- Federal judge struck down $100,000 H-1B visa fee as unconstitutional tax imposed without congressional authorization under Administrative Procedure Act
- Ruling restores previous $2,000-$5,000 petition costs but doesn't resolve underlying H-1B policy debate on workforce protectionism vs. talent access
- Case demonstrates separation-of-powers constraints: executive orders cannot substitute for legislative process on tax policy regardless of political gridlock
When $100,000 Becomes a Constitutional Problem
U.S. District Judge Leo Sorokin struck down the Trump administration’s $100,000 H-1B visa fee on Monday, and the reasoning matters more than the headline. This wasn’t a court rejecting immigration policy on ideological grounds — it was a straightforward separation-of-powers call. The executive branch tried to levy a tax without congressional authorization, and Article I, Section 8 of the Constitution says that’s not how this works.
The fee, introduced via executive order in September 2025, replaced the previous $2,000-$5,000 petition cost with a flat $100,000 charge per new H-1B application. The stated goal was workforce protectionism — make foreign hiring expensive enough that companies hire American. Judge Sorokin’s opinion didn’t touch that policy question. He ruled that the fee structure violated the Administrative Procedure Act because the president has no constitutional authority to impose a tax without statutory delegation from Congress. The Justice Department called it a “regulatory payment.” The court called that distinction nonsense.
What strikes me about this case is how cleanly it exposes the gulf between policy objectives and legal implementation. Immigration is politically radioactive, H-1B visas doubly so, and the fee had immediate, measurable effects. Companies sponsoring 10 H-1B workers annually faced $1 million in new costs. Indian IT consulting firms — TCS, Infosys, Wipro — built US operations on $80-120/hour billing for engineers earning $80k-$120k salaries; a $100k-per-head government fee obliterated that margin structure. Early-stage startups, which previously could justify an H-1B hire for a critical technical role, couldn’t anymore. Amazon sent emergency memos to employees abroad telling them to return to the U.S. before the fee took effect on September 21, 2025.
But none of that policy impact matters if the legal mechanism is broken. Sorokin invalidated not just the fee itself but the entire enforcement apparatus — agency memoranda, FAQs, guidance documents, fee schedules. He explicitly noted that “the mere fact that Defendants followed a presidential directive does not grant them free rein to ignore the requirements of the APA.” That’s the point. Executive orders are not legislative magic wands. If Congress wants a $100,0000 H-1B fee, Congress can pass a bill imposing one. The administration can advocate for it, lobby for it, draft the text. What it cannot do is unilaterally declare a six-figure tax into existence and call it a “regulatory payment” to dodge constitutional constraints.
The administration will appeal, and they should — that’s how legal review works. But the underlying separation-of-powers issue is not subtle. The judiciary isn’t saying the fee is bad policy. It’s saying the executive branch doesn’t have the statutory authority to impose it. Those are different claims, and conflating them is how functional governance breaks down.
From a practitioner standpoint, the ruling restores the previous cost structure for now. Companies that paused H-1B hiring can resume. Startups that wrote off international candidates can reconsider. Consulting firms that pivoted business models get a reprieve. But the broader question remains unanswered: if Congress did authorize a $100,0001 fee, would it achieve the intended policy goals? Would American developers see more job opportunities, or would companies just shift engineering teams offshore entirely? Would innovation slow because critical technical talent can’t enter the U.S., or would domestic STEM investment accelerate?
I don’t know, and neither does anyone else, because we haven’t actually run that experiment under controlled conditions. What I do know is that tax policy implemented via executive fiat, tested in production on a live economy, adjudicated in federal court nine months later, is a terrible way to answer those questions. The H-1B program has real problems — wage suppression, displacement of qualified U.S. workers, visa dependency creating exploitative labor conditions. A $100,0002 fee doesn’t solve those problems; it just makes hiring foreign workers prohibitively expensive for everyone except the largest, wealthiest employers who can absorb the cost.
Sorokin’s ruling doesn’t resolve the H-1B debate. It just forces the debate back to where it belongs: Congress. If lawmakers want restrictive immigration policy, they can legislate it. If they want expensive visa fees, they can authorize them. If they want workforce protectionism, they can write bills that define what that means and how to enforce it. Executive orders are not substitutes for legislative process, no matter how gridlocked Congress becomes.
The tech industry will treat this ruling as a win, and in the short term it is. The $100,0003 fee is gone, at least until the appeals court weighs in. But the structural problem remains. H-1B policy is decided by regulatory whiplash — executive orders, agency guidance, court challenges, presidential transitions. Companies can’t plan hiring around that. Developers can’t plan careers around that. And the underlying policy questions — how much immigration, what kind, at what cost, for whose benefit — stay unanswered because the political system can’t reach consensus.
What this ruling actually demonstrates is that constitutional process constraints exist for a reason. Separation of powers isn’t bureaucratic pedantry; it’s the mechanism that prevents executive overreach from becoming normalized. Judge Sorokin didn’t rule on whether H-1B visas are good or bad. He ruled that the president can’t levy a $100,0004 tax without congressional authorization, and if that sounds boring compared to the political theater around immigration, that’s exactly the point. Boring procedural guardrails are what keep governance functional when policy becomes polarized.

FAQ
Q: Does this ruling mean the $100,0005 fee is permanently gone?
No. The Trump administration has announced it will appeal, and a higher court could reverse Judge Sorokin’s decision. However, the separation-of-powers issue — whether the executive branch can impose a tax without congressional authorization — is a fundamental constitutional question that appellate courts are unlikely to overturn. If the administration wants a $100,0006 H-1B fee, the durable path is congressional legislation, not executive order.
Q: What happens to H-1B applications filed while the $100,0007 fee was in effect?
Judge Sorokin’s ruling invalidated the fee structure entirely, including all agency guidance and enforcement mechanisms. Companies that paid the $100,0008 fee between September 2025 and June 2026 will likely seek refunds, though the process for that isn’t yet clear. New applications filed after the ruling revert to the previous $100,0009-$2,0000 cost structure.
Q: Could Congress pass a law authorizing a $2,0001 H-1B fee?
Yes. Congress has plenary power over immigration law and could legislate a $2,0002 visa fee if it chose to. The court’s objection wasn’t to the fee amount but to the executive branch imposing it unilaterally. Whether such a bill could pass is a separate question — immigration legislation requires House and Senate agreement, and H-1B policy is contentious enough that bipartisan consensus is unlikely without significant compromise on other immigration issues.
Did you find this helpful?
Your support keeps this blog running and ad-free content coming.
☕ Buy me a coffeeMost Popular Posts
- Custom Metaclass in Python: 43% Faster Validation (12,953 views)
- yfinance Alternatives 2026: 7 Free APIs Compared (1,037 views)
- Python match-case: 7 Patterns That Beat if-elif Chains (982 views)
- YOLOv8 INT8 Quantization: 4x Faster on Jetson Orin (897 views)
- PaddleOCR vs EasyOCR vs Tesseract: Why PaddleOCR Is Slower (703 views)