The State of America's Academic Medical Centers, 2026
Explore the interactive dashboard →
Aggregate statistics for 221 U.S. academic medical centers, compiled October 2026 from public sources: reported operating margins, days cash on hand, cited credit-rating actions, and the risk and opportunity themes documented across patient care, education and research. No institution is named or ranked. Each figure states the number of AMCs it is based on.
Data freshness: 50% of records (111 of 221) rest on financial data 12 months old or newer; 71 are 13 to 24 months old, 32 are older, and 7 have no public financials. Of the 119 reported margins, 66 (55%) are from the last 12 months. Fiscal years differ by institution.
50% of reported operating margins are below 2%
Operating margin bands for the 119 AMCs with a reported operating margin (n=119; latest fiscal period available, compiled October 2026). Red bars are below break-even.
- 50% below 2%60 of 119 AMCs report an operating margin under 2% (compiled October 2026).
- 0.3% to 5.5%The middle half of reported margins (25th to 75th percentile, n=119), around a median of 1.9%.
- 23% at 6% or more27 of 119 AMCs report margins of 6% or higher (compiled October 2026).
- 22% under 100 days cash11 of the 50 AMCs with days cash reported hold less than 100 days (compiled October 2026).
By ownership type
Public AMCs: 26% report a negative operating margin (8 of 31), median days cash 123 (n=11), and 7% of rated AMCs have a cited downgrade or negative outlook (1 of 14). Private not-for-profit AMCs: 19% negative (16 of 86), median days cash 186 (n=39), and 27% with a cited downgrade or negative outlook (22 of 83). Groups are small and differences are not tested for statistical significance.
Table: n per cell as shown; compiled October 2026.
| Ownership | AMCs | Median operating margin | Negative margin | Median days cash | Ratings pressure |
|---|---|---|---|---|---|
| Private not-for-profit | 134 | 1.9% n=86 | 19% 16 of 86 | 186 n=39 | 27% 22 of 83 rated |
| Public | 77 | 1.9% n=31 | 26% 8 of 31 | 123 n=11 | 7% 1 of 14 rated |
| For-profit | 10 | too few reported (n=2) | n/a | too few reported (n=0) | too few rated (n=0) |
Medians are shown only where at least 5 AMCs report the figure. Ratings pressure counts AMCs whose cited Moody's, Fitch or S&P Global history includes a downgrade, review for downgrade or negative outlook, as a share of AMCs with any cited rating. Many public AMCs are rated through a state university or not rated separately.
By Census region
The Northeast has the lowest median reported margin (0.9%, n=33) and the West the highest (5.7%, n=19). Regional samples are small; differences are descriptive and not tested for statistical significance.
Table: n per cell as shown; compiled October 2026.
| Region | AMCs | Median operating margin | Negative margin | Median days cash | Ratings pressure |
|---|---|---|---|---|---|
| Northeast all 9 states | 50 | 0.9% n=33 | 39% 13 of 33 | 116 n=13 | 43% 9 of 21 rated |
| Midwest 11 of 12 states | 48 | 1.6% n=27 | 4% 1 of 27 | 157 n=12 | 14% 3 of 22 rated |
| South 16 states and DC | 83 | 4.2% n=40 | 15% 6 of 40 | 200 n=18 | 17% 7 of 41 rated |
| West 9 of 13 states | 40 | 5.7% n=19 | 21% 4 of 19 | 247 n=7 | 31% 4 of 13 rated |
U.S. Census Bureau regions; DC is counted in the South. State counts show how many of each region's states have at least one AMC in this set.
Patient care dominates the documented risks and opportunities
Share of the 221 AMCs (n=221, compiled October 2026) with at least one analyst-documented risk or opportunity matched to each mission by keyword rules. Because clinical margins help fund teaching and research, pressure on patient care can spill over to the other missions. 28 AMCs (13%) have NIH or research-funding exposure documented as a risk.
The most frequently documented risks and opportunities
735 risks and 726 opportunities documented by our analysts from public sources (compiled October 2026), classified into themes with keyword rules. Each bar is the share of the 221 AMCs with at least one item in that theme; an item can fall into more than one theme.
Risksn=221 AMCs
Opportunitiesn=221 AMCs
These items are analyst assessments, not statements by the institutions. Unclassified items: 104 risks and 111 opportunities did not match any theme rule. Notes that only flag missing data are excluded.
Workforce and burnout
Burnout is measured in national surveys as a share of clinicians reporting symptoms, and it is studied as an operating cost through turnover and reduced clinical hours. The national figures below come from the most recent editions available in October 2026. Physician burnout has fallen for three consecutive years in the AMA series but remains above 40%; hospital RN turnover rose in 2025 after falling in 2024.
Prevalence and trend
- Residents and fellows: 28.6% reported burnout symptoms in 2025, down 5.9 points from the prior year (more than 3,000 trainees, 29 organizations).2
- Among nurses who intend to leave for reasons other than retirement, 41.5% named stress and burnout; more than 138,000 nurses have left the workforce since 2022.3
- Hospital RN turnover was 17.6% in 2025, up 1.2 points; first-year RN turnover was 22.7%.4
- The direction is mixed: physician and trainee burnout rates are declining in the AMA series, while nurse turnover and stated intent to leave remain high. These surveys use different instruments and samples, so the rates are not directly comparable.
Workload drivers and cost
- EHR time: among 141 academic primary care physicians, EHR time per 8 hours of scheduled patient time rose 28.4 minutes (7.8%) from 2019 to 2023; inbox time rose 24.4%, and work outside scheduled hours rose 8.2% on clinic days and 19.9% on non-clinic days.6
- Inbox: across 280,712 ambulatory physicians, patient advice messages rose in March 2020 and stayed elevated through April 2022; primary care physicians averaged 16 patient messages and 24 patient calls a week.7
- Prior authorization: 94% of surveyed physicians said it contributes to burnout, and 26% reported a serious adverse event they linked to it (self-reported).5
- Cost: a 2019 model estimated about $4.6 billion a year in U.S. costs from physician turnover and reduced clinical hours attributable to burnout (range $2.6 billion to $6.3 billion), about $7,600 per employed physician per year. It is a modeled estimate from 2019 inputs, not a measured outcome.8
- NSI estimates each 1-point change in RN turnover costs or saves the average hospital about $295,000 a year.4
What our AMC data shows
Share of the 221 AMCs whose public record mentions each workforce signal. Records were classified with keyword rules over the people, finance, risk and transaction notes; event signals are excluded when they cite only years before 2024 or are phrased as a possibility; undated mentions and announced 2027 successions are included. An AMC can carry more than one signal.
| Ownership | AMCs | Leadership turnover | Layoffs | Union or contract dispute | Agency labor | Strike |
|---|---|---|---|---|---|---|
| Private not-for-profit | 134 | 40% 53 of 134 | 22% 29 of 134 | 7% 10 of 134 | 3% 4 of 134 | 2% 3 of 134 |
| Public (state/university) | 41 | 39% 16 of 41 | 12% 5 of 41 | 17% 7 of 41 | 2% 1 of 41 | 5% 2 of 41 |
| Public (county/city/district/authority) | 36 | 33% 12 of 36 | 14% 5 of 36 | 8% 3 of 36 | 17% 6 of 36 | 0% 0 of 36 |
| For-profit (investor-owned) | 10 | 50% 5 of 10 | 0% 0 of 10 | 10% 1 of 10 | 0% 0 of 10 | 0% 0 of 10 |
Independence and limits. These counts come only from public reporting (press coverage, filings and system websites) compiled through October 2026, not from the institutions' own workforce data. A missing flag does not mean the event did not occur, so most shares are likely undercounts, especially for agency-labor spending and union activity, which are rarely disclosed. Leadership turnover is reported here as a workforce signal; this dataset cannot show whether it is related to burnout. Ownership groups of 10 to 41 AMCs are small, so treat differences between them as directional. A few flags reflect forward-looking risk items or events at a parent system, university or health plan rather than the hospital itself.
What the evidence on AI documentation shows so far
- A five-system observational study of 8,581 ambulatory clinicians (2023 to 2025) found AI scribe adoption associated with 16.0 fewer documentation minutes per 8 scheduled patient hours (about 10%) and 13.4 fewer total EHR minutes (about 3%). The change in after-hours EHR work was not statistically significant. The authors describe the effects as modest.9
- A randomized trial of 238 physicians at one academic center found one AI scribe reduced time per note by about 10% while the other showed no statistically significant change; both were associated with about a 7% improvement in burnout scores. Physicians reported occasional clinically significant note inaccuracies, and the authors call for larger, longer studies.10
Reading the evidence
- Survey rates are self-reported and depend on the instrument; the AMA figures come from organizations that chose to take part in its survey program.
- The cost estimates are modeled (physician) or based on a hospital benchmarking survey (nursing); they are not audited results at any single AMC.
- The studies above report associations between workload, burnout and turnover; most do not establish that one causes the other.
Where agentic AI may help with workforce strain
The workload items above are where agentic AI could return clinician hours. Early results are modest and mixed, so each use would need local measurement against a baseline. No dollar savings are assumed.
- Ambient documentation and note draftingDraft notes for clinician review may reduce documentation time; published effects so far are about 10% of note time, with after-hours effects not yet shown.
- Inbox triageAgents could sort patient messages, route non-clinical requests to staff and pre-draft replies for clinician sign-off, which may reduce inbox minutes.
- Prior-authorization packetsAgents could assemble payer-specific documentation and track status, which may reduce part of the reported 13 hours a week, with a person approving every submission.
- Scheduling and staffingDemand forecasts and internal float-pool matching could help fill shifts before agency, which may reduce premium labor and schedule churn.
- GME duty hours and education administrationDuty-hour monitoring, rotation scheduling and evaluation follow-up could take routine tasks off program coordinators and trainees.
- American Medical Association, "AMA: Physician burnout rates are falling, specialty gaps remain," April 16, 2026. ama-assn.org; 2024 and 2022 rates: "Burnout eases for doctors at every career stage as support rises," July 22, 2025.
- American Medical Association, "Latest data shows 28.6% burnout rate among resident physicians," April 23, 2026. ama-assn.org
- National Council of State Boards of Nursing, "NCSBN Research Highlights Small Steps Toward Nursing Workforce Recovery; Burnout and Staffing Challenges Persist" (2024 National Nursing Workforce Study), April 17, 2025. ncsbn.org
- NSI Nursing Solutions, "2026 NSI National Health Care Retention & RN Staffing Report" (calendar 2025 data), March 2026. nsinursingsolutions.com (PDF)
- American Medical Association, "AMA survey: Prior authorization reform pledge falls short for physicians," May 13, 2026. ama-assn.org
- Arndt BG, Micek MA, Rule A, et al. "More Tethered to the EHR: EHR Workload Trends Among Academic Primary Care Physicians, 2019-2023." Annals of Family Medicine, 2024;22(1). annfammed.org
- Holmgren AJ, Apathy NC, Sinsky CA, et al. "Trends in Physician Electronic Health Record Time and Message Volume." JAMA Internal Medicine, 2025;185(4). jamanetwork.com
- Han S, Shanafelt TD, Sinsky CA, et al. "Estimating the Attributable Cost of Physician Burnout in the United States." Annals of Internal Medicine, 2019;170(11). acpjournals.org
- Rotenstein LS, Holmgren AJ, Thombley R, et al. "Changes in Clinician Time Expenditure and Visit Quantity With Adoption of Artificial Intelligence-Powered Scribes: A Multisite Study." JAMA, 2026;335(16). jamanetwork.com
- UCLA Health, "UCLA study finds AI scribes may reduce documentation time and improve physician well-being," November 26, 2025, reporting a randomized trial published in NEJM AI (doi:10.1056/AIoa2501000). uclahealth.org
AMC counts: Elaria analysis of 221 AMC records, keyword rules over public reporting, compiled October 2026. Figures checked against each source page in October 2026.
How this report was built
- Scope. 221 U.S. academic medical centers in 45 states and DC, compiled October 2026: hospitals and health systems serving as the principal clinical partner of a medical school. This is a compiled set, not a complete census of U.S. academic medical centers.
- Sources. Public information only: health-system leadership and financial-disclosure pages, rating-agency notices (Moody's, Fitch, S&P Global), audited financial statements and bond disclosures, IRS Form 990 filings, SEC 10-K filings, and healthcare trade press.
- Figures. Every figure on this page is computed by script from the compiled dataset. Missing values are excluded, and each figure states the number of AMCs it is based on. Percentages are rounded to whole numbers and margins to one decimal, with halves rounded up.
- Limits. Fiscal years and reporting bases differ (system versus hospital, audited versus interim), and some margins are quarterly or derived from Form 990 totals. 50% of records use financial data 12 months old or newer. Ratings history reflects the most recent action cited in public sources and may not reflect current status.
- Themes. Risks and opportunities are free text written by our analysts from public sources, then classified into themes with keyword rules; counts are indicative rather than exact.
- Not advice. This report is for general information. It is not investment, credit, legal or financial advice. AI-assisted research; verify before relying.
Where agentic AI may help
Elaria Agentic, the publisher of this report, advises health-system executives on agentic AI. This box lists possible applications mapped to themes in this report, grouped by People, Product and Process. It does not report results: whether any of these would help depends on each institution's data, systems, staffing and governance, and any agent would work with a human approving consequential actions. Theme shares are our analysts' assessments (compiled October 2026), not the institutions' stated priorities.
Labor is among the largest cost lines. Agents could take some routine work off clinicians and staff rather than adding headcount.
- Staffing and agency-labor agentCould forecast census-driven demand, suggest internal float-pool coverage before agency labor, and flag premium-pay drift for review.Workforce flagged as an opportunity by our analysts at 32% of AMCs (71 of 221); labor cost flagged as a risk by our analysts at 31% of AMCs (69 of 221).
- Administrative workload reliefCould help with inbox triage, prior-auth packet assembly and documentation drafts, with licensed staff reviewing the output.Cost-out and administrative automation flagged as an opportunity by our analysts at 38% of AMCs (85 of 221).
- Leadership-transition supportCould assemble a KPI baseline, board-packet drafts and an open-decision log for an incoming executive.Leadership transition flagged as a risk by our analysts at 25% of AMCs (55 of 221).
For an AMC the product is the three missions. Education and research administration are possible areas for agent support.
- GME and education administration agentCould support rotation scheduling, duty-hour monitoring, evaluation follow-up and accreditation-report assembly for program coordinators.GME and education administration flagged as an opportunity by our analysts at 18% of AMCs (39 of 221).
- Research administration and grants agentCould support pre-award checklists, effort-reporting reconciliation, sponsor-rule monitoring and post-award close-out.Research administration flagged as an opportunity by our analysts at 17% of AMCs (37 of 221); research funding flagged as a risk by our analysts at 16% of AMCs (35 of 221).
- Care management and value-based agentCould prepare care-gap and outreach lists and assemble utilization-management reviews for staff to complete.Care management and value-based programs flagged as an opportunity by our analysts at 11% of AMCs (24 of 221).
Process fixes are often the quickest to show margin impact. These themes were flagged most often in the dataset.
- Revenue cycle, denials and prior-auth agentsCould support eligibility checks, prior-auth submission, denial root-cause coding and first-draft appeals, with a human approving every submission.The opportunity our analysts flagged most often: 86% of AMCs (190 of 221).
- Throughput and capacity agentCould prepare discharge-readiness lists, triage transfer-center requests and suggest OR block releases for staff review.Throughput, capacity and access flagged as an opportunity by our analysts at 65% of AMCs (143 of 221).
- Supply chain, pharmacy and 340B agentCould flag price variance, check contract compliance and support 340B eligibility audits across sites.Supply chain, pharmacy and 340B flagged as an opportunity by our analysts at 20% of AMCs (45 of 221).
- Integration and shared-services playbookCould help apply common revenue-cycle, supply and reporting processes at newly affiliated hospitals.Integration and operating model flagged as an opportunity by our analysts at 39% of AMCs (87 of 221).
A private briefing could benchmark an institution's margins, liquidity and operating themes against these aggregates and identify the 3 agentic plays worth evaluating first.
Request a briefingGet the executive briefing
Occasional executive briefings on agentic AI, academic medical center finance and health policy. One field, one click.