The Proposed $70K OPT Fee: What It Would Do to Entry-Level Engineering Hiring

Last updated: October 9, 2026

By Mike Carter, Managing Director, KORE1

DHS has proposed charging schools $70,000 the first time they recommend an F-1 student for Optional Practical Training, plus $30,000 for each later recommendation. Nothing is in effect. Comments close November 9, 2026.

That is the whole mechanism, and it is worth reading twice, because almost every headline about it has the payer wrong. The fee does not land on the student. It does not land on you. It lands on the university, which the rule then explicitly permits to hand the bill to international students, to all students, or to employers.

A VP of engineering at a semiconductor capital-equipment supplier in San Jose called me on October 8, about four hours after the proposed rule hit the Federal Register. Eleven open engineering reqs. Four of them entry-level, and two of those four already had verbal offers out to master’s candidates whose OPT start dates were set for January. The question was not about policy. It was whether to pull the offers.

No. Not yet, and probably not at all.

Most of my week at KORE1 goes to hardware, semiconductor, and medical device companies filling seats through our engineering staffing agency, so I have a direct stake in you continuing to hire. Fine. The numbers below are all from DHS, NSF, and BLS, and every one of them is linked, so you can run the arithmetic yourself and reach whatever conclusion it supports.

What the Rule Actually Proposes

DHS published the notice of proposed rulemaking on October 8, 2026, under docket ICEB-2026-0100. It amends 8 CFR Part 214 and does one narrow thing: it attaches a price to the act of a designated school official recommending a student for OPT in SEVIS.

Read the refund clause closely if you do nothing else. A school pays before the recommendation, and once the employment authorization document issues, that money is gone under any circumstances, and the determination on a refund request is not subject to administrative appeal. For a public university with a 40-student OPT cohort, that is $2.8 million of non-recoverable exposure committed in advance of an immigration adjudication nobody at the university controls.

The full text sits on the Federal Register site. It runs 35 pages. The regulatory text itself is about a page and a half of that, which tells you where the argument is.

Why This Lands on Engineering Harder Than Anywhere Else

Because that is where the people are.

The National Science Foundation’s 2026 State of U.S. Science and Engineering report puts temporary visa holders at 42% of all science and engineering master’s degrees awarded in 2024, and at 54% of engineering doctorates, 61% of computer and information sciences doctorates, and 52% in mathematics and statistics. Those are not edge cases at the margin of a graduate program. In a lot of departments that is the program.

The pipeline numbers line up the same way. The Institute of International Education’s Open Doors 2025 release counted 294,253 students on OPT in 2024/25, up 21% in one year, with math and computer science at 26% of all international students and engineering at another 18%.

Now put DHS’s own participation estimates next to that. The agency reviewed OPT authorizations with 2022, 2023, and 2024 start dates and found 2,478 schools with students in the program. Its high scenario, which assumes nothing changes, uses 194,554 initial OPT and 95,384 STEM OPT participants. Its low scenario assumes something very specific and very revealing: that schools stop recommending non-STEM students entirely, because $70,000 for twelve months of work authorization is a bad trade while $100,000 for up to three years is a defensible one.

DHS is projecting, in its own regulatory impact analysis, that the rule prices non-STEM OPT out of existence and leaves engineering and CS as the only fields where the math still works. Fifty-six percent of OPT participants took a STEM extension the following year, under the 2021 to 2024 data the agency used. Those are your candidates. Your competitors’ candidates too.

The Arithmetic on One Entry-Level Engineering Seat

Set the policy argument aside and price a single req.

If the full $70,000 reaches the employer, a first-year engineering hire at the national median stops being a $99,520 decision and becomes a $169,520 one before benefits, equipment, or a mentor’s time. That is a 70% increase in the cost of the least expensive person on your team. Mid-level engineers do not cost 70% more than entry-level engineers.

Which is the part worth sitting with. A fee designed to protect U.S. workers, applied to a stack of engineering reqs, makes the junior seat the single worst value on the org chart. Hiring managers respond to that the way they respond to any cost signal. They stop writing junior reqs and write a senior software engineer one instead, and then they compete with you for a candidate who does not exist in sufficient numbers, which is the loop we wrote about when the entry-level market cratered on its own, without help from a rulemaking.

We have watched this cost math play out on H-1B for a year now. The $100,000 proclamation fee from September 2025 has been stopped in court twice, by injunction in Massachusetts and then by a judge in the Northern District of California who vacated it outright on September 30, 2026. A separate USCIS rule proposing a $103,265 cap-subject fee is still pending. None of it is being collected. All of it is already shaping reqs, because budget holders plan against announced costs, not against collected ones.

What DHS Admits in Its Own Analysis

The regulatory impact analysis is unusually candid, and it is the strongest material available to anyone filing a comment.

On magnitude: annualized costs ranging from $32 million to $8.2 billion, with a primary estimate of $4.1 billion, and ten-year costs with a primary estimate of $35.1 billion at a 3% discount rate. Annualized transfers run $8.7 billion to $16.9 billion. These are not small-ball numbers for a fee rule, and DHS does not pretend otherwise.

On employers, in one sentence the agency wrote itself: employers may face productivity losses if they are unable to replace F-1 student OPT participants with other qualified workers. There is no analysis behind it. DHS says plainly that it cannot allocate the opportunity costs by stakeholder group, and uses the unpaid fees as a stand-in for the value of lost work.

Then there are the small schools. Fifty-six percent of regulated SEVP-certified institutions are small entities by DHS’s own count. The agency considered exempting them and declined, reasoning that fraud could still continue because most of them are small. Read that twice if it did not land the first time.

The paperwork burden runs to 10,800 designated school officials and roughly 12,390 administrators, at an estimated 65 hours each just to learn the rule and rebuild the process around it.

And DHS asks for comment on a harsher version. It floats re-charging the full $70,000 whenever a student moves to a higher educational level, so a bachelor’s OPT followed by a master’s OPT would cost $140,000 rather than $100,000. That alternative is in the rule as a question, which is the stage at which a comment can still move it.

What to Do About Your Open Entry-Level Reqs

I told the San Jose VP to keep both offers. Here is the reasoning, in the order I gave it.

1. Anything with a DSO recommendation dated before the transition date is untouched. The transition provision applies the fee to recommendations dated on or after 60 days past publication of a final rule. There is no final rule. Candidates recommended now are outside it.
2. Stop treating a proposed rule as a hiring freeze. An NPRM is a question the government is asking. This one had drawn 774 comments on Regulations.gov within its first day and will draw many thousands more before November 9, and the two closest analogues on the H-1B side are both currently blocked in federal court.
3. Ask your immigration counsel one question, not ten: which of our current and pipeline hires would carry a DSO recommendation dated after the transition date. That list is short and it is the only list that matters for budgeting.
4. Price the downside case into 2027 comp bands, not into 2026 req decisions. If a final rule lands and schools pass the cost to employers, the realistic response is a sponsorship cost line, not an empty seat. Budget it. Do not cancel the req against a cost nobody is charging yet.
5. Build the domestic entry-level bench in parallel, and do it this quarter rather than next. Not as a hedge against a rule. As the thing you should already be doing, which this rule has merely made urgent in a way a spreadsheet can see.
6. File a comment. Twenty minutes, Regulations.gov, docket ICEB-2026-0100. Agencies weight comments that contain specific data over comments that contain sentiment, and the one thing you have that no university or law firm has is a count of the engineering seats you filled this way and what happened to the work.

On that fifth point, contract and contract staffing structures do real work here, and not for the reason people assume. A contract engagement does not solve a work authorization problem. What it does is let you test whether a junior seat produces before you commit permanent headcount to it in a year when the authorization cost of one hiring channel is genuinely unknowable, and conversion into direct hire stays available once the regulatory picture settles.

For comp bands on the domestic side of that bench, our salary benchmark assistant will get you closer than a national median will.

The Window Closes November 9

Comments are due on or before November 9, 2026, at docket ICEB-2026-0100. Comments sent by email or letter will not be considered and will not receive a response, which the rule states twice. Separate comments on the information collection burden are due December 7.

After that, DHS reads the record and either issues a final rule, modifies it, or lets it sit. A final rule takes effect 60 days after publication. On the fastest realistic path, that puts first collections somewhere in 2027, assuming no injunction, which given the H-1B litigation history is a generous assumption.

We have placed engineering and IT talent for 20+ years across 30+ U.S. metros, with an average time-to-hire of 17 days on IT roles and 92% twelve-month retention, through H-1B cap years that opened and closed, a pandemic, and the better part of a dozen rulemakings that looked existential in October and were footnotes by March. This one is bigger than most. It is still a proposal.

The OPT proposal was not the only thing that moved that week. The same crackdown produced a PERM suspension at Microsoft, Adobe and six IT outsourcers, which reaches employers who sponsor nobody at all.

What Employers Are Asking Us This Week

Is the $70,000 fee being charged right now?

No. It is a notice of proposed rulemaking published October 8, 2026, and nothing is collected unless DHS issues a final rule, which would then take effect 60 days after publication. Plenty of vendors are already selling urgency against it. There is no fee today, and there may never be one in this form.

Would my company actually get the bill?

Only if the school passes it along, which the rule permits but does not require. DHS wrote that institutions may pass the obligation to F-1 students, to all students, or to employers, and declined to regulate how schools fund it. Expect that to vary enormously by institution. A large private university with a $4 billion endowment and a public land-grant school with a 300-student OPT cohort are going to land in different places, and some will simply stop recommending non-STEM students, which DHS assumed in its own low-participation scenario.

We have two offers out with January OPT starts. Pull them?

Keep them. The transition provision applies the fee only to DSO recommendations dated on or after 60 days past publication of a final rule, and no final rule exists. January is months inside that gap.

So a STEM engineer ends up costing $100,000 in fees?

$100,000 over three years, as the proposal is written. Seventy thousand for the initial twelve months, thirty for the 24-month extension. Odd detail, and it is the hinge of the whole rule. DHS built its low-participation scenario on exactly that spread, reasoning that a school weighing one year of authorization at $70,000 walks away, while a school weighing three years at $100,000 pays. Engineering and CS students are the ones who qualify for the extension. So the rule as drafted thins out every field except the ones you hire from.

Everyone keeps conflating this with the $100,000 H-1B fee. Same thing?

No. Earlier step in the hiring timeline, a different payer, and a much stronger legal footing. OPT is the work authorization a graduate uses before anyone files an H-1B petition, so this rule reaches candidates a year or two upstream of the lottery. The payer differs too. Schools here, employers there. What should worry you is the third part. The H-1B charge came out of a presidential proclamation, and courts have knocked it down twice, most recently in the Northern District of California on September 30, 2026, on the ground that the administration skipped notice-and-comment rulemaking. This one did not skip it. That is the point of the November 9 deadline.

If this passes, where do entry-level engineers come from?

Domestic programs, community college transfer pipelines, and apprenticeship structures that most engineering employers have never seriously staffed against. BLS projects architecture and engineering employment growing much faster than average through 2035, and 106,100 annual openings in software development alone. The supply side was already tight. Pricing out a chunk of the master’s-degree cohort does not create domestic candidates. It just moves who is competing for the ones who exist, and the companies that start building that bench now will spend 2028 hiring from it instead of bidding against everyone else for the same forty resumes.

Twenty minutes on a comment. Does that actually do anything?

Sometimes, and only when the comment carries data. DHS says in the rule which comments help it most. The ones that point at a specific provision, say what should change, and show the evidence. Sentiment gets counted. Evidence gets cited in the final rule’s response to comments, and that record is what a court reads later. Your entry-level engineering fill rate is evidence. So is the headcount you sourced this way and what those people shipped. No university in that docket has your numbers.

If you are deciding what to do with entry-level engineering reqs between now and a final rule, talk to our team. We can scope the contract or direct-hire path that keeps the seat filled while the rulemaking runs, and tell you honestly which of your open reqs this would actually touch.