← 返回 Avalaches

中國人工智慧公司在面臨美國高階晶片出口管制的情況下,轉向開發高效率且資源消耗較少的模型,例如透過「專家混合」技術讓 DeepSeek 和阿里巴巴等模型在運算力有限的條件下,仍能達到與美國領先模型相近的效能。

中國的 AI 發展策略側重於開源權重模型及廣泛的產業應用,政府不僅補貼基礎設施,還積極推動 AI 在製造、物流與電子商務等領域的落地,旨在提升國家整體經濟生產力,而非單純追求個別企業的高額利潤。

儘管面臨歐美市場的資安疑慮與限制,中國的 AI 模型憑藉低成本和高靈活性的優勢,已開始在全球市場擴展市占率,這對依賴高投入、高收費商業模式的美國 AI 巨頭構成了潛在的定價與競爭壓力。



Facing US export controls on advanced chips, Chinese AI companies have pivoted to developing highly efficient models that require less computing power, utilizing techniques like mixture-of-experts to allow platforms such as DeepSeek and Alibaba to perform comparably to leading American models.

China's AI strategy emphasizes open-weight models and broad industrial application, with the government subsidizing infrastructure and promoting the integration of AI into manufacturing, logistics, and e-commerce to boost national economic productivity rather than simply maximizing individual corporate profits.

Despite security concerns and restrictions in Western markets, Chinese AI models are expanding their global market share, leveraging their low cost and flexibility to pose a potential pricing and competitive threat to the high-investment, premium-priced business models of US AI giants.
2026-08-19 (Wednesday) · af6760a19a1799705a4181840fd76c205bdfa962