Health Care’s AI Bubble Needs to Burst

Health care’s artificial intelligence (AI) boom is on a collision course with clinical reality, but it’s a necessary market recalibration, argues a new commentary published in Health Affairs on February 4, 2026, by associate professors Maxim Topaz, PhD, and Jing Dong, BS, and postdoctoral researcher Zhihong Zhang, PhD, of Columbia University, along with associate professor Laura-Maria Peltonen, PhD, of the University of Eastern Finland. 

Currently, AI startups capture 62% of digital health venture funding, but 95% of enterprise AI pilots fail to deliver returns, signaling what the authors describe as a bubble poised to burst. They argue that health care has become Silicon Valley’s unlikely lifeline to justify $250 billion in annual AI infrastructure spending despite little evidence of proven impact on patient outcomes. When AI vendors fail, hospitals and patients are left dealing with the aftermath of costly, ineffective systems, eroding trust in innovation. 

Rather than fear a downturn, the authors call for a “clinical correction” that prioritizes real-world effectiveness, governance safeguards, and financial accountability. To prepare, they urge health systems to stress-test vendors, build operational contingencies, and deploy AI intentionally to solve real problems in patient care. 

A market reset, they contend, could weed out ineffective products, elevate tools that reliably reduce clinician burden, and push capital toward innovations that notably improve care.