China’s medical AI battle shifts from changing doctors to courting patients

Ant’s big consumer gamble

Ant Group (蚂蚁集团) is pushing medical AI straight into people’s pockets. Its Afu app (阿福) — positioned around elderly checkups and chronic‑disease management — ran ubiquitous ads, even appearing on the Spring Festival Gala, and reportedly spent "a few hundred million yuan" on marketing in December alone. It has been reported that Afu topped 100 million users in late January, with more than half of new Lunar‑New‑Year registrants coming from third‑tier and lower cities. Why the consumer push? After a decade of hard, expensive work trying to change physician behaviour, a growing number of Chinese firms now believe it may be easier to change patient pathways instead.

Policy and data are the new infrastructure

The timing is no accident. In October 2025 China’s National Health Commission set targets for “AI + healthcare” applications and for building trusted health data spaces; in January 2026 the National Medical Security Administration launched pilots for a “personal medical cloud” to pool diagnostics, treatment and health‑consumption data for all 1.3 billion insured residents. Those government moves turn data into national infrastructure that no single company could assemble alone. Data desensitization and authorization remain technical challenges, but they are solvable — and with domestic data flows becoming strategically vital amid U.S.–China tech competition and export controls, Chinese tech firms are racing to lock down home‑market advantage.

A new round of winners and losers

Big‑tech entrants such as Baichuan Smart (百川智能) and Ant are now directly courting consumers; it has been reported that Ant aggressively recruited talent from rivals as it scaled the Afu push. Hospitals are watching nervously: will patients choose a provider recommended by an app? The industry remembers an earlier decade when medical AI focused on B2B tools — image readers, screening aids and hospital systems — and found that penetrating hospital walls is costly and unrewarding. Regulatory approvals and a growing catalog of certified AI medical software (more than 110 products approved by October 2025) show technical progress, but the central question remains commercial: can firms turn consumer engagement into sustainable revenue, or will the “do‑good” halo of health products again collide with thin business models? The next few years will tell whether this patient‑first turn produces a durable market or another round of dashed hopes.