Walmart Is a Top-3 Medicaid Employer in 6 States

Introduction

A federal audit released on July 22 puts Walmart among the top three employers of Medicaid enrollees in all six states where investigators could get usable data, and Amazon in the top three in five of those six. H.R. 1, the budget law Congress passed a year before that audit landed, tightens who is allowed to keep that coverage, with its Medicaid work requirement starting January 1, 2027. Nothing in the law changed the rule that decides whether an employer has to offer you health insurance in the first place. That rule is 30 hours a week, so if your schedule runs 28, your employer owes you nothing and the government now wants documentation that you're working.

What GAO Could and Couldn't Measure

The report is GAO-26-108703, 46 pages, requested by Vermont Senator Bernie Sanders as ranking member of the Senate HELP committee and released publicly on July 22. The employer names come from 15 state agencies across 11 states, the only ones that could hand over reliable employer-of-record data — CNN put those states' combined population at roughly a fifth of the U.S. total. Six of those states had usable Medicaid data, nine usable SNAP data.

Walmart ranked first among employers of Medicaid enrollees in Georgia (4,595 workers), Indiana (3,844) and Oklahoma (4,684), second in Maine, third in Massachusetts and Rhode Island. Amazon came in second in Georgia, Massachusetts, Oklahoma and Rhode Island, third in Indiana, and doesn't appear anywhere in Maine's top 25. On the SNAP side, Walmart lands in the top six in all nine states with usable data, Amazon in the top four in six of them. Add the six Medicaid tables together and you get 16,055 Walmart workers and 11,338 Amazon workers enrolled across those states, plus 15,515 and 12,346 on SNAP across the nine. Those are sample totals; GAO's transmittal letter says the estimates "are not generalizable to all states."

States supplied employer of record to GAO, not necessarily current employer, and the names were messy enough that every count carries a confidence interval. The largest caveat is size: GAO didn't adjust for company size, so the biggest employers are likelier to top any list than smaller ones, wages and industry held equal. In Table 5, 35.2% of wage-earning Medicaid enrollees and 41.5% of wage-earning SNAP recipients work for employers with 1,000 or more employees, against 48.2% and 47.4% of everyone else, meaning benefit recipients are less concentrated at big companies than the general workforce, not more.

80 Hours a Month, 30 Hours a Week

H.R. 1's Medicaid work requirement, as KFF's breakdown of the provisions lays it out, makes adults in the expansion group document 80 hours a month of work, school, volunteering or job training, at application and at every renewal, starting January 1, 2027. Eighty hours a month is a little over 18 hours a week. The ACA's employer mandate, the rule that obligates a large employer to offer coverage at all, only reaches workers averaging 30 hours a week. Congress moved one of those numbers last year and left the other alone.

KFF's March analysis of working adults on Medicaid found 65% are either at an employer offering no coverage (52%) or offered coverage they don't qualify for (13%), and 69% of that ineligible group blame insufficient hours. Only 21% of part-time Medicaid workers are eligible for their employer's plan, against 42% of full-timers. So you can clear 80 hours a month, keep your Medicaid, and still sit far enough under 30 hours a week that your employer never owes you a plan. What gets me is that both thresholds were Congress's to set.

That work requirement is the single largest piece of the law's Medicaid savings, $326 billion of roughly $911 billion over ten years by CBO's estimate. SNAP moved too: the work-requirement age ceiling rose from 54 to 64, the dependent-child exemption dropped from under 18 to under 14, and veterans, homeless adults, and young adults aged out of the child welfare system lost their automatic exemptions.

Oregon's Intake Forms Name the Same Companies

Oregon isn't one of GAO's six Medicaid states, which makes its own paperwork a useful check. The Oregon Capital Chronicle asked the Oregon Health Authority for the 500 employers listed most often on Medicaid income-verification forms filed during 2025; those companies accounted for more than 118,000 listings (not unique people, per the paper's same-day clarification). Walmart was the single most-listed employer, Safeway/Albertsons close behind, Amazon/Whole Foods fourth, Fred Meyer/Kroger fifth. Kroger is a real name in this data and a smaller one than the shorthand suggests: it turns up in 6 of GAO's 15 state tables, never above fifth.

The same reporter ran the SNAP version in May, with Safeway/Albertsons first, Walmart second and Amazon third among 1,000 companies, and found Oregon State University itself listed more than 2,300 times, mostly part-time student workers: the same hours story with a public university's name on it.

Amanda Dalton of the Northwest Grocery Retail Association told the Capital Chronicle that many grocery employees "work part-time or seasonally" and belong to unions with health benefits, so "SNAP participation should not be viewed as a simple measure of employer wage levels alone." She's right on the merits, and she's describing the mechanism: part-time and seasonal is the population the 30-hour rule was drawn to exclude.

Walmart's own corporate statement puts its U.S. average at $18.25 an hour, and spokesperson Jimmy Carter gave the Capital Chronicle an Oregon average of $20.51. Amazon's April proxy reports a $53,211 median for U.S. full-time employees and average base pay above $23 an hour in fulfillment and transportation. For a full-time worker those figures sit above Medicaid and SNAP thresholds, which is exactly why the question is who gets counted as full-time.

Illinois, One Year In

Illinois SNAP enrollment fell 16% in the year after the law took effect, from nearly 1.9 million people to nearly 1.6 million (the lowest level since 2009), per state Department of Human Services data reported by the Chicago Sun-Times.

Handle that 16% carefully. IDHS ties roughly 148,000 of the 305,000-person drop specifically to the new work requirements; the rest came from income changes, the law's separate restrictions on immigrant eligibility, and ordinary administrative churn. State officials had projected 400,000 losses from the work rules alone. Nearly 97% of the people still enrolled are exempt from those rules anyway.

Mercatus's Jack Salmon told PBS NewsHour in June that SNAP is returning to normal after years above its pre-emergency baseline; Harvard's Sara Bleich answered on the same segment that "eligible people are leaving the program because it's become too difficult to document." USDA Secretary Brooke Rollins calls the drop a win against fraud, but the error rate she cites measures over- and underpayments; actual fraud runs closer to 1.6%.

Who Benefits

Walmart and Amazon benefit from an obligation that doesn't exist rather than a payment that arrives. No federal law requires either company to offer health coverage to a worker scheduled under 30 hours a week, sets a minimum benefit for part-time workers, or charges an employer anything when its workers enroll in Medicaid. Public coverage makes a schedule that comes with no insurance survivable, which makes those schedules easier to staff.

On the tax side, the same bill moved in the opposite direction, restoring 100% bonus depreciation and immediate R&D expensing. Amazon's cash federal income tax payments fell from $7.63 billion in 2024 to $2.75 billion in 2025, figures the company itself supplied to Politico, and Amazon calls that a change in "the timing of our tax payments," not the total.

Walmart's FY2026 proxy, filed with the SEC, reports CEO pay of $29,240,930 against global median associate pay of $30,520, a ratio of 958 to 1, belonging to Doug McMillon, who ran the company all fiscal year (John Furner only became CEO February 1, 2026). Amazon spokesperson Alisa Carroll told KFF Health News (republished by The Nevada Independent) that "pointing fingers at Amazon over Medicaid is a red herring" and that the real fix is a much higher federal minimum wage, handing the problem to the Congress that just tightened the benefit.

Where the Subsidy Argument Breaks Down

The obvious way to tell this story is that your taxes subsidize Walmart's payroll, and that version has a problem. Eric Levitz made the case against it in Vox in April: the ACA's 2014 Medicaid expansion covered millions of previously ineligible workers, and no case study from that natural experiment found employers cutting wages in response. His warning is political too: telling people Medicaid subsidizes Walmart gives some of them a reason to support cutting it. Gideon Lukens at CBPP, which is fighting these cuts hard, opposes employer Medicaid charges on a related ground: charging per enrolled worker gives a company an incentive to avoid hiring people likely to qualify.

Both hold up, and neither touches the hours point. Levitz is answering whether wages fall when benefits expand. The threshold gap is a different question, and Congress, holding both numbers, moved only the one that removes people from coverage.

Governor Mikie Sherrill signed a law in June charging New Jersey employers $325 a year per Medicaid-enrolled worker at companies with 50 to 249 of them, and $725 at companies with 500 or more, the one state testing the other line. It took effect July 1 and hasn't collected a dollar yet; the state projects $145 million a year from roughly 750 entities, per New Jersey Monitor and CBS Philadelphia. Then read the exemptions: the fee doesn't apply to temporary, seasonal or part-time employees, so the only law in the country charging employers for Medicaid enrollment skips precisely the workers the 30-hour rule already skips.

Precedent for it is thin: Massachusetts charged employers up to $750 per nondisabled Medicaid-covered worker starting in 2018 and let it lapse the next year, and Maryland's 2006 version, which in practice reached only Walmart, was struck down in federal court on ERISA preemption grounds.

The Bottom Line

Sanders requested this audit and the 2020 one before it. His office's comparison of the two, relayed by CNN rather than published by GAO, puts Walmart's Medicaid count in the sampled states up about 55% since 2020 and Amazon's Medicaid and SNAP counts up nearly threefold. Congress had the earlier version on the shelf when it wrote H.R. 1; it has this one now, five months before the work requirement takes effect.

Come January, the workers in GAO's Medicaid tables start proving their hours to a state agency to keep coverage at jobs that still owe them nothing at 29 hours a week. New Jersey is the only government still trying to send those employers a bill, and it hasn't collected a dollar yet.