Chinese AI rivalry threatens US margins, not its ecosystem

key takeaways.

• Chinese AI models are narrowing the performance gap with US rivals, increasing competition and putting pressure on prices
• The leading US models still outperform on the hardest tasks, and many are competitive on cost-per-task, rather than simply price-per-token measures
• Cheaper, open-weight, locally-deployed models may squeeze the margins of leading AI labs, but should accelerate adoption and expand demand across the value chain
• Cloud providers, or hyperscalers, and semiconductor firms are more likely to capture AI upside than pure-play model providers. Their infrastructure, data and distribution advantages should become more valuable as AI adoption accelerates and intelligence becomes cheaper.