American universities use H-1B and J-1 visas as cheap labor to feed grant-driven revenue, not to fill real expertise gaps

Source: Chris Bray. "Universities Are Grant Money Factories Relying On Foreign Labor." October 16, 2026. thefederalist.com

The Gist

The author argues that universities say they need foreign workers for rare, specialized skills, but the evidence suggests they use visas to hire cheaper labor. Stanford's hiring for an ordinary finance job, plus a glut of American PhDs who can't find research jobs, point to a money-driven system built around grants rather than a real talent shortage.

Conclusion

American universities use temporary foreign labor visas (H-1B and J-1) primarily to obtain cheap labor that supports grant-money generation, rather than to fill genuine gaps in specialized expertise that no American can fill.

Premises

  1. The H-1B visa was designed as a gap-filler for highly specialized work requiring esoteric knowledge that Americans cannot do, and the J-1 visa is a nonimmigrant visa for temporary work such as research.
  2. Stanford is seeking an H-1B Financial Analyst 3 at $149,000, a role that by the university's own comparable postings requires only a bachelor's degree, accounting-focused analysis, and strong Excel skills, which are not rare or esoteric qualifications.
  3. American universities overproduce STEM PhDs relative to available academic jobs: only about 25% of recent STEM PhD cohorts ever reach tenure-track positions (about 12% in engineering), and postdoc rates are well below 100% in most fields.
  4. Despite the supply of American-trained PhDs, Stanford reports that about 66 percent of its postdocs are international.
  5. Junior faculty in science and engineering depend on federal grant money, creating perverse incentives to produce revenue for the university and therefore to minimize labor costs.
  6. Cheap foreign labor fits this financial-incentive environment, whereas the rhetoric about importing foreign researchers for their unique expertise does not match the evidence.

Assumptions

View this argument on LogicFirst.ai