Washington has spent a year rewriting both the price of an American tech job and the odds of getting one. For engineers in India, the practical result is that the route you take now depends far less on luck and far more on how senior you already are.

Ask a recruiter in Bengaluru what changed about American immigration in the last twelve months and you will almost certainly hear one number: one hundred thousand dollars. It is the figure that ran on every front page when the White House signed Proclamation 10973 in September 2025, imposing a $100,000 payment on certain new H-1B petitions for workers entering from outside the United States.

That number is the least useful thing to know right now.

Start with where it actually stands. A federal district court in Massachusetts vacated the policy implementing the fee on June 8, 2026, on four independent grounds, holding that the statutes behind the H-1B program say nothing about a payment requirement and that Congress would have to authorise a burden of that size clearly. Judge Leo Sorokin stayed his own ruling four days later, which briefly put the fee back in force, and the First Circuit lifted that stay on July 24. The merits appeal is still live. The government could still go to the Supreme Court on an emergency basis. The proclamation itself was never struck down, only the machinery for collecting under it, and it carries a twelve month term that runs out on September 20 unless it is renewed or reissued.

So as of this week, the fee is not being collected, and the instrument that created it is days from expiry. Anyone reading that as a reprieve should read the Federal Register instead.

On August 25, the Department of Homeland Security published a proposed rule that would charge $103,265 on every cap-subject H-1B petition, filed at the front end, on top of existing fees, including petitions filed under the advanced degree exemption for holders of American master’s degrees. Universities and non-profit research employers would be carved out. Everyone else would pay. Comments close on September 24, and the earliest realistic effective date is some point in 2027. DHS projects the rule would raise $8.8 billion a year and concedes in its own analysis that it would land hard on roughly 11,051 small entities.

The difference between the proclamation and the proposed rule is not the amount. It is the process. The first attempt failed because it was executive fiat dressed as a fee. The second is being built through notice and comment rulemaking, which is exactly the procedural armour the Massachusetts court found missing. Treat the current pause as a gap in the weather.

The change that matters more, and got a fraction of the coverage

While the six figure headline was moving through the courts, DHS rewrote the lottery itself. The final rule landed in the Federal Register on December 29, 2025 and took effect on February 27, 2026, in time for the FY2027 cap season. Selection is no longer random. Each registration now gets entries according to the Department of Labor wage level attached to the job: four entries at Level IV, three at Level III, two at Level II, and one at Level I.

DHS published its own modelling of what that does. Under the old random draw, everybody sat at roughly 30 percent. Under weighting, a Level IV registration carries around a 61 percent chance of selection, Level III around 46 percent, Level II around 31 percent, and Level I around 15 percent. The department further projects about 10,099 fewer Level I selections a year. Level I positions historically made up about 28 percent of H-1B petitions.

Level I is where junior engineers live. That is what the wage tier is for. It describes someone with basic understanding of the role who works under close supervision, which is an accurate description of a good engineer two years out of college and no insult to anybody.

Put the pieces together from the employer’s side of the desk, because that is where the decision is actually made. To sponsor a junior, a company registers at $215 a head, up from $10 two years ago, accepts roughly one in seven odds, pays counsel, and, if the DHS rule survives, writes a cheque for $103,265 at filing for a candidate who has not yet shipped anything for them. The same company can register a senior engineer at Level IV, take 61 percent odds on the same fee, or hire the junior in Bengaluru tomorrow with no visa exposure at all.

The market has already answered. H-1B registrations have fallen from roughly 780,000 for FY2024 to 470,342 for FY2025 to 343,981 for FY2026. New approvals dropped 37 percent in the most recent USCIS reporting period. Indian nationals still take about 71 percent of approvals, which tells you the pipeline is still overwhelmingly Indian; it is simply a much narrower pipe, and it now favours people who are further along.

The cap was still reached for FY2027 and there was no second lottery, so demand has not vanished. It has been sorted.

What still works: the intracompany transfer

The L-1 is the route that survives all of this, and it is worth understanding precisely, because most of the advice floating around Indian tech circles about it is wrong in the details.

The L-1 moves an existing employee from a foreign entity to a related American one. You need one continuous year of employment with the qualifying foreign company within the three years before the filing, and time already spent working in the United States does not count toward it. The two companies have to be genuinely related as parent, subsidiary, affiliate or branch. Your role abroad must have been managerial, executive, or specialised knowledge work, and the American role has to fit one of those categories too.

L-1A covers managers and executives and runs up to seven years. L-1B covers specialised knowledge and runs up to five. Both allow dual intent, meaning you can pursue a green card without endangering your status. Spouses on L-2 get work authorisation in their own right, which is a material difference from the H-4 situation that has caused so much grief in Indian households over the past decade.

Crucially, there is no lottery and no cap. And neither the vacated $100,000 proclamation nor the proposed $103,265 fee touches the L-1. Both target cap-subject H-1B petitions.

Approval rates hold up well. Recent government data puts L-1A around 91.4 to 91.8 percent and L-1B around 91.6 to 92.3 percent, with blanket L petitions filed by large qualifying multinationals at roughly 97 to 99 percent. What those figures hide is friction. Around one in four L-1B filings draws a request for evidence, which typically costs two to four months. Adjudicators consistently reject petitions that describe industry expertise rather than company specific expertise, that lean on job duties instead of documenting what the employee actually knows, or that cannot show what the company invested in building that knowledge. Organisational charts without payroll records behind them, and job descriptions that blend managing people with doing the work yourself, are the two fastest ways to trigger scrutiny on the L-1A side.

One honest caveat, because the approval numbers invite complacency. L-1B denial rates sat at 6 percent in FY2006 and reached 35 percent by FY2014 with no change in the underlying statute. Adjudication standards on specialised knowledge swing with the political weather, and nothing in the current posture suggests the L-1 is permanently off the table for tightening. Today’s approval rate is a forecast, not a guarantee.

Where this leaves a 25 year old engineer in Pune

Your visa strategy is now an employer strategy. That is the whole shift.

The infrastructure for it is already sitting in India. There are 2,117 global capability centres operating across 3,728 sites, employing about 2.36 million people and booking $98.4 billion in revenue, up from 1.9 million people and $64.6 billion in FY2024. Bengaluru alone holds more than 900 units. Every one of those centres is, by definition, the foreign entity in a qualifying relationship with an American parent.

Which means the practical sequence is not to win a lottery. It is to join a company with a real American parent and a demonstrable record of moving people, do work that is genuinely proprietary to that company rather than generic to the industry, keep documentation of the training and the systems you were built into, bank your qualifying year, and then ask. It is entirely reasonable to ask a hiring manager how many people that centre has transferred to the US in the last two years. The answer is informative whichever way it goes.

For anyone with ambitions beyond a few years abroad, one more piece of arithmetic deserves to be on the table. The National Foundation for American Policy reported in August that 996,599 people from India were waiting in the EB-1, EB-2 and EB-3 green card categories as of December 2025, about 79 percent of the entire employment based backlog. For someone filing today, NFAP estimates a wait of roughly 179 years in EB-2 and 38 years in EB-3. EB-1 comes in at four to five years.

Those projections assume current law and visa supply hold, and they count dependents against the quota, so treat them as a description of how badly the system is jammed rather than a literal forecast of anyone’s life. But the ranking is the point. EB-1C, the multinational manager category, is the one employment based path for Indian nationals that is measured in years rather than lifetimes, and it maps directly onto the L-1A. That is not a coincidence, and it is the strongest argument for pointing your career at managing a function rather than only at deepening a technical specialism.

None of this is a closed door. It is a repriced one, with a bouncer who now checks your title before your luck. The people who get through in 2027 will mostly be people who were already inside a company that happened to have an American address.