China’s Low-Cost AI Models Challenge U.S. Lead as Open-Weight Releases Gain Ground

Chinese AI developers are widening the reach of low-cost, high-capability models, intensifying the contest with U.S. labs over whether frontier AI advantage depends most on compute scale, model efficiency or distribution. Recent releases from DeepSeek, Alibaba’s Qwen team and Moonshot AI show how Chinese firms are using open-weight models, aggressive pricing and hardware-aware engineering to compete despite U.S. restrictions on advanced chips.