Trump’s Deportation Machine Is Being Built Through Budgets, Beds and Deputies

The infrastructure of mass deportation does not look like a secret bunker. It looks like a contract, a county sheriff’s memorandum of agreement, a warehouse lease, a budget line and a shortened legal form. That is what makes the current immigration crackdown so easy to misread. The danger is not that the Trump administration has discovered some entirely hidden power outside the law. The danger is that it is combining lawful tools at a scale and speed that can make legal accountability arrive too late.

I think that is the right way to understand the second Trump administration’s immigration project. It is not just tougher enforcement. It is an attempt to build a semi-permanent deportation apparatus whose main parts are visible in plain sight: more detention beds, more local police deputized into federal immigration work, wider use of expedited removal, and weaker practical oversight at the exact moment the system is growing.

Start with the legal baseline, because it matters. The strongest defense of the administration’s approach is not frivolous. Congress did authorize several of these tools. Section 287(g) of the Immigration and Nationality Act allows federal officials to enter written agreements with state and local agencies so trained officers can perform immigration functions related to investigation, apprehension or detention, under federal supervision and consistent with state and local law, according to 8 U.S.C. § 1357⁴. Expedited removal is also statutory. Under 8 U.S.C. § 1225³, certain people deemed inadmissible can be removed “without further hearing or review,” with limited asylum-related screening and narrow review channels. Federal law also directs the government to remove people after final removal orders and generally to detain them during the removal period, as 8 U.S.C. § 1231⁵ provides.

There is also a real enforcement backlog. A Federal Register public-inspection document said that, as of November 24, 2024, there were 1,445,549 people on ICE’s non-detained docket with final orders of removal⁶. ICE’s fiscal 2024 annual report said Enforcement and Removal Operations removed 271,484 people in FY2024⁷, while the non-detained docket grew from more than 6.1 million to more than 7.6 million. A government that cannot execute final judgments has a legitimacy problem of its own. If immigration courts issue orders that the executive branch cannot carry out, law starts to look like advice.

But this is where the defense becomes too neat. Formal legality is not the same thing as operational accountability. A tool can be authorized by statute and still become dangerous when it is expanded beyond the system’s capacity to check mistakes. A court order matters only if a person can reach court before removal. A detention standard matters only if inspectors, lawyers and lawmakers can see conditions in time. A training requirement matters only if thousands of deputized local officers are actually supervised in the field.

The administration’s own order shows the design. On January 20, 2025, Trump signed Executive Order 14159, “Protecting the American People Against Invasion.” It instructed agencies to use lawful means to enforce immigration law against inadmissible and removable people, prioritize final removal orders, use expedited removal authorities, expand detention facilities and broaden federal-state agreements under 287(g), according to the American Presidency Project’s text of the order¹. The order told DHS to allocate legally available resources or establish contracts to “construct, operate, control, or use facilities” for removable people and to authorize qualified state and local officials to perform immigration-officer functions under federal direction.

That is not campaign theater. It is an implementation manual.

Congress then supplied the fuel. The Brennan Center reported that lawmakers approved a roughly $75 billion cash injection for ICE in July 2025⁸, including $45 billion for new immigrant detention centers, which more than tripled ICE’s annual budget. The Council on Foreign Relations described the broader package as about $170 billion in new funding for border and immigration enforcement⁹. Money of that size does not merely fund operations. It creates institutions: beds, contracts, transport systems, staffing pipelines, local dependencies and private-sector constituencies.

The detention numbers show the shift. Vera’s April 2026 analysis of ICE detention data found that the daily detained population reached more than 73,400 people in mid-January 2026¹¹, a record high. Vera also found that ICE was detaining people in 456 facilities in February 2026, excluding medical facilities, while the agency’s website acknowledged using only 220 facilities. That gap is the kind of fact that should stop the conversation for a moment. The public-facing map of detention was smaller than the operational reality.

The network is not just bigger. It is more diffuse. DHS said in September 2025 that 287(g) partnerships had increased from 135 agreements to more than 1,000, and ICE’s own public page, captured in search results, listed 1,036 memorandums of agreement across 40 states as of October 1, 2025. FWD.us, an immigration and criminal-justice advocacy organization, later estimated that by January 26, 2026, there were 1,372 agreements across 1,169 agencies¹⁵. One need not accept every advocacy conclusion in that report to see the institutional point: federal immigration enforcement is being embedded into local policing on a scale that makes it harder for any single court, mayor, state legislature or protest movement to constrain the system as a whole.

The most important legal lever is expedited removal. The phrase sounds bureaucratic, but the mechanism is stark. In ordinary removal proceedings, a person can present a case before an immigration judge. In expedited removal, an immigration officer can order removal without that full hearing for certain categories, subject to limited exceptions and screening, especially for asylum claims. The Congressional Research Service explained that Trump’s January 2025 order called for expanding expedited removal “to the fullest extent authorized by Congress,” and that litigation challenged the expansion’s application to people inside the United States on due-process and statutory grounds, according to CRS².

That interior use is the crux. If expedited removal stays near the border and applies to recent arrivals, its risks are still serious but more bounded. If it applies broadly inside the country to people who cannot prove two years of continuous presence, the sorting problem changes. People with weak paperwork, limited English, trauma, poor legal access or complicated status histories can be routed through a fast track before anyone independent has checked the facts. The issue is not whether every removal is wrongful. The issue is whether the system is being designed so that wrongful cases can be identified before they become irreversible.

The oversight picture makes that worry sharper. In March 2025, DHS delivered reduction-in-force notices to the majority of staff in the Office for Civil Rights and Civil Liberties, the Citizenship and Immigration Services Ombudsman and the Office of the Immigration Detention Ombudsman, according to GAO¹². GAO did not find an Impoundment Control Act violation, but it also noted DHS had not provided requested factual and legal information. In May 2025, DHS said the offices would not be abolished after litigation pressure, according to Public Citizen¹³. Then, on May 5, 2026, NOTUS reported that DHS had confirmed it had shut down the Office of the Immigration Detention Ombudsman¹⁴, with the office empty, its website archived and inspections ending.

That sequence matters because detention is where legal theory meets human dependence. People in custody rely on the government for medical care, communication, access to counsel, transfer information and basic safety. Brennan reported that 2025 was the deadliest year for people in ICE custody in two decades, with at least 32 deaths⁸. NOTUS reported that, from Trump’s second inauguration through early May 2026, ICE data showed 49 detainees had died in custody¹⁴. These numbers do not prove every death was caused by policy. They do prove that the oversight burden is rising, not falling.

The counterargument deserves its full weight. A country can reasonably decide that final removal orders should mean something. The immigration system cannot function if legal outcomes are unenforced for years. Detention capacity, transport, hiring and state-local cooperation are not inherently sinister. Courts still exist, and some policies have been blocked. CFR reported, for example, that a federal judge in February 2026 ruled the administration’s third-country deportation policy unlawful on statutory and due-process grounds, after the administration had used that approach to send migrants to countries other than their own, according to CFR¹⁰.

I take that seriously. But it does not answer the institutional problem. After-the-fact court review is not the same as timely process. If a person is transferred, detained in poor conditions, moved again, denied meaningful counsel access, or removed before a judge can see the case, the later existence of judicial power is a weak remedy. The question is not whether the legal system has vanished. It has not. The question is whether the enforcement machine can move faster than the legal system can correct it.

Right now, the answer is yes, or close enough that the burden should shift to the administration. If officials want the public to believe this buildout is lawful enforcement rather than visible shadow infrastructure, they should publish audited case-level data showing who is being detained and removed, how many removals involve final orders versus expedited proceedings, how often people obtain counsel, how often credible-fear claims are made and sustained, how many U.S. citizens or lawful residents are mistakenly detained, how quickly members of Congress and independent inspectors can enter facilities, and what happens after complaints are filed.

Absent that evidence, I do not see a normal enforcement reset. I see a deportation system being hardened into place before the country has decided what level of error, secrecy and coercion it is willing to tolerate.

The indicator to watch next is not the loudest raid. It is the ratio between capacity and review: detention beds, 287(g) agencies and expedited-removal orders on one side; immigration judges, counsel access, facility inspections and published error data on the other. If the first column keeps growing while the second stays flat or disappears, the machine will not need to hide. It will already have won.