The Dominoes: What Fewer International Students Means for American Higher Education

For years, part of my job was tracking CPT, Curricular Practical Training. CPT is what allows an international student to do an internship or co-op as part of their degree. Each request passed through three sign-offs: first the student's college, then an advisor reviewing the paperwork, and last the international student office, whose approval was the one that let the student begin.

That last step mattered more than it looks. It meant the people deciding were people who knew the student, knew the program, and could pick up the phone. In 2026, the rules about what a school is even allowed to approve got much narrower, and the people closest to the student lost a lot of their room to say yes. I think it's the first domino in a chain that a lot of people outside higher education haven't noticed yet.

I'm not a policy researcher. I'm a data person who spent years watching international students move through university systems, and who was one of them. So this is the chain as I see it, with the best numbers I could find for each link, and sources at the end.

The first domino: the rules got narrower and less predictable

In August 2026, the government office that oversees student visas, SEVP, sent schools two pieces of guidance on CPT. The short version: CPT now has to be essential to finishing the degree, and it can't be authorized unless that practical training is required of every student in the program, international and American alike (Boston University ISSO, Ogletree). The same month, SEVP sent schools a reminder about the liability their officials carry when they authorize CPT (ICE broadcast message).

Read that from inside a university. A co-op that's optional, the kind many students choose because it's the best part of their education, may no longer qualify. An internship a student found on their own, in their field, at a company that wants them, may no longer qualify either, unless every student in the program is required to do the same kind of training.

That's the part that gets lost in the headlines. This isn't only a paperwork change for schools. It decides what work an international student is allowed to go to at all. Before, the question was whether a job fit the student's studies. Now, for many programs, the question is whether the program requires it of everyone, and if it doesn't, the answer is no, however good the opportunity. The students lose the experience that turns a degree into a career, and the employers who wanted them lose the hire. And the officials who approve CPT have just been reminded, in writing, of the liability they carry. The rational response is caution, and caution in this pipeline means fewer approvals. Some universities paused or narrowed CPT approvals while they worked out what the guidance meant. Harvard's international office, for one, halted CPT authorizations for several of its schools (The Harvard Crimson).

Then there's the bigger rule. In July 2026, the Department of Homeland Security published a final rule ending "duration of status," the system that lets a student stay for as long as they're enrolled in good standing. It would have replaced that with fixed admission periods, generally capped at four years, after which students would have to ask USCIS, the federal immigration agency, for an extension, with no guarantee of approval. It would also have cut the grace period after graduation from sixty days to thirty. A federal judge blocked it on September 14, 2026, one day before it took effect, and the case is ongoing (Day1CPT summary, Auburn University).

So today the rules are partly tightened, partly frozen in court, and entirely uncertain. For a student deciding whether to spend four years and a lot of money here, uncertain is often enough to say no.

The second domino: fewer students come

They already are saying no. In fall 2025, new international enrollment at US colleges fell 17%, and international graduate students fell 12%. Of the institutions that reported declines, 96% named visa concerns as a reason (IIE Fall 2025 Snapshot).

The visa numbers point the same way. F-1 student visa issuances fell by about a third in 2025 compared with 2024, and visas issued to students from India dropped by 57% (ICEF Monitor). For fall 2026, NAFSA projects as many as 111,000 fewer international students than the year before (NAFSA Fall 2026 Outlook).

Pushing from the other side: AI and the layoffs

If the visa rules were the only problem, a strong job market might still make the gamble worth it. It isn't strong, and AI is a big part of why.

Tech layoffs in 2026 passed the total for all of 2025 before September was over: 128,536 people at 299 companies by September 10, against 122,606 for the whole previous year, according to layoffs.fyi (Futurism). A growing share of those cuts are being explained, at least publicly, as AI.

The pressure is sharpest at the start of a career. Stanford's Digital Economy Lab finds that employment for 22 to 25 year olds in the occupations most exposed to AI, like software development and customer service, is now about 19% below where it would be if it had kept pace with their peers in less exposed jobs, up from a 15% gap a year earlier. To be fair to the data, the same researchers don't see widespread job loss across the whole economy. The damage is concentrated, and it's concentrated on exactly the entry-level, early-career roles that graduates compete for (Stanford Digital Economy Lab).

Students have noticed. In a May 2026 CNBC and SurveyMonkey survey, 65% of students said AI is taking away job opportunities for entry-level workers, and 62% said AI has made college less worthwhile (CNBC and SurveyMonkey).

Now do that math as an international student. You'd be paying full tuition, often with family savings or loans. You'd be counting on an internship you may no longer be allowed to take, then on finding a job within the short window your work authorization allows after graduation, then on an employer willing to sponsor a visa, all in the part of the job market AI is squeezing hardest. Every one of those steps was already uncertain. Stacked together, they turn a hopeful plan into a bet a lot of families can't afford to lose. Hesitating isn't irrational. It's the sensible response.

There's an irony in this I can't get past. Some of the most valuable AI companies in America right now, including Anthropic and OpenAI, have immigrant founders, according to the same NFAP research I cite below. The technology making the job market harder for the next generation of international students was built, in part, by people who once came here the way they're trying to.

The third domino: university budgets, right when domestic students are shrinking too

Here's the part that makes the timing so bad. American high school graduating classes peaked in 2025, at about 3.9 million students, and are projected to fall to about 3.37 million by 2041, a 13% decline (WICHE, Knocking at the College Door). Universities were already bracing for fewer domestic students. International students were part of how many planned to absorb that.

Because so many international students pay the full sticker price without need-based aid, their tuition quietly helps pay for aid that goes to domestic students. In the 2024 to 2025 academic year, international students contributed $42.9 billion to the US economy and supported about 356,000 jobs (NAFSA). NAFSA's fall 2026 projection puts the loss at up to $3.4 billion and nearly 40,000 jobs, driven mostly by graduate enrollment (NAFSA Fall 2026 Outlook).

When that money shrinks, it isn't abstract. It's sections that don't get staffed, programs that get cut, and financial aid budgets that get tighter for everyone.

The fourth domino: graduate programs and research

Graduate programs feel it first, because that's where international students are most concentrated. In 2023, temporary visa holders earned 34% of all US doctorates, and 86% of those were in science and engineering (NSF NCSES). In 2019, they earned 59% of doctorates in computer science and 58% in engineering.

Those students aren't just enrolled. They're the research assistants running experiments and the teaching assistants running labs and recitations for undergraduates. A lab that can't fill those roles does less research, and a department that can't fill them teaches fewer American students, too.

The fifth domino: the skilled workforce

The people who come to study here tend to stay highly educated. Among college graduates, 60% of immigrants hold at least a master's degree, compared with 53% of US-born graduates, and 51% of immigrant graduates' degrees are in STEM and health fields, compared with 36% for the US born (Migration Policy Institute).

Those are the fields the country keeps saying it can't hire enough people for. Fewer international students today means fewer of those graduates in five years, and there's no domestic pipeline growing fast enough to replace them, especially with the domestic college-age population shrinking.

The sixth domino: innovation

Immigrants are about 16% of US inventors but produce 23% of US patents, and once you count how much they raise the output of the American colleagues they work with, they account for about a third of aggregate US innovation (Bernstein et al., NBER). Their share is highest in the fast-moving fields: computing, communications, electronics, and medicine.

The last domino: companies and jobs

And then the companies. Immigrants are about 80% more likely to start a business than people born in the US, and the firms they start create more jobs relative to population (Azoulay, Jones, Kim and Miranda). As of April 2026, immigrants founded or cofounded 59% of America's privately held billion-dollar startups, 455 of 775. Almost one in four of those companies has a founder who first came to America as an international student, and on average they've each created more than 800 jobs (NFAP, June 2026).

I'm a very small data point in that story. I came here as a student. palStack only became a company after I had a green card, because a work visa tied me to a single sponsoring employer. It's a small company, not a unicorn. But it exists, and it wouldn't if the path in had been closed.

Why it's a chain, not a list

Each of these could be read as a separate statistic. I don't think they are. It's a cascade. A student who doesn't come this year isn't in a lab in three years, isn't in the workforce in five, and isn't founding a company in ten. The dominoes fall slowly, which is exactly why they're easy to ignore, and why they're so hard to stand back up.

What I keep coming back to is the thing I saw from inside that CPT pipeline. The rules themselves matter, but the bigger damage is to trust. A student deciding where to spend the most important years of their education doesn't need every rule to be generous. They need the rules to be stable, and the people closest to them to have the authority to help. When that goes, they don't argue. They just go somewhere else, and the first domino quietly tips the next one.