Ping An Insurance (Group) Company of China said its chief technology officer, Ray Wang, was invited to speak at the 2026 Apsara Conference, where he set out the group's experience of deploying AI at scale. Mr Wang said the central question for enterprises is how to put AI to work in real business scenarios at a manageable cost while creating measurable value, the company said.
Over the past year the group's computing capacity rose from 800 PF to 1,500 PF, while average daily token consumption climbed from 30 billion at the end of last year to more than 300 billion. Mr Wang said that by roughly doubling computing capacity the group supported a tenfold rise in token consumption while keeping costs under control. Ping An said token output per unit of computing power rose 368% and cost per token fell 76%.
The group's AI-powered Express Service, launched four months ago, now supports more than 300 commonly used services covering transactions, financing and claims, and has recorded over 100 million user interactions. Ping An said the service reaches an end-to-end resolution rate of 92% and has cut average processing time by more than 50% compared with traditional service models.
Ping An said its AI Doctor supports the full healthcare journey, from consultations and multidisciplinary diagnosis and treatment to ongoing health management, drawing on medical resources from PKU Healthcare Group and work with institutions including PUMC Innovation Port. In multidisciplinary diagnosis and treatment for breast and gastric cancer, AI assessments matched expert conclusions in 91% of cases, and the AI hallucination rate stayed below 0.3%. The group said AI Doctor is the first point of contact for 25% of in-hospital services and that its annual service volume is expected to exceed 100 million by the end of the year.
Inside the group, Ping An said AI-generated code accounts for about 80% of total code output and that around 20% of routine development tasks can now be delivered end to end by AI. AI service representatives handled 81% of customer service volume in the first half of the year, quick claims accounted for 59% of life insurance claims, and Ping An P&C recorded RMB7.11 billion in claims savings through AI-powered fraud detection. Mr Wang said highly specialised sectors such as finance and healthcare require domain-specific intelligence, and that the group has built a framework combining proprietary data, expert-defined standards, production-grade quality controls and feedback from real business scenarios.