Company logo
Search
Company logo
Part 3 of 3

Who wins the H-1B lottery?

The H-1B is the main US temporary work visa for college-educated foreign professionals. Employers want far more of them than Congress allows, so the visas are allocated via lottery. Federal records obtained by Bloomberg under the Freedom of Information Act (FOIA) cover petitions from the four FY2021–24 lotteries. Part three of this series examines where in the country that work is based and which employers file for it.

H-1B Worker Geography

Approved H-1B petitions, FY2021–24
50%
in four states
California, Texas, New Jersey, and New York, which hold 29% of US private-sector jobs
99.1%
of the work is in metro areas
compared to 89.7% of US private-sector jobs
44%
in five metro areas
New York, Dallas, San Jose, Seattle, San Francisco
27%
won by just 29 employers
the 10 largest filers took 17.5% of all approved petitions; 82% of employers won fewer than five
How the lottery works

The odds are one in five

The H-1B is the largest US temporary work visa for foreign professionals in “specialty occupations.” This means a job that requires a bachelor’s degree or higher in that specific field. H-1B visa holders range from software engineers to accountants to doctors. Congress caps new visas at 85,000 a year: 65,000, plus 20,000 for people holding a graduate degree from a US university. Employers submit registrations for far more, so each spring the government runs a lottery. Across the four FY2021–24 lotteries, employers submitted 1.8 million registrations; about one in five ended in an approved H-1B petition. Those 363,024 approvals are the basis for the analysis that follows.

Read more

Employers, not workers, enter the lottery. A company submits a registration naming each person it wants to sponsor (the “beneficiary”), and US Citizenship and Immigration Services (USCIS) draws winners from eligible registrations. Each registration drawn allows one petition to be filed for that worker, and the worker can take the job only if USCIS ultimately approves the petition.

Employers submitted 1,847,661 registrations across the four FY2021–24 lotteries, of which 1,804,286 met USCIS eligibility requirements. Just under a third were drawn in the lottery. A petition was filed on two-thirds of those draws—some employers did not end up submitting petitions, despite winning the draw—and one registration in five ended in an approved petition. The following analysis is based on the 363,024 approved petitions that meet our sample criteria, which the notes set out in full. Universities and research organizations are not part of this data as they are exempt from the cap and hire outside the lottery.

The states

Half the petitions are in four states

If you want to know where H-1B workers are based, you can skip most of the map. Half of approved H-1B petition holders work in just four states: California, Texas, New Jersey, and New York. You might assume this just tracks where jobs are. It doesn't: those four states hold only 29 percent of US private-sector jobs.

Two states that stand out when comparing the ratio of H-1B jobs and private-sector jobs are Washington and Florida, near-mirror images of each other. Washington holds 6.9 percent of the H-1B work but only 2.2 percent of the country's private-sector jobs. Florida is the reverse: It holds 7.1 percent of the private-sector jobs but only 2.8 percent of the H-1B work. Only eleven of the fifty states, plus the District of Columbia, hold as large a share of the country's approved H-1B petitions as they do of its private-sector jobs. Everywhere else the H-1B share falls short, often by a lot.

Where the H-1B work is versus where private-sector jobs are, by state
FY2021–24
Approved H-1B petitions, by work-site state
Share of approved H-1B petitions, up to3.9%7.8%11.7%15.6%19.5%
H-1B work compared with private-sector jobs, by state
Share of approved H-1B petitions ÷ share of US private-sector jobs¼×½×1×2×3× or more
Source: H-1B: USCIS petition records obtained by Bloomberg under the Freedom of Information Act, FY2021–24. US benchmark: Census Bureau, County Business Patterns 2022. State outlines: the Census Bureau's 2022 cartographic boundary file, on an Albers equal-area projection.
Additional notes & methods

What the maps show and for which petitions. The map on the left shows each state's share of the 362,758 approved H-1B petitions naming a work-site state in the fifty states and DC, excluding Alaska and Hawaii (both listed in the table). The map on the right shows, for each state, its share of the approved H-1B petitions divided by its share of US private-sector payroll jobs. Two small groups of petitions appear on neither map: 65 petitions at work sites in the US territories and 201 petitions that name no work-site state at all.

US private-sector jobs. This count excludes government, the self-employed, and most farm and railroad workers.

Metro areas

H-1B jobs are concentrated in big metro areas

H-1B work concentrates in metropolitan areas far more tightly than US private-sector employment. Of the approved petitions, 99.1 percent sit in a metro area, compared to 89.7 percent of US private-sector jobs.

Other areas barely register. Micro areas—which hold 6.9 percent of the nation's jobs—have only 0.7 percent of the H-1B work. Smaller areas outside any metro or micro area hold 3.4 percent of the private-sector jobs and 0.2 percent of the H-1B work.

H-1B work versus private-sector jobs, by type of area
Share of each total · metro and micro areas drawn · FY2021–24
Metropolitan areas99.1%89.7%1.10Ă—Micropolitan areas0.7%6.9%0.10Ă—H-1B share Ă· US shareApproved H-1B petitionsUS private-sector jobs
Source: H-1B: USCIS petition records obtained by Bloomberg under the Freedom of Information Act, FY2021–24. US jobs: Census Bureau, County Business Patterns 2022. Statistical areas: US Office of Management and Budget, list of July 2023.
Additional notes & methods

The graph uses the Office of Management and Budget's July 2023 definition of statistical areas, which sorts counties into statistical areas built around population centers:

  • A metropolitan statistical area, or metro area for short, is an urban area of 50,000 people or more plus the surrounding counties whose commuters drive into it.
  • A micropolitan statistical area, or micro area for short, is an urban area of 10,000 to 50,000 people plus the surrounding counties whose commuters drive into it.
  • A county that belongs to neither metro nor micro area is a non-CBSA area.

 

We exclude from the data jobs in the US territories and jobs that cannot be located. The denominator in each chart includes jobs in metro, micro, and non-CBSA areas. For reference, the petitions in non-CBSA areas are 741 (0.2 percent of located work) and the US jobs in these areas are 3.4 percent of private-sector jobs.

The H-1B sample. From 363,024 approved petitions we remove the 65 at work sites in the US territories and the 761 that cannot be placed in any location, leaving a total of 362,198 approved petitions, including jobs in metro, micro, and non-CBSA areas.

The US benchmark. From 135,748,407 private-sector jobs we remove 7,534,819 jobs the Census Bureau cannot place in any county, leaving a total of 128,213,588 jobs located in metro, micro, and non-CBSA areas.

H-1B work piles up in a handful of metropolitan areas. The five biggest are New York, Dallas, San Jose, Seattle, and San Francisco. Together they hold 44 percent of the H-1B work while they employ only 13 percent of the overall country's private-sector workers. New York alone accounts for 14 percent of the H-1B work.

Zooming out a little farther, the top 10 percent of metro areas (by approved H-1B petitions) hosts 88 percent of the H-1B jobs.

Assessing the country as a whole, H-1B work sites span 385 metro areas, but half of them recorded fewer than thirty-seven approved petitions across all four years.

Approved H-1B petitions by metro area
One circle per metro area, sized by approved H-1B petitions · FY2021–24
approved petitions50k10k1k
Click a column heading to sort.
New York-Newark-Jersey City, NY-NJ 50,535 13.95% 6.40% 2.18Ă—
Dallas-Fort Worth-Arlington, TX 32,773 9.05% 2.68% 3.37Ă—
San Jose-Sunnyvale-Santa Clara, CA 29,578 8.17% 0.89% 9.13Ă—
Seattle-Tacoma-Bellevue, WA 24,793 6.85% 1.44% 4.76Ă—
San Francisco-Oakland-Fremont, CA 20,661 5.70% 1.84% 3.10Ă—
Chicago-Naperville-Elgin, IL-IN 15,677 4.33% 3.26% 1.33Ă—
Washington-Arlington-Alexandria, DC-VA-MD-WV 15,543 4.29% 2.16% 1.98Ă—
Boston-Cambridge-Newton, MA-NH 13,111 3.62% 2.03% 1.79Ă—
Los Angeles-Long Beach-Anaheim, CA 12,105 3.34% 4.26% 0.78Ă—
Atlanta-Sandy Springs-Roswell, GA 11,829 3.27% 2.00% 1.63Ă—
Source: H-1B: USCIS petition records obtained by Bloomberg under the Freedom of Information Act, FY2021–24. US jobs: Census Bureau, County Business Patterns 2022. Metro areas: US Office of Management and Budget, list of July 2023.
Additional notes & methods

The denominator, and why neither column adds to 100. In column three, the numerator only includes metro areas (the remainder 2,414 petitions in micropolitan areas and the 741 in non-CBSA areas are excluded); the denominator includes the 362,198 approved petitions that can be located in the fifty states and DC, excluding the US territories. The fourth column's numerator includes only private-sector jobs in the metro areas; the denominator includes 128,213,588 private-sector jobs that can be placed in a county.

How a petition is assigned to a metro area. It takes two steps. First, postal code to county, using the Department of Housing and Urban Development (HUD) postal-code-to-county crosswalk: each code goes to the county holding most of its addresses. Then county to metro area, using the government's own list.

The employers

The biggest filers file by the thousands

A handful of employers file at a scale no one else comes close to. Across the four lotteries, 50,618 different employers won at least one approved H-1B petition, but the ten largest took 17.5% of them. Amazon alone won 13,468 approved petitions, or 3.7 percent of the total. Most employers—82 percent—had fewer than five approved petitions.

Largest H-1B employers, by metro area, by state, and nationwide
Approved H-1B petitions, counted at the work site · FY2021–24
Amazon8,788Infosys4,859Google3,278Cognizant3,240TCS2,895Meta2,382Wipro1,755Capgemini1,389Intel1,305IBM1,260Qualcomm1,226Infosys815Walmart498Infosys496Amazon434Deloitte432Management Health430MathWorks405Infosys371Intel371UST Global347Eficens304Intel277Goldman Sachs265Syntel249Citibank239Infosys232Strategic213Coforge195Cerner182Infosys173Amazon155TCS153Deloitte145HCL134TCS129Tech Mahindra115Wells Fargo107ASML100Infosys95Capital One87Mesons69TCS59Variyas Global54Each dot is a metro area with at least 1,000 approved petitions, labeled with its single largest H-1B employerand sized and shaded by that employer's approved petitions.
The largest H-1B employer in each of the 44 metro areas with at least 1,000 approved petitions.Click a metro area for its five largest employers.
Click a column heading to sort.
AMAZON.COM SERVICES LLC Amazon 8,788 35.4%
INFOSYS LIMITED Infosys 4,859 14.8%
GOOGLE LLC Google 3,278 11.1%
COGNIZANT TECHNOLOGY SOLUTIONS US CORP Cognizant 3,240 96.7%
TATA CONSULTANCY SVCS LTD TCS 2,895 18.6%
META PLATFORMS INC Meta 2,382 11.5%
WIPRO LIMITED Wipro 1,755 3.5%
CAPGEMINI AMERICA INC Capgemini 1,389 8.9%
INTEL CORPORATION Intel 1,305 44.8%
IBM CORPORATION IBM 1,260 50.5%
Source: H-1B: USCIS petition records obtained under the Freedom of Information Act, FY2021–24, by Bloomberg. Map geometry: the Census Bureau's cartographic boundary files for states and for metro areas, on an Albers equal-area projection.
Additional notes & methods

Which petitions are included. Our data include 362,959 approved petitions that can be located at work sites in the fifty states and DC.

Employers are grouped by federal tax ID, not by name. Every company observation is grouped by federal tax ID, the legal entity behind the filings. The name on a petition varies in spelling, punctuation, and legal suffix, as well as companies renaming (for example, one tax ID filed as FACEBOOK INC in FY2021 and as META PLATFORMS INC from FY2022 on).

Which name is shown. A tax ID that filed under more than one name is shown under the name on its most recently received petition, which is why Facebook appears here as Meta.

The bottom line

H-1B petitions are concentrated in a handful of places and filed at scale by a few employers:

  • 50% of the H-1B work is based in four states: California, Texas, New Jersey, and New York.
  • H-1B work is more highly concentrated in metro areas than private-sector jobs: 99.1 percent of H-1B work sits in metro areas, compared to 89.7 percent of US private-sector jobs.
  • A handful of metro areas dominate. H-1B work sites span 385 metro areas, but the top five (New York, Dallas, San Jose, Seattle, and San Francisco) hold 44 percent of the work.
  • A few employers win most H-1B petitions. The ten largest took 17.5 percent of all approved petitions.
Notes
About these figures

The main data on this page come from USCIS petition records FY2021–24 obtained by Bloomberg under the Freedom of Information Act.

The H-1B process begins with a registration, submitted by an employer into an online system, that puts a candidate’s name forward for consideration. Because there are far more candidates than visas, US Citizenship and Immigration Services (USCIS) runs a lottery to select who is allowed to move forward. An employer whose registration is selected may then file a petition, and the worker can take the job only once USCIS approves that petition. The FY2021–24 lotteries described here drew registrations at random.

The data are individual approved petitions filed by US employers that meet the sample criteria below. One worker could be registered by several employers and in more than one year, so a single person can appear more than once.

How do we construct the sample?

In total, 1,847,661 registrations were submitted across the four lotteries. USCIS removed 43,375 before releasing the data (same-employer duplicates, registrations deleted before the filing period closed, invalid passport or travel-document information, and failed payments), leaving 1,804,286 eligible registrations.

We drop 139 rows redacted by USCIS, since they contain no information. We drop 907 observations whose beneficiaries were not of working age (20–64) at registration: 638 with a plausible birth year but an age outside the band; and 269 whose birth year is unusable. That leaves 1,803,240 working-age registrations.

Of those, 571,926 were selected via the lottery. Selection is only an invitation to file: Employers actually filed an H-1B petition (Form I-129) for 376,348 of them. From these we keep only ordinary new cap filings: petitions for new employment, in the standard H-1B classification, entered under either the regular (bachelor’s) cap or the US advanced-degree (master’s) cap. We drop 740 petitions that fail one of these checks (extensions, transfers, and amendments; other classifications; or a cap question that marks neither cap), leaving 375,608. Finally, we keep only the petitions USCIS approved, dropping 12,584 (11,030 denied and 1,554 still pending). That leaves the 363,024 approved petitions shown throughout. We drop 65 petitions at work sites in the US territories.

The denominator in each figure.

The state map and the state chart. H-1B: the 362,758 approved petitions naming a work-site state in the fifty states and DC. US jobs: all 135,748,407 private-sector jobs.

The chart by type of area, the metro chart, and the metro table. H-1B: the 362,198 petitions in the fifty states and DC whose work site can be placed on the map. US jobs: the 128,213,588 the Census Bureau can place in a county.

The employer figures. All 362,959 approved petitions at work sites in the fifty states and DC.

Definitions

Work site: the address where the job is, as recorded on the petition. A work-site state is recorded on 99.9 percent of approved H-1B petitions and a work-site postal code on 99.9 percent.

Metropolitan and micropolitan statistical areas: The Office of Management and Budget sorts counties into statistical areas (July 2023 list). A metropolitan statistical area, or metro area for short, is an urban area of 50,000 people or more plus the surrounding counties whose commuters drive into it. A micropolitan statistical area, or micro area for short, is an urban area of 10,000 to 50,000 people plus the surrounding counties whose commuters drive into it. A county that belongs to neither metro nor micro area is a non-CBSA area. Each work site is placed in two steps: its postal code is mapped to the county that holds most of that code's addresses, using the HUD USPS ZIP Code Crosswalk, and that county then maps to its metropolitan or micropolitan area on the July 2023 list.

Employer: a federal tax ID, shown under the name on its most recently received petition. A tax ID is recorded on every approved H-1B petition.

US private-sector jobs: paid employees at private establishments in County Business Patterns 2022, which excludes government, the self-employed, and most farm and railroad workers. The US file this page uses covers the fifty states and DC; the Census Bureau publishes Puerto Rico and the island areas as a separate series, which this page does not consider.

Sources

US private-sector employment by state and by metro area: Census Bureau, County Business Patterns 2022. Metro area definitions: US Office of Management and Budget, July 2023. Work-site ZIP code to county: HUD USPS ZIP Code Crosswalk. State and metro outlines: the Census Bureau's cartographic boundary files. H-1B: USCIS petition records obtained by Bloomberg under the Freedom of Information Act, FY2021–24; see the H-1B program.

Replication

The underlying data, the code behind every figure on this page, and the full method notes are available on request.

For questions or comments, please reach out to the Hoover Immigration team at hooverimmigration@stanford.edu.

Expand
overlay image