Who actually sponsors work visas

These are not the same measurement. The UK register lists who holds a licence and publishes no volumes at all. The US Data Hub publishes petition approvals and denials for a single fiscal year. Canada publishes approved LMIA positions. They are reported separately below and never summed into one “sponsorships” total, because that number would not mean anything.

161,302 register rows resolve to 159,965 distinct employers, because some companies hold a licence in more than one country.

Across 26,901 US employers in FY2023, USCIS recorded 176,693 approvals against 5,326 denials: an approval rate of 97.1%.

Sponsorship is usually described through the handful of companies that file thousands of petitions. The register says the opposite is typical: the great majority of licensed employers sponsor a handful of people, or one.

The 16 employers with 1000+ approvals account for 34,307 of them, more than the 21,683 smallest sponsors put together (32,969).

Counts describe petition volume, not company size. The register publishes no headcount, so a small number of approvals is not evidence of a small company.

NAICS sector as reported by the USCIS Data Hub, ranked by how many employers appear. An employer can report more than one sector.

126,520 UK employers hold a licence. The licence categories look varied until they are counted.

Where those employers are. The register records a town for effectively every row, so this is close to complete; its county field is blank on most rows and is deliberately not used here.

The positive-LMIA list is the most misread of the three. It is dominated by seasonal and food-production work, which is worth knowing before reading it as a map of demand for office roles.

The occupations those employers were approved for:

Joining the registers on employer name shows how rare a genuinely multi-country sponsor is.

• 1,287employers hold a licence in 2 of the 3 countries
• 25employers hold a licence in 3 of the 3 countries

The 25 on all three: accenture, airbnb, astek, autodesk, blackberry, brillio, cargill, ciena, electronic arts, epam systems, epic games, ericsson, insight global, invesco, ivalua, kinaxis, methodica technologies, pinterest, ringcentral, rockwell automation, sophos, sulzer pumps, tetra tech, ts imagine, valtech.

Matched on normalised employer name, so a company that files under different legal names in different countries is missed. Treat the overlap as a floor, not an exact count.

Figures come from the three official registers listed above, ingested into a single table and aggregated at build time. Every number on this page is computed from that table; none is typed in by hand.

• Register rows only. The same table also holds a Y Combinator company directory used for job matching. It is not a register of licensed sponsors and is excluded from every figure here. Counting it would give 165,927 rather than 161,302.
• Volumes are not comparable. UK licences carry no volume, US figures are FY2023 petition approvals, Canadian figures are approved LMIA positions. They are never added together.
• Snapshot dates are the last ingest that completed successfully, not the last time a row was touched. Where no ingest has ever succeeded, the date rows first landed is shown and flagged.
• Multi-value fields (route, sector, province, occupation) let one employer appear under more than one value, so those columns sum to more than the employer count.

The aggregates behind every table are downloadable as CSV under CC BY 4.0. If a figure here looks wrong, the underlying registers are public and linked above.

Free to reuse with attribution. Related: the Skilled-Visa Cost & Speed Index, and per-country breakdowns at United Kingdom, United States, Canada.