American health care has a workforce problem — and it is driving up costs.

When nurse practitioners cannot practice independently, patients who need a routine primary care visit end up paying for more expensive care. When prescription renewals require a physician visit that an AI tool could handle, the system is more expensive, for no clinical benefit. The United States has spent decades restricting the supply of health care providers in the name of quality, and the result has been higher prices, worse distribution, and a workforce that isn’t organized around the needs of patients.

Two changes would go a long way toward fixing that: rethinking who is allowed to do what, and expanding the supply of providers at every level.

Rethinking who is allowed to do what

American health care has largely refused to ask a key question: What level of training does a specific task actually require?

Managing a stable patient’s diabetes medications does not require the same skills as diagnosing a rare autoimmune condition. Performing a routine colonoscopy does not require the same training as performing cardiac surgery. A system that insists otherwise is wasteful.

The evidence is unambiguous. Nurse practitioners deliver care comparable in quality to physicians for many common conditions such as chronic disease management. Pharmacists, with years of post-graduate training focused precisely on medications already manage medication therapy, order labs, and treat common infections in many states — often through cooperative agreements with physicians. Each tier of this workforce is substantially cheaper than the one above — primary care physicians earn around $287,000 annually, nurse practitioners and pharmacists roughly half that — potentially generating substantial reductions in the cost of delivering care. Meanwhile, many of the tasks a registered nurse does can be easily managed by licensed practical nurses, medical assistants, and trained aides. Allowing registered nurses to focus on assessment, judgment, and complex care, while support roles absorb routine tasks, would reduce costs and make better use of a workforce in short supply.

Twenty-three states still impose physician supervision requirements on nurse practitioners, with no basis in evidence. Congress should establish a national minimum standard giving nurse practitioners the legal authority to diagnose, treat, and prescribe for a specific set of defined conditions without physician oversight — in every state, not just the ones that have already acted.

Some other countries take a more practical approach: The Netherlands ran a five-year national experiment with nurse practitioners and physician assistants independently performing endoscopies and minor surgical procedures, evaluated with input from the Dutch Royal Medical Association, and made independent nurse practitioner and physician assistant practice permanent law in 2018. The United Kingdom gave nurses full prescribing authority nearly two decades ago. Thirteen European countries now have nurse prescribing laws.

The United States should also identify technically straightforward and high-volume procedures, train non-physicians on validated surgical simulators for those procedures, and allow those who meet objective benchmarks to perform them, measuring outcomes to ensure the care is both high quality and safe. AI can now assess surgical skill from video footage with accuracy, making objective competency assessment scalable in ways that weren’t possible before. The question of which procedures can be safely delegated to well-trained, simulation-validated non-physicians is an empirical one and can be answered with high-quality studies, rather than assuming the answer is none. Aravind Eye Hospital in India built one of the world’s highest-volume, highest-quality cataract services program on a single principle: No one should do a job that someone with less training could do just as well. That principle is unremarkable in so many industries — but not in health care.

Technology has opened up new opportunities here. The average physician spends less than half of a 58-hour workweek with patients. The rest disappears into documentation, prior authorization, and administrative tasks. AI scribes, software that listens to patient visits and automatically generates clinical notes, are reclaiming that time in current real-world deployments. AI-assisted clinical decision support tools have reduced diagnostic errors by 16 percent in primary care settings. In January, Utah became the first state to authorize an AI platform to autonomously renew routine chronic medications — at $4 per renewal, with rigorous safety monitoring. Prescription renewals constitute roughly 80 percent of all medication activity. The cost savings implications are substantial.

But realizing those savings requires getting the payment environment right. Under fee-for-service payment, if AI makes a service more efficient but reimbursement stays the same, providers profit, while the system saves nothing. If reimbursement falls to reflect the efficiency, the incentive to adopt disappears. The only model that resolves this dilemma is value-based care, in which health systems accountable for total cost of care capture every efficiency gain as dollars saved: the AI prescription renewal or the nurse practitioner managing chronic disease follow-ups at half the physician cost. With value-based care, these reforms could lower costs for society.

Expanding the supply of providers

The United States faces a shortage of up to 86,000 physicians by 2036, with an aging population placing growing demands on a workforce already stretched thin. We need more providers at every level. Congress capped Medicare-funded residency slots in 1997, with only modest changes since. In 2025, nearly 47,000 medical school graduates applied for fewer than 38,000 positions, preventing qualified medical school graduates from completing their training. The bipartisan Resident Physician Shortage Reduction Act would add 14,000 slots over seven years. Congress has failed to pass it for years — it should finally act.

For nurses, the bottleneck is school capacity. In 2024-2025 alone, more than 80,000 qualified nursing school applicants were turned away due to limited faculty and clinical placement spots. Investment in faculty, simulation labs, and clinical placements — accelerated through health system partnerships with nursing schools — would expand the nursing pipeline.

The physician and nursing shortages are largely a consequence of bad policies. The tools to fix this problem exist. The rest is a choice.