BABA(BABA) CNBC·2025-03-24 05:10Core Insights - Ant Group is utilizing both Chinese and U.S.-made semiconductors to enhance the efficiency of its artificial intelligence models, which helps in reducing training time and costs while minimizing dependence on a single supplier like Nvidia [1][3] - The company reported a 20% reduction in computing costs by employing lower-cost hardware for training its mixture of experts (MoE) models [2] - Ant Group has announced significant upgrades to its AI solutions for healthcare, which are currently being implemented in seven major hospitals and healthcare institutions across several cities in China [4] Semiconductor Usage - Ant Group is leveraging chips from Alibaba and Huawei for AI model training, while also incorporating alternatives from Advanced Micro Devices and other Chinese manufacturers, reducing reliance on Nvidia [3] - The trend in the industry is moving towards using a mixture of networks to train AI models more efficiently [1] AI Solutions in Healthcare - The healthcare AI model developed by Ant Group is based on DeepSeek's R1 and V3 models, as well as Alibaba's Qwen and Ant's BaiLing, aimed at improving patient services and answering medical inquiries [4] - The deployment of these AI solutions is part of a broader strategy to enhance healthcare services in major Chinese cities [4] Regulatory Environment - The U.S. has imposed restrictions on China's access to advanced semiconductors, impacting the development of AI technologies within the country, although Nvidia can still sell lower-end chips to Chinese firms [5]