FY2025the evidence on pay
H-1B pay vs. the local market
The clearest finding: 55.8% of matched filings offer less than local median pay. That comparison uses 404,917 H-1B filings from FY2025. We compare the low end of each offered range with the median for the same occupation and area. That median includes all experience levels. See the pay comparison
Why it matters. A six-figure salary can still fall below local pay. The useful comparison is the same occupation in the same area.
What this does not prove. A below-median offer alone does not establish a wage violation or job replacement. These filings report offers, not actual paychecks or visa grants.
Finding 01
55.8% of matched offers fall below local median pay.
Compare an offer with what the same occupation pays in the same area. A national salary average can miss that difference.
The median is the midpoint: half of local workers in that occupation earn more, and half earn less.
55.8% of matched filings offer less than the local median. When a filing is below the median, the median gap is −$21,924 a year.
An “H-1B dependent” employer reports a large share of H-1B workers on its staff. See the definition and wage-floor comparison.
View the pay comparison data
The shortfall cited above uses only offers below the local median. The table's median gap includes all matched offers in each group.
| Employer type | Matched filings | Below local median | Median gap |
|---|---|---|---|
| All employers | 404,917 | 55.8% | −$4,427 |
| “H-1B dependent” employers | 96,325 | 69.9% | −$18,221 |
| Other employers | 308,592 | 51.3% | $0 |
Explore 15 large sponsors with the highest shares below the median
These large sponsors offer less than the local median most often. Each has 500 or more matched filings.
| Employer | Matched filings | Below local median | Median gap |
|---|---|---|---|
| WIPRO LIMITED | 1,655 | 90.3% | −$19,198 |
| TATA CONSULTANCY SERVICES LIMITED | 6,708 | 90.3% | −$20,425 |
| TECH MAHINDRA (AMERICAS), INC | 1,065 | 85.4% | −$19,531 |
| COMPUNNEL SOFTWARE GROUP, INC | 1,572 | 78.9% | −$16,689 |
| Qualcomm Technologies, Inc. | 1,684 | 77.3% | −$23,192 |
| Hexaware Technologies, Inc. | 505 | 75.8% | −$18,470 |
| Deloitte Tax LLP | 833 | 71.3% | −$17,389 |
| Deloitte & Touche LLP | 1,224 | 70.3% | −$17,888 |
| Intel Corporation | 2,490 | 69.9% | −$10,777 |
| INFOSYS LIMITED | 3,460 | 67.1% | −$16,744 |
| LTIMindtree Limited | 2,131 | 66.9% | −$14,852 |
| Fidelity Technology Group, LLC d/b/a Fidelity Investments | 1,156 | 66.7% | −$18,587 |
| Amazon Development Center U.S., Inc. | 2,311 | 66.4% | −$18,846 |
| ServiceNow, Inc. | 794 | 65.1% | −$13,092 |
| UST Global Inc | 736 | 65.1% | −$9,438 |
Finding 02
Meeting the wage floor can still mean below-median pay.
An offer can meet the listed wage floor and still fall below the midpoint of local pay.
DOL's four wage levels reflect different job requirements. The first two sit below the local median. Level III is at the median. Level IV is above it.
In FY2025, 62.4% of filings with a known wage level were at Level I or II. That is the wage floor, not necessarily the offered pay.
Finding 03
H-1B dependent employers offer the minimum more often.
Employers that report H-1B dependence more often offer exactly the listed wage floor.
DOL calls an employer “H-1B dependent” when H-1B workers are a large part of its staff. For an employer with 51 or more full-time workers, the limit is 15%.
View the data by employer type
| Employer type | Filings | Level I or II | Offered exactly the wage floor | Median offered pay |
|---|---|---|---|---|
| “H-1B dependent” employers | 131,461 | 73.6% | 52.0% | $105,997 |
| Other employers | 407,983 | 58.6% | 27.9% | $127,878 |
Finding 04
Offers tied to private surveys fall below the median more often.
Offers tied to private salary surveys fall below the local median more often than offers tied to government wage data.
Each filing states where its wage floor came from. Most filings use government data. Some employers choose a private salary survey instead.
View all wage sources
| Where the wage floor came from | Filings | Share | Below local median (of matched filings) |
|---|---|---|---|
| Government data (OEWS) | 498,480 | 92.4% | 54.1% |
| Private survey, chosen by the employer | 33,759 | 6.3% | 75.8% |
| Union contract | 6,531 | 1.2% | 77.5% |
| DOL determination | 590 | 0.1% | 75.9% |
| Federal contract wage | 84 | <0.1% | 92.3% |
View the most-used survey publishers
The survey publishers on the most filings:
| Survey publisher | Filings | Below local median (of matched filings) |
|---|---|---|
| Willis Towers Watson | 11,241 | 72.3% |
| Radford (Aon) | 9,521 | 77.6% |
| Aon | 6,242 | 77.6% |
| Mercer | 3,092 | 72.1% |
| AAMC | 1,165 | 97.3% |
| Pearl Meyer | 1,030 | 76.8% |
| U.S. Bureau of Labor Statistics | 209 | 8.5% |
| CUPA-HR | 190 | 93.7% |
Beyond the salary
Different records. Different questions.
Pay comparisons show a gap. Enforcement records show documented violations. Layoffs, lottery entries, and green card filings add context, with limits of their own.
Finding 05
$123.2M in back wages from documented violations.
These cases establish that violations occurred. They cannot tell us how common violations are across all H-1B employers.
DOL's Wage and Hour Division found H-1B violations in 2,886 cases. In these cases, employers agreed to pay $123.2M in back wages to 13,595 workers. DOL also assessed $11.7M in penalties.
View the employers with the most back wages
| Employer (as named by WHD) | State | Cases | Back wages | Workers owed |
|---|---|---|---|---|
| Prince George's County Public Schools | MD | 1 | $4,222,146 | 1,046 |
| Advanced Prfessional Marketing, Inc. | NY | 1 | $2,865,216 | 156 |
| Patni Computer Systems Inc. | MA | 2 | $2,411,537 | 609 |
| Computech Corporation | MI | 1 | $2,250,000 | 36 |
| Byton North America Corporation | CA | 1 | $1,749,184 | 76 |
| GlobalCynex, Inc. | VA | 2 | $1,683,583 | 343 |
| Populus Group, LLC | MI | 3 | $1,159,056 | 596 |
| Metropolitan Center for Mental Health | NY | 2 | $1,113,677 | 15 |
| Smartsoft International, Inc. | GA | 1 | $997,732 | 135 |
| Sirsai Inc. | WA | 1 | $762,806 | 115 |
Finding 06
Companies laid off workers, then filed for new H-1B workers.
929 companies filed 2,956 layoff notices for 285,788 workers. In the 12 months after those notices, the same companies filed 84,720 H-1B filings for workers new to the company.
- The layoff. Before a large layoff, a company must send a WARN notice to the state. We use 14,753 notices from 45 states, dated Oct 2020 to Jun 2025.
- The match. 4,850 of these notices are from companies that also file for H-1B workers. The company names must match exactly.
- The filings after. For each notice, we count the company's certified H-1B filings in the next 12 months for workers who are new to the company. Extensions for current workers do not count.
After 60.9% of the matched notices (2,956 of 4,850), the company filed for at least one new H-1B worker within 12 months. Those notices are the ones we count below.
The layoffs did not stop the H-1B filings. In the 12 months after any of their layoff notices, these companies filed 83.1% as many new-worker filings as in the 12 months before any of them. 39.5% of the filings after a notice were for a job in the same state as the layoff.
Are these extensions? No. Each counted filing asks for at least one worker who is new to the company: 58,427 for new employment and 26,293 for workers who moved from another H-1B employer. In the same months, the companies also filed 109,246 H-1B filings that only extend or amend the visas of current workers. We do not count those.
| New employment (a worker new to the company) | 58,427 | 30.1% |
| Change of employer (a worker who moves from another H-1B employer) | 26,293 | 13.6% |
| Extensions and amendments (current workers only; not counted) | 109,246 | 56.3% |
An example: of the companies in the table below, Google filed the most after its notices. Its layoff notices that a new filing followed (CA, NY, the first in Jan 2023) were for 2,030 workers. In the 12 months after them, it filed 11,926 H-1B filings for new workers, 9,811 of them within 90 days. In the 12 months before any of its layoff notices, it filed 17,065.
The records do not show that the new H-1B workers replaced the workers who were laid off.
View the companies with the most workers in these layoff notices
Select a company name to see its employer page.
| Company (name in the notice) | States | Workers in layoff notices | First notice | New-worker H-1B filings, 12 months after | Of those, within 90 days | Same filings, 12 months before any notice | Of the filings after: new employment | Of the filings after: from another employer | Extensions and amendments after (not counted) |
|---|---|---|---|---|---|---|---|---|---|
| Meta Platforms, Inc. | CA, NJ, NY, WA | 8,068 | Nov 2022 | 2,764 | 719 | 3,474 | 1,636 | 1,128 | 4,494 |
| Tesla, Inc. | CA, MD, NV, NY, TX | 8,008 | Jun 2022 | 2,744 | 986 | 2,030 | 1,843 | 901 | 2,556 |
| FCA US LLC | MI, OH | 5,821 | Jul 2023 | 139 | 19 | 146 | 49 | 90 | 376 |
| Microsoft | CA, WA | 5,716 | Jan 2023 | 5,520 | 1,613 | 8,428 | 3,352 | 2,168 | 14,518 |
| BOEING | AL, CA, CO, FL, LA, MO, VA, WA | 4,692 | Nov 2023 | 55 | 33 | 48 | 40 | 15 | 26 |
| Jabil Inc. | CA, NM, TX, WA | 4,485 | Oct 2020 | 93 | 41 | 114 | 62 | 31 | 120 |
| American Airlines, Inc. | AZ, CA, FL, MO, PA | 4,053 | Feb 2021 | 283 | 48 | 254 | 116 | 167 | 517 |
| Sodexo | AL, CA, CO, CT, DC, FL, HI, IL, LA, MO, NC, NJ, NV, NY, OH, OK, TN, VA, WV | 3,771 | Oct 2020 | 37 | 22 | 36 | 26 | 11 | 62 |
| GENERAL MOTORS LLC | KS, MI | 3,643 | Oct 2023 | 412 | 52 | 294 | 147 | 265 | 1,074 |
| Intel Corporation | CA, TX | 3,148 | Dec 2022 | 6,007 | 3,863 | 7,605 | 5,696 | 311 | 5,222 |
| Cisco Systems, Inc. | CA | 2,750 | Dec 2020 | 3,780 | 1,265 | 4,087 | 3,048 | 732 | 2,004 |
| Wayfair, LLC | CA, MA | 2,556 | Aug 2022 | 172 | 22 | 461 | 107 | 65 | 386 |
| Cepheid | CA | 2,530 | Aug 2022 | 69 | 30 | 163 | 47 | 22 | 186 |
| Qualcomm Incorporated | CA, NJ | 2,356 | Dec 2022 | 159 | 103 | 207 | 141 | 18 | 83 |
| Bristol Myers Squibb | CA, IL, NJ | 2,107 | Sep 2022 | 395 | 273 | 436 | 148 | 247 | 429 |
| Genpact LLC | CA, CT, FL, IL, OH, TX | 2,089 | Oct 2020 | 777 | 475 | 962 | 484 | 293 | 391 |
| CA, NY | 2,030 | Jan 2023 | 11,926 | 9,811 | 17,065 | 10,019 | 1,907 | 11,149 | |
| Tyson Foods | FL, IL, IN, MO, SD, VA | 2,019 | Mar 2023 | 51 | 21 | 54 | 35 | 16 | 31 |
| Pitney Bowes Inc. | CA, FL, GA, IL, IN, MD, NJ, NV, OH | 1,956 | Feb 2021 | 44 | 20 | 67 | 23 | 21 | 92 |
| Charter Communications | CA, HI, MA, MI, MT, NC, NY, OH, WI | 1,930 | Jun 2021 | 454 | 193 | 641 | 127 | 327 | 1,295 |
Finding 07
Multiple registrations fell after the lottery changed.
A person once gained more chances through multiple employer registrations. Since FY2025, selection is by person.
Each year, employers register the people they want to hire. USCIS selects registrations at random. Up to the FY2024 lottery, each registration was one chance of selection. A person with registrations from several employers had several chances.
In FY2024, 408,891 of 758,994 eligible registrations were for people with more than one registration (53.9%). USCIS cites evidence from the FY2023 and FY2024 lotteries. It opened fraud investigations and changed the rule.
From FY2025, USCIS selects by person, so more registrations do not give more chances. In FY2026, the share fell to 2.3%.
View lottery registration data by year
| Cap year | Eligible registrations | For people with multiple | Share | Selected |
|---|---|---|---|---|
| FY2021 | 269,424 | 28,125 | 10.4% | 124,415 |
| FY2022 | 301,447 | 90,143 | 29.9% | 131,924 |
| FY2023 | 474,421 | 165,180 | 34.8% | 127,600 |
| FY2024 | 758,994 | 408,891 | 53.9% | 188,400 |
| FY2025new rule | 470,342 | 47,314 | 10.1% | 135,137 |
| FY2026 | 343,981 | 7,828 | 2.3% | 120,141 |
Finding 08
Most professional green card filings are for existing employees.
The employer often tests the U.S. labor market for a job that its sponsored worker already holds. These records do not show H-1B status.
To sponsor a worker for a green card, the employer must show that no qualified U.S. worker applied for the job. The employer places job ads and reviews the U.S. applicants.
In FY2025, the worker already worked for the employer on 91.5% of certified filings for professional jobs. For all jobs, the share was 67.1%. The employer tests the U.S. labor market while its sponsored worker already works for it.
6,408 certified filings came from 147 employers that reported a layoff in the area, in the occupation or a related occupation, in the 6 months before the filing.
View green card sponsors that reported prior layoffs
| Employer | Certified filings | After a layoff | Share |
|---|---|---|---|
| Microsoft Corporation | 2,926 | 2,296 | 78.5% |
| Oracle America, Inc. | 1,190 | 759 | 63.8% |
| TATA CONSULTANCY SERVICES LIMITED | 506 | 506 | 100.0% |
| Wipro Limited | 506 | 504 | 99.6% |
| Intel Corporation | 936 | 477 | 51.0% |
| Capgemini America, Inc. | 321 | 320 | 99.7% |
| CGI Technologies and Solutions Inc. | 133 | 122 | 91.7% |
| Concentrix CVG Customer Management Group Inc. | 117 | 117 | 100.0% |
| Wells Fargo Bank, N.A. | 167 | 108 | 64.7% |
| AT&T SERVICES, INC. | 168 | 89 | 53.0% |
| American Express Travel Related Services Company, Inc. | 82 | 79 | 96.3% |
| JPMORGAN CHASE & CO. | 711 | 73 | 10.3% |
| Elevance Health, Inc. | 107 | 64 | 59.8% |
| Marvell Semiconductor, Inc. | 60 | 57 | 95.0% |
| Hexaware Technologies, Inc. | 48 | 48 | 100.0% |
The DOL approval is for one employer only. A worker who moves to a different employer must start a new case with that employer. The DOL step alone took a median of 483 days. For 90% of filings, it took 510 days or less.
A note on the evidence
Keep three distinctions in mind.
- An offer is not a paycheck.
- Filings report an offered wage. They do not show actual earnings or prove that a worker received a visa.
- A benchmark is not a legal minimum.
- The local median includes all experience levels. A below-median offer alone does not establish a violation.
- Overlap is not replacement.
- A company can report layoffs and H-1B filings for different jobs. These records cannot show who replaced whom.
