The $70 Billion ICE Bill Isn't About Immigration. It's About Building a Machine That Can't Be Turned Off.

Last night, the U.S. Senate voted 50-48 to adopt a budget resolution that begins the process of funneling another $70 billion to ICE and Customs and Border Protection¹ through reconciliation, bypassing the Democratic filibuster that has kept DHS partially shut down since February. The money is expected to fund the agencies for roughly 3.5 years, through the remainder of Trump's term. This comes on top of the approximately $170 billion in enforcement funding² already authorized by the One Big Beautiful Bill Act last July, which included $45 billion for detention expansion and nearly $30 billion for ICE enforcement operations alone.

I want to think carefully about what this means, because the headline number is the least interesting part of the story. The real question is structural: what kind of enforcement apparatus is being built, and how reversible is it?

Let me start with the money, because the scale is genuinely hard to process. ICE's annual budget hovered around $10 billion for years. The OBBBA effectively tripled it³, making ICE the highest-funded law enforcement agency in the country. According to the Brennan Center for Justice⁴, the $11.25 billion annual increase to ICE's detention budget alone exceeds the Department of Justice's entire budget request for the federal prison system, which holds 155,000 people. And now the Senate is moving to add another $70 billion on top of that — even as the Senate Budget Committee's own Democratic staff reports⁵ that ICE and CBP are sitting on more than $103 billion in unobligated funds from the OBBBA as of March. The agencies haven't spent most of what they were already given. Congress is now moving to give them more.

This is where the infrastructure argument gets interesting. Defenders of the spending make a legitimate point about enforcement capacity: immigration courts have a backlog approaching 4 million cases, the non-detained docket grew to over 6 million people, and deportation numbers swung from 409,000 under Obama to roughly 72,000 under Biden — all under the same statute. If you believe immigration law should be enforced consistently, there is a real argument that the system has been structurally undersized for the obligations Congress created. I take that argument seriously.

But here's where I break from the "just building capacity" framing. The mechanism of the spending tells you more than the amount. Nearly 90 percent of ICE detainees⁶ are already held in facilities operated by for-profit companies — principally GEO Group and CoreCivic. Trump's declaration of a border emergency has allowed ICE to largely bypass competitive bidding⁷ and award no-bid contracts to these firms. A 15-year contract with GEO Group⁸ for the Delaney Hall facility in Newark, New Jersey, is valued at $1 billion. CoreCivic's Dilley, Texas facility was reopened under a contract running through at least 2030. GEO Group's CEO told investors the budget reconciliation process was a "key element" in how fast the company can build up privately run detention capacity⁹.

This matters because of how private detention contracts typically work. A report from In the Public Interest¹⁰ found that roughly 65 percent of private prison contracts contain occupancy guarantee clauses — "bed guarantees" — requiring the government to maintain 80 to 100 percent occupancy or pay for empty beds. The most common guarantee is 90 percent. GEO Group and CoreCivic both use these clauses extensively in their contracts with federal and state governments. That means if a future administration wants to reduce the detained population, it doesn't just face political opposition. It faces contractual obligations to pay for capacity it isn't using. The cost of policy reversal is engineered into the contract structure itself.

Think about what that means for democratic governance. When a future president wants to change enforcement priorities — which is a core executive function recognized by the Supreme Court in Heckler v. Chaney (1985), holding that agency decisions not to enforce are presumptively unreviewable — they would inherit a system where empty beds still generate payments to private contractors. The economic incentives point in one direction: fill the beds, because you're paying for them either way. CoreCivic said it plainly in a 2014 annual report¹¹: demand for their services "could be adversely affected by the relaxation of enforcement efforts." Their business model depends on the government being structurally unable or unwilling to reduce enforcement.

The parallel expansion makes the picture even starker. Beyond the private contracts, ICE is now purchasing commercial warehouses¹³ and retrofitting them into detention centers — what Acting ICE Director Todd Lyons described, remarkably, as "Prime, but with human beings." The agency plans up to two dozen warehouse "mega centers" that could detain 5,000 or more people each¹⁴. DHS is purchasing Boeing 737s for deportation flights. Thousands of new agents are being hired with $50,000 signing bonuses¹⁵. The American Immigration Council reports¹² that with OBBBA funding, ICE could potentially operate 135,000 detention beds through FY 2029 — roughly the population of the entire federal prison system.

All of this is being built at remarkable speed, and the infrastructure carries its own momentum. Federal employees hired into this expansion will have civil service protections. Facilities purchased or leased will have long-term obligations. Political constituencies — the towns where these facilities provide jobs, the contractors who depend on federal dollars, the congressional districts that receive the spending — will resist any future drawdown. This is exactly what happened with post-9/11 fusion centers, which proved nearly impossible to dismantle even when they were demonstrably ineffective.

I want to be clear about my position. I am not arguing that immigration enforcement is illegitimate or that the enforcement system didn't need more capacity. There is a genuine enforcement gap that reflects a real policy failure. But the way this infrastructure is being built — through no-bid contracts to companies that donated nearly $2.8 million to Trump's campaign and inaugural fund¹⁶, through occupancy guarantees that penalize future administrations for reducing detention, through reconciliation that bypasses normal legislative debate, and at a scale that far exceeds what the agencies can even spend (they're sitting on $103 billion in unobligated funds!) — this is not capacity building. This is institutional capture.

The strongest counterargument is the precedent one: Democrats have used exactly the same reconciliation process to lock in their priorities, from the Inflation Reduction Act's climate spending to the Affordable Care Act's coverage expansion. Both parties build durable infrastructure to survive electoral cycles. That's true. But the IRA's tax credits are reversible through ordinary legislation (Republicans have already partially clawed them back). The key distinction is between programs where the cost of reversal is political — you have to beat the opposition in elections and pass new laws — and programs where the cost of reversal is contractual — you have to pay private companies for capacity you don't want to use. The first is democracy. The second is something else.

The real context for this $70 billion push is what triggered it: the DHS shutdown that began after ICE agents killed two American citizens in Minneapolis¹⁷ — Renee Good and Alex Pretti — in January 2026, prompting Democrats to refuse further ICE funding without reforms. Rather than negotiate guardrails, Republicans are using reconciliation to fund the agency without any Democratic votes at all, ensuring the enforcement machine not only continues but accelerates, with zero accountability provisions attached.

What to watch: the final reconciliation bill's text, expected by June. The critical details will be (1) whether it includes sunset clauses requiring reauthorization, (2) whether it restricts or permits occupancy-guarantee contracts, and (3) whether any oversight mechanisms survive the drafting process. My prediction: none of those safeguards will make it into the final bill. The entire architecture of this spending push is designed to make the enforcement apparatus a permanent feature of American governance, regardless of what future voters decide. That should worry everyone — including people who want strong immigration enforcement — because a system insulated from democratic correction is a system that cannot self-correct when it fails. And this system, which has already produced the deadliest year in ICE detention history and the killing of American citizens by federal agents, is failing.