Who's Actually Filing for H-1B Workers in 2026? Almost Never the Government

Methodology

Source: our local mirror of DOL's FY2026 H-1B LCA disclosure file (2026h1b_combined). DOL's quarterly FY2026 files are cumulative rather than incremental, so this report uses only the Q3 snapshot (437,496 distinct CASE_NUMBER rows, covering applications received through June 30, 2026) rather than summing Q1+Q2+Q3, which would double- and triple-count the same filings. All case statuses are included — Certified, Certified-Withdrawn, Denied, and Withdrawn — because the question here is what employers applied for, not only what was ultimately approved.

Employer type was assigned by pattern-matching each EMPLOYER_NAME against government and education naming conventions (e.g. "City of," "County of," "Department of," "School District," "ISD," "University," "College," "Board of Regents"), since the disclosure data has no employer-type field of its own. Candidate matches were manually reviewed; several false positives were found and excluded by name — most notably City of Hope (a private cancer hospital, not a municipal government), two private military academies, a company literally named "State of Escape Communications," and a "port of" pattern that was catching the word "import." Residual noise after this cleanup is expected to be small but not zero.

Position counts use TOTAL_WORKER_POSITIONS, the number of workers an employer requests on the LCA. As documented elsewhere on this site, these are batch-filing ceilings — often suspiciously round numbers (25, 50, 100) — not confirmed hires or a real headcount, so treat "positions requested" as an upper bound on hiring intent, not a count of people actually placed.

Contractor placements at government/school worksites are captured separately using the LCA's SECONDARY_ENTITY / SECONDARY_ENTITY_BUSINESS_NAME fields — the form's own disclosure of a third-party client site when the sponsoring employer isn't where the work actually happens. 77,597 of 437,496 FY2026 filings (17.7%) name a real third-party client this way (the remaining filings show "No," with placeholder values like "Remote" or "Beneficiary's Residence" rather than a client name, so this population is fully captured, not under-sampled). Applying the same name-pattern matching used for EMPLOYER_NAME above to SECONDARY_ENTITY_BUSINESS_NAME finds several thousand filings where the LCA's employer of record is an ordinary staffing or consulting firm, but the actual worksite is a government agency, public university, or school district — see "The Hidden Government Footprint" below.

The Employer-Type Breakdown

Splitting all 437,496 FY2026 applications by who's actually filing them:

Universities and school systems are counted regardless of public/private status, matching the "schools, universities" framing this report was built to answer. "True government agencies" excludes all education employers, however they're organized (e.g. a city "Department of Education" is counted as a K-12 school system, not a generic government agency). This table counts each filing under its LCA employer of record only — it does not yet include contractor-placed workers whose actual worksite is a government agency or school; see the next section for that additional layer.

Where the Government/Education Filings Actually Are

The true government-agency list (179 employers, all filing under 25 positions apiece — mostly state DOTs, city treasuries, and corrections departments) is small enough that even summing it entirely comes to 499 positions nationwide: less than one-tenth of one percent of the FY2026 total. That's the LCA employer-of-record view only, though — the next section adds contractor-placed workers.

The Hidden Government Footprint: Workers Placed by a Staffing Contractor

A large share of H-1B filings list a private staffing or consulting firm as the "employer," with the actual worksite disclosed separately as a third-party client. Matching that client field against the same government/education patterns used above finds thousands of filings where the sponsoring company is ordinary private business, but the work itself happens inside a state agency, public university, or school district:

So the "hidden" contractor layer is real — it multiplies the direct government-agency count by roughly 6.4× (499 → 3,202 filings) — but it still doesn't change the headline: even counting every contractor placement, government agencies account for well under 1% of FY2026's H-1B filings. Two clear patterns emerge in who's doing the placing:

The K-12 layer is a different niche almost entirely: therapy and special-education staffing agencies placing H-1B occupational and speech therapists in school districts — The Stepping Stones Group LLC (93 filings), EdTheory LLC (38), Pristine Rehab Care, LLC (30), BMR Health Services, Inc. (23), and SpEd Therapy Services, Inc. (16 filings covering 122 positions — one of the largest single position counts in this entire hidden layer). This is a distinct, largely non-tech corner of H-1B usage inside public schools, worth its own follow-up.

Same caveat as the primary employer-name matching: this is regex-based name matching against SECONDARY_ENTITY_BUSINESS_NAME, manually spot-checked but not exhaustively hand-verified, so treat these as a solid estimate rather than an exact count.

Inside the Private Sector: Size Concentration

The disclosure data has no employee-count or revenue field, so "large" and "small" business here is a proxy — total positions an employer requested across all its FY2026 filings, tiered at 500+ (large), 50–499 (mid), and under 50 (small). By that measure, positions are heavily concentrated at the top:

The Better Axis: DOL's Own "H-1B Dependent" Legal Flag

Company size doesn't actually separate "Intel good, Infosys bad" — Infosys, Cognizant, TCS, and Wipro aren't small, they're huge; the distinction Matloff is drawing is about business model, not headcount. DOL already tracks something closer to that: H_1B_DEPENDENT, a statutory designation (based on the employer's ratio of H-1B workers to its total U.S. workforce) that triggers extra attestation requirements. In FY2026, 97,711 private-sector filings — 23.8% — carry the "Yes" flag:

That's the opposite of what a "small body shop" narrative predicts: the Large tier has the highest H-1B-dependent share of positions, not the lowest. A few of the largest filers by volume make this concrete:

† Grandison Management averages roughly 40 positions per filing across 1,988 filings — a pattern consistent with the round-number batch filing this site has flagged elsewhere — yet is not DOL-flagged as H-1B dependent. That flag is a workforce-ratio formula, not a catch-all for filing behavior that looks like a body shop; it can legitimately miss firms whose overall U.S. workforce is large enough to dilute their H-1B ratio, even if their H-1B filing volume looks similar to a dependent employer's.

Investigative Assessment

💬 Comments