and explains how open sourcing is a key route for Chinese AI to reach foreign users. They also discuss Alibaba’s Qwen shakeup and why Chinese firms may lead in AI-driven shopping and commerce.
AI Snips
Chapters
Transcript
Episode notes
Chinese AI Jobs Have Much Lower Upside
- Working in Chinese AI offers far less upside and far more pressure than the West due to orders-of-magnitude less access to compute and smaller rewards.
- Kevin Xu and Jordan Schneider contrast US labs like OpenAI with Chinese labs facing tighter resources, corporate constraints, and weaker enterprise markets.
Open Source Is The Overseas Breakout Channel
- Open source is one of the few practical routes for Chinese AI firms to reach overseas users despite geopolitical barriers.
- Kevin Xu argues open sourcing lets foreign developers evaluate models on merit before commercial ties form, bypassing the China label.
Open Source Requires Western-Style Governance
- Sustaining open source projects requires liberal, democratic governance practices that mirror Western foundations.
- Kevin Xu notes Chinese contributors are adopting technical committees, roadmaps, and community engagement similar to Apache or CNCF projects.