Summary
Large companies are increasingly building on Chinese AI models, and this video explains why. Nick Saraev points to Coinbase, Airbnb, DoorDash, Cursor, Ecosia and Lindy as examples of businesses running on models such as Alibaba’s Qwen and Moonshot’s Kimi, with Coinbase reportedly cutting its ad bill by roughly half.
The video cites marketplace data to show the scale of the shift. On OpenRouter, American models have fallen from about 70% of traffic a year ago to around 30%, Chinese models took all five top token-volume spots in July, and Qwen has passed 2 billion Hugging Face downloads compared with roughly 230 million for Meta. More than 150,000 models have been fine-tuned from Qwen alone.
Three main drivers are discussed. The first is price: Kimi K3.1 Pro costs about $2 per million input tokens, against much higher rates from Anthropic and OpenAI. The second is that open-weight models are good enough for many coding and automation tasks. The third is privacy, since companies can download the weights, run them on their own hardware and fine-tune them without sending proprietary data to an outside lab. Moonshot says the K3.1 Pro weights arrive on Hugging Face on November 4.
📺 Source: Nick Saraev · Published October 31, 2026
🏷️ Format: Opinion Editorial
