Ant Group has successfully used domestic chips including those from Alibaba and Huawei, combined with a mixed expert (MoE) machine learning method, to train artificial intelligence (AI) models, reducing costs by about 20%. Test results show that the technology is comparable to the performance of Nvidia's H800 chip. Although Ant Group is still using Nvidia chips for AI development, its latest models have mainly relied on alternatives including AMD and Chinese domestic chips. A research paper released by Ant Group this month showed that some of its models surpassed Meta in certain benchmarks, but the data has not been independently verified. This shows that Ant Group is accelerating the localization of AI technology to reduce costs and reduce dependence on high-end chips from the United States.