Foreign-born workers comprise 18.6% of the U.S. civilian labor force β approximately 30.4 million workers β according to BLS data from 2025. These workers are disproportionately concentrated at both ends of the skill spectrum: in low-wage service and manual labor jobs AND in high-skill STEM occupations. Both segments face significant AI displacement risk, yet immigration and automation are rarely analyzed together. This oversight leaves one of America's most vulnerable worker populations without adequate policy attention.
The Immigrant Workforce: A Dual Concentration
The immigrant workforce is not evenly distributed across occupations. BLS Current Population Survey data reveals stark concentration patterns:
| Occupation Group | Total Employment | Foreign-Born Share | National Average | Overrepresentation | ADI Range |
|---|---|---|---|---|---|
| Building/Grounds Maintenance (37-0000) | 5,920,000 | 38.4% | 18.6% | 2.06x | 25β45 |
| Construction/Extraction (47-0000) | 7,120,000 | 31.2% | 18.6% | 1.68x | 18β35 |
| Food Prep/Serving (35-0000) | 13,290,000 | 27.8% | 18.6% | 1.49x | 30β65 |
| Production (51-0000) | 8,940,000 | 25.6% | 18.6% | 1.38x | 40β70 |
| Computer/Mathematical (15-0000) | 4,920,000 | 28.5% | 18.6% | 1.53x | 25β55 |
| Healthcare Support (31-0000) | 7,480,000 | 24.2% | 18.6% | 1.30x | 20β45 |
| Transportation/Moving (53-0000) | 10,080,000 | 22.8% | 18.6% | 1.23x | 35β65 |
| Office/Admin Support (43-0000) | 18,940,000 | 12.4% | 18.6% | 0.67x | 55β80 |
Low-Wage Immigrant Workers and AI
The conventional wisdom holds that low-wage immigrant workers are "safe" from AI because their work is physical and manual. This is partially true but dangerously oversimplified.
Where Low-Wage Immigrant Workers Are Protected
- Construction: 2.22 million foreign-born construction workers perform physical tasks (framing, roofing, masonry, electrical, plumbing) that are extremely difficult to automate. ADI scores for most construction occupations are 18β35. This is the most AI-resilient major immigrant employment sector.
- Building/grounds maintenance: Landscaping, cleaning in varied environments, and building maintenance involve physical dexterity and environmental adaptation that robotics cannot yet match at cost parity.
- Agriculture: 1.1 million foreign-born agricultural workers harvest crops, tend livestock, and perform varied physical tasks. Harvesting automation exists for some crops but remains uneconomical for most fruits and vegetables.
Where Low-Wage Immigrant Workers Are Vulnerable
- Food service: 3.7 million foreign-born food service workers face growing automation. Self-ordering kiosks (now in 85% of major fast-food chains), robotic food preparation (Flippy, Makr Shakr), and AI-managed kitchen operations are reducing per-restaurant staffing by 15β25%.
- Production/manufacturing: 2.29 million foreign-born production workers face continued robotics + AI displacement. AI-guided robotic systems are expanding into tasks previously requiring human dexterity β garment sewing, food processing, electronics assembly.
- Transportation: 2.3 million foreign-born transportation/moving workers face autonomous vehicle displacement. Foreign-born workers comprise 24% of taxi/rideshare drivers (SOC 53-3041) and 20% of heavy truck drivers β roles directly threatened by autonomous vehicles.
- Retail: 1.8 million foreign-born retail workers face the same cashier automation and e-commerce pressures affecting all retail workers, compounded by potential language barriers in transitioning to more complex roles.
High-Skill Immigrant Workers: H-1B and the AI Paradox
The other end of the immigrant worker spectrum β high-skill visa holders, predominantly H-1B β faces a different but equally consequential AI challenge.
BLS data shows approximately 1.4 million foreign-born workers in computer and mathematical occupations, representing 28.5% of the field. Many hold H-1B visas (approximately 583,000 active H-1B workers in STEM as of 2025, per USCIS data). These workers face unique vulnerabilities:
| Vulnerability | Detail | Impact |
|---|---|---|
| Visa tied to employment | H-1B holders have 60 days to find new employment after layoff or must leave the U.S. | Layoff = potential deportation; less leverage to negotiate or resist |
| Concentrated in AI-vulnerable roles | H-1B workers are overrepresented in software development, QA, IT support | Disproportionate exposure to AI coding tools displacing junior roles |
| Green card backlogs | Indian-born applicants face 80+ year backlogs in EB-2/EB-3 categories | Workers trapped in H-1B status with no path to permanent residency; employer dependency intensified |
| Wage suppression tool | Some employers use H-1B workers at lower wages; AI gives justification to eliminate these positions | H-1B holders are "easy" layoff targets β no severance obligation complications |
The H-1B Displacement Data
USCIS data combined with layoff trackers reveals concerning patterns:
- H-1B transfer applications dropped 28% from 2023 to 2025, suggesting fewer job opportunities for visa-dependent workers to transition to
- An estimated 45,000 H-1B workers were affected by tech layoffs in 2023β2025
- The 60-day grace period forces H-1B holders to accept any available position, often at lower wages, or leave the country β creating a race to the bottom during layoffs
- Companies sponsoring fewer new H-1Bs: initial H-1B applications declined 12% for FY2026, despite overall visa cap remaining at 85,000
Undocumented Workers: The Most Vulnerable
An estimated 7.5 million undocumented workers are in the U.S. labor force (Pew Research Center, 2024). They are concentrated in the most precarious employment sectors:
| Sector | Undocumented Worker Share | Estimated Workers | AI/Automation Risk |
|---|---|---|---|
| Agriculture | ~44% | ~1,100,000 | Moderate (crop-dependent) |
| Construction | ~13% | ~1,400,000 | Low-Moderate |
| Hospitality/Food Service | ~10% | ~1,350,000 | Moderate-High |
| Manufacturing | ~8% | ~950,000 | Moderate-High |
| Domestic Work/Cleaning | ~25% | ~850,000 | Low-Moderate |
| Other Services | Varies | ~1,850,000 | Varies |
Undocumented workers are completely excluded from government retraining programs, unemployment insurance, and most safety net benefits. When their jobs are automated, they have essentially zero institutional support.
Language Barriers and the Retraining Gap
Even documented immigrant workers face significant barriers to retraining and reskilling:
- Limited English Proficiency (LEP): 21.6 million working-age adults have LEP (ACS 2024). Most retraining programs, online courses, and certification exams are English-only.
- Credential non-recognition: Foreign degrees and professional credentials are often not recognized, forcing immigrants into lower-skill work even when they have relevant qualifications
- Digital literacy gaps: First-generation immigrants over 45 have lower rates of digital fluency, making AI-adjacent retraining more challenging
- Navigational barriers: Complex bureaucratic systems for accessing workforce development, community college programs, and benefits create friction that discourages participation
- Work schedule conflicts: Many immigrant workers hold multiple jobs or irregular schedules, making it difficult to attend structured training programs
Policy Blind Spots
Current workforce policy fails immigrant workers at multiple levels:
| Policy Area | Current Status | Gap |
|---|---|---|
| Workforce Innovation and Opportunity Act (WIOA) | Federal retraining funding available to authorized workers | Programs rarely offer multilingual services; cultural competency lacking |
| Trade Adjustment Assistance (TAA) | Retraining for trade-displaced workers | No equivalent for AI-displaced workers; excludes undocumented workers entirely |
| H-1B portability | 60-day grace period after layoff | Insufficient time for job search; creates coercive employer dependency |
| Community college access | Varies by state; some offer in-state tuition to undocumented residents | Most AI/tech programs require English proficiency and prior digital literacy |
| Unemployment insurance | Available to authorized workers with qualifying wages | Excludes undocumented workers; H-1B holders face status complications |
What Other Countries Do Differently
Several nations have integrated immigrant workforce transitions into their AI strategies:
- Canada: The Global Talent Stream provides expedited work permits for AI-sector workers, while provincial nominee programs offer permanent residency pathways not tied to a single employer. Canada's Immigrant Settlement Program funds multilingual digital literacy training.
- Germany: The Skilled Immigration Act (FachkrΓ€fteeinwanderungsgesetz) recognizes foreign credentials more broadly and funds German-language technical training. Integration courses include digital skills modules.
- Singapore: The SkillsFuture program provides $500 annual training credits to all residents, including permanent residents. Programs are offered in Mandarin, Malay, and Tamil alongside English.
- Australia: The Skills Priority List explicitly considers AI displacement risk in immigration planning, steering immigrant workers toward AI-resilient occupations.
Recommendations
- Multilingual retraining programs: WIOA-funded programs should be required to offer training in languages representing at least 80% of the local LEP workforce β in most areas, this means Spanish and one or two additional languages
- H-1B reform: Extend the post-layoff grace period to 180 days; allow H-1B holders to access WIOA retraining; create a startup visa pathway for displaced H-1B workers
- Portable credentials: Develop nationally recognized micro-credentials that are language-accessible and transferable across states and employers
- Undocumented worker inclusion: At minimum, state-level retraining programs should not require immigration status verification β workforce stability benefits all workers and employers
- Immigrant entrepreneurship support: Displaced immigrant workers have high entrepreneurship rates when given access to capital and mentorship; expand SBA programs targeted at immigrant entrepreneurs
- Data collection: BLS and Census Bureau should collect more granular data on AI displacement by immigration status and national origin to inform targeted policy responses
The Numbers Don't Lie
Our modeling estimates that 4.2β6.8 million foreign-born workers hold positions with ADI scores above 50, placing them in the elevated-to-very-high risk categories. Of these:
- 2.1β3.4 million are in low-wage service and production roles where displacement means falling into poverty
- 800,000β1.2 million are visa-dependent tech workers whose displacement triggers immigration consequences
- 1.3β2.2 million are undocumented workers with no institutional safety net
Conclusion
The intersection of AI displacement and immigration creates a crisis that neither workforce policy nor immigration policy is designed to address. Immigrant workers are concentrated in exactly the occupations most vulnerable to automation β at both the low-wage and high-skill ends β while facing additional barriers (language, visa status, credential recognition) that make adaptation harder. Until policymakers recognize that AI displacement is an immigration issue and immigration is an AI policy issue, millions of the most vulnerable workers in America will fall through the cracks.