Particle.news

Download on the App Store

Ant Group Achieves AI Breakthrough with Cost-Effective Chinese and U.S. Chips

The Alibaba-backed firm reduces AI training costs by 20% while matching Nvidia's performance, expanding applications in healthcare across seven major Chinese hospitals.

Ant Group headquarters
The results attained were even similar to those from Nvidia Corp. chips like the H800.
Image
Image

Overview

  • Ant Group has reduced AI training costs by 20% by using a mix of Chinese-made chips from Alibaba and Huawei, alongside U.S. chips from AMD.
  • The company's AI models, Ling-Plus and Ling-Lite, achieve performance comparable to Nvidia's H800 GPUs, showcasing China's growing AI capabilities.
  • The Mixture of Experts (MoE) technique underpins Ant's cost-efficient AI training, enabling high performance with less powerful hardware.
  • Ant's AI advancements are being applied in healthcare, with its models deployed across seven major hospitals in China to improve patient services.
  • These developments highlight China's push for technological independence and innovation in AI, navigating U.S. export restrictions on advanced semiconductors.