China tops elite AI research as insurers blame AI for $942m in added health costs

The Carnegie China study, reported by The Next Web on 28 September, found that 41% of the top AI researchers it tracked worked in China in 2025, against 34% in the US. Three years earlier the order was reversed: 46% in the US, 27% in China.

Damien Ma and Binyi Yang defined the group by papers accepted at NeurIPS 2025, calling them the field's "Navy Seals." They counted 25,677 authors across 5,823 papers and traced more than 10,000 career histories from first degree to current employer using data researchers report on OpenReview. AI coding tools helped process the data, with human checks on the results. Two numbers stand out for health technology, where AI talent is now applied to diagnostics, drug discovery and clinical trial design. First, 57% of the tracked researchers took their first degree in China in 2025, up from 46% in 2022. Second, Peking University and Tsinghua University now rank first and second as workplaces, with South Korea's KAIST third. Google held first place in 2022. ETH Zurich and the University of Oxford dropped out of the top rankings, and Europe retained 60% of its own talent, against 89% for the US.

Money and immigration policy sit behind the shift. The study points to job openings in China's AI industry and tighter US visa rules, especially for Chinese graduate students in science. The Trump administration has proposed a $103,265 fee on new H-1B visas, and Chinese AI and chip firms are handing out shares to keep engineers.

Insurers put a number on AI-driven costs

On the payer side, an analysis by the Blue Cross Blue Shield Association says hospitals' use of AI tools in insurance claims led to an additional $942 million in healthcare spending over a two-year period. TechCrunch reported the figure on 26 September. The BCBSA analysis found a sharp increase in patients documented as having complex conditions, but argued there is a clear disconnect between medical coding and treatment, with no evidence of a corresponding change in care delivered. The New York Times, cited by TechCrunch, described the analysis as the latest sign that AI is contributing to rising healthcare costs, noting that disputes between hospitals and insurers over treatments and payments are not new but appear to be getting worse as both sides use AI.

"It's not a war. It's a completely one-sided blood bath," BCBSA senior vice president Luke Chalker told TechCrunch.

Not everyone frames it that way. Dr. Shiv Rao, founder of AI startup Abridge, acknowledged to TechCrunch that the use of AI could lead to "a horrible dystopic future nobody wants to live in," with bots fighting bots and agents fighting agents, but said it might also reduce tensions and cut costs. The insurers' figure is an estimate from one industry group, not an audited national statistic, and hospitals have not accepted it.

China's health tech push

The talent shift lands in a health system that already runs on apps. Rest of World reported on 24 September that patients in China book hospital appointments through WeChat or Alipay, speak with doctors on telemedicine platforms, or chat with AI doctor avatars about conditions from insomnia to diabetes. The government has encouraged digital healthcare expansion to reach the country's 1.4 billion people and address an urban-rural gap. Ruby Wang, a physician and consultant whose book China Cure is forthcoming, told Rest of World that in the UK and US technology is always an add-on to the health system, whereas in China the digital rails are enabled enough that society treats them as normal. She also noted limits: China is strongest in oncology but less mature in immunology, and Chinese biotech firms are weak at commercialising, which is why out-licensing deals are so common.

Two developments suggest where the underlying technology is heading next. IEEE Spectrum reported on 21 September that Rhizomatica's open-sourced HERMES shortwave radio set, with its Mercury modem, sends encrypted files over the 3 to 30 megahertz band using the ionosphere instead of satellites or relay towers. Founder Peter Bloom describes reliable 400 to 600 kilometre links at about 20 watts. Omnesis, posted to Hacker News on 28 September, is a local-first index that syncs email, messages, notes, calendar, health and file data onto a user's own hardware and exposes it to agents such as Claude or ChatGPT through a policy gate.

Neither is a clinical tool. Both point the same direction as the Carnegie China numbers: the infrastructure and the data pipelines around care are being rebuilt faster than the rules and the payments around them.