From Coinbase to Airbnb: U.S. companies switch to China’s big model to reduce costs

Against the background of pressure on the rate of return on AI capital expenditures in the United States, "switching to China's large models to reduce costs" is moving from an individual case to a trend. The two latest samples are: Coinbase switched to a Chinese model to reduce AI expenses and ease reasoning cost pressure; Airbnb used Alibaba's model (Qwen series) to support AI capabilities in some business scenarios.

From "dare not use" to "actively use"

After cases such as DoorDash, well-known U.S. stock companies such as Coinbase and Airbnb have also joined the camp of adopting the Chinese model, indicating that the demand for "cost reduction" is overwhelming "source concerns." The logic is actually very simple: on the premise that the effect is similar, the cost is significantly lower - this cost-effectiveness drive puts real pressure on the financial reports of listed companies.

Open source is an accelerator of diffusion

Dark Side of the Moon's open source Kimi K3 has received widespread attention among overseas developers and companies, and is another driver of this trend. Open source lowers the adoption threshold and the psychological cost of compliance, and also allows Chinese models to penetrate faster into the overseas developer ecosystem. The reason why Coinbase and Airbnb voted with their feet is that "similar results and lower costs" are becoming an increasingly hard standard in overseas technology selection.

Of course, the other side of the trend is geographical tension: commercial adoption is advancing in parallel with regulatory scrutiny (such as the DoorDash investigation), which means that the expansion of the Chinese model overseas is destined to find a balance between cost-effective advantages and compliance constraints. For U.S. closed-source model manufacturers such as OpenAI and Anthropic, the price-performance pressure from Chinese models may force them to re-examine their pricing systems.

Several directions worth tracking in the future:

1. Expansion of Adoption List: The penetration scale and new cases of Chinese models in American companies.
2. Regulatory Binding Level: The boundary between Congress and export control on "model adoption".
3. Overseas commercialization: The progress of API and cloud channel layout determines how far the trend can go.