Six hospitals in six years, $13.8 billion spent. The VA's new health-record plan calls for thirty hospitals a year.

The finding, in one paragraph: the Department of Veterans Affairs is this year resuming — and radically accelerating — the largest health-IT project in American history, on the strength of a schedule that requires it to do every year, five years running, five times what it accomplished in its best six years combined; with a total price that its own numbers put somewhere between $16 billion and $50 billion depending on the document; while the program's auditor reports that the department has implemented 2 of its 18 recommendations, has no approved change-management strategy of its own, and — as of the hearing where VA told Congress the new price — had still not been given that price itself.

Five documents, read directly.▸ The current assessment is GAO's VA Electronic Health Record Modernization: Critical Actions Needed to Support Accelerated System Deployments (GAO-26-108812, December 15, 2025), delivered as testimony to the House Veterans' Affairs technology subcommittee the day VA presented its restart plan. The money history is GAO's March 2025 cost-and-schedule report (GAO-25-106874). The patient-safety record is the VA Inspector General's The New Electronic Health Record's Unknown Queue Caused Multiple Events of Patient Harm (VAOIG-22-01137-204, July 2022). The ~$37 billion figure and hearing exchanges are from contemporaneous reporting on the December 15 hearing, read against VA's deployment announcements.

The program has one contract, one mission, and three prices:

The $16.1 billion figure is the one the program was justified with in January 2019.▸ The $49.8 billion figure is what the Institute for Defense Analyses calculated in October 2022 when VA finally commissioned an independent look — three times the original, more than a third of it sustainment costs the 2019 estimate barely addressed. The ~$37 billion figure is what VA told the House subcommittee in December 2025. GAO's first recommendation, open since 2021 and concurred with by VA, is to produce a reliable updated lifecycle estimate; GAO's December testimony records that "as of December 2025, the department has not provided an updated cost estimate." The auditor charged with checking the number learned it from the hearing. Meanwhile $13.84 billion — 86 percent of the original all-in price — is already obligated, with six hospitals converted.

The pause this restart ends was not budgetary.▸ It was clinical. At the first site, Mann-Grandstaff in Spokane, the new system delivered more than 11,000 orders for care into an "unknown queue" — an undetectable holding pen for orders the software couldn't match — without telling VA the queue existed. Clinicians' lab orders and referrals simply vanished. Working through 1,286 case assessments, VHA identified 149 adverse events for patients. By September 2024, per GAO's survey data, 23 percent of users agreed the system enables quality care; 13 percent said it made them efficient; 69 percent were dissatisfied overall. That is the workforce now scheduled to absorb thirty go-lives a year:

Set the restart plan against the auditor's checklist.▸ GAO's 18 recommendations are the program's repair manual: an honest cost estimate, a reliable master schedule, a VA-specific change-management strategy, an independent operational assessment of whether the system is suitable for users, and — most pointed — user-satisfaction targets established before future deployments. Sixteen of eighteen are unimplemented, including 11 of the 12 GAO designated priority. VA's organizational change management was partially consistent with seven of eight leading practices and consistent with none fully documented in a VA-approved strategy. VA concurred with all of it — concurrence dates back years — and then scheduled the acceleration anyway. The department's public case, made by its acting program chief at the December hearing, is that the system has "made significant strides in stabilizing," with new governance and a renegotiated Oracle contract carrying "additional accountability measures." That claim is not audited; the 23 percent figure is. The program's official record thus contains both assertions at once: stabilized enough to go five times faster, and too unmeasured to say what satisfied would mean.

The restart is live: the four Michigan facilities (Detroit, Ann Arbor, Battle Creek, Saginaw) went first in mid-2026, with nine more sites — Ohio, Indiana, Kentucky, Alaska — slated to bring the year to thirteen, and roughly 170 total by 2031.▸ Each go-live is a hospital switching every order, prescription, and referral to the new system in a weekend. GAO's open recommendations remain the scoreboard to watch, along with the next user-satisfaction survey and whatever lifecycle estimate finally reaches the auditor. The Oracle contract's five one-year option structure gives Congress an annual pressure point; the appropriations committees fund the program one year at a time.

All figures are from the documents cited in-line, each read directly. The ~$37 billion estimate is as reported from the December 15, 2025 hearing; GAO had not received or evaluated it as of that date. The 149 adverse events cover the unknown-queue review period (October 2020–June 2021) at initial sites.