凭借远低于美国同类产品的成本，中国大模型企业正在吸引美国企业和初创公司大规模转向使用。根据OpenRouter数据，今年中国AI模型的整体使用量已超过美国同类产品[1]。开源的中国模型相比Anthropic和OpenAI的解决方案便宜60%到90%[1]，这种价格优势正驱动美国企业做出实质性转变。技术公司Lindy已将业务从Anthropic迁移至DeepSeek，预计在数月内节省数百万美元[1]；加密货币交易所Coinbase表示正在转向使用中国人工智能模型以降低成本[1]；出行平台爱彼迎则采用了阿里巴巴的Qwen模型，并赞其"快速且便宜"[1]。
这一趋势激化了中美科技政策的对立。美国财政部长贝森特威胁采取制裁和"实体清单"措施应对中国企业[1]。然而，美国政府遭遇来自产业端的集体反对。7月24日，近200家美国科技初创企业敦促政府不要切断对中国开源模型的访问[1]，英伟达首席执行官黄仁勋也表示"美国不应该禁止这类大模型"[1]，认为限制措施将削弱美国企业竞争力。
Chinese artificial intelligence companies including Moonshot AI and DeepSeek have captured significant market share among American businesses by offering models priced 60 to 90 percent cheaper than their U.S. counterparts [1]. According to OpenRouter data, Chinese AI models have already surpassed American equivalents in overall usage volume this year [1]. Companies including Lindy, Coinbase, and Airbnb have begun transitioning to Chinese alternatives to reduce expenses, with Lindy alone projecting savings of millions of dollars over the coming months [1]. Airbnb adopted Alibaba's Qwen model, describing it as both "fast and inexpensive" [1].
The cost advantage has prompted a policy response from U.S. authorities. Treasury Secretary Bessen threatened to deploy sanctions and add Chinese companies to an "entity list" in response to model distillation practices [1]. However, this stance faces significant opposition within the American tech sector. On July 24, a coalition of nearly 200 U.S. technology startups released a joint statement urging the government not to restrict access to Chinese open-source models [1]. Nvidia CEO Jensen Huang stated that "the United States should not ban these types of large models," warning that such restrictions would undermine American competitiveness [1].