Two reprinted versions of the same analysis compare China's two routes for AI radiology: a disease-specific, clearance-heavy path represented by Shukun Technology (about RMB 560M revenue in 2025 and more than 70% gross margin) and United Imaging Intelligence (23 NMPA Class III clearances), versus Alibaba DAMO Academy's RADAR, a general model that flags 146 conditions across 18 organs on abdominal contrast CT, with an average AUC of 0.913 over nearly 40,000 real-world exams, open-sourced the day it appeared in Science on Sept 18. The article argues the bottleneck is payment rather than algorithms: medical insurance treats AI-assisted diagnosis as an add-on priced with the main exam and not billed separately, and 12 AI-assisted diagnosis items entered the national Class B insurance catalogue on March 31, 2026, closing off the patient-pays route. The most concrete buyers today are local finance departments, health commissions and hospitals, including a RMB 205.7M provincial 'medical AI innovation service platform' contract won by Alipay in Zhejiang and a RMB 958,000 'one-scan-multiple-screens' AI imaging purchase by a county hospital.