Research Note
Researched
New ICE facilities running far under capacity (2026) — judicial checks, not slackening enforcement
US
Overview
A pattern visible across multiple states researched in May 2026: newly opened or expanded ICE detention sites are holding far fewer people than their contracted capacity, even as arrests surge. The gap is driven by judicial checks (habeas releases, bond-hearing orders, Zadvydas indefinite-detention limits) and intake bottlenecks — not by any slackening of enforcement. Several of these facilities run on a revenue-dependent model (per-diem federal payments projected into county budgets), so under-population is also a fiscal problem for the host jurisdiction.
The Data Points
- KS — Leavenworth (CoreCivic): opened mid-March 2026, holding ~240–249 of a 1,104-bed capacity. Partly attributed to Judge Lungstrum’s 23+ Zadvydas habeas releases (see kansas-lungstrum-zadvydas-habeas-releases).
- IN — Miami Correctional: ~550 detainees against an up-to-1,000-bed contract; feds ~5 months behind on payment ($291/bed/day, ~4x state-inmate cost).
- IA — Woodbury County LEC: ~41 federal inmates against a 125 goal — a third of target, straining the revenue-dependent budget the county built the jail on.
- MS — Adams County CC (CoreCivic): population collapsed ~50% in three weeks in April 2026 (~2,100 → ~1,400), units vacant, with no WARN/closure notice — cause unexplained (transfers, quiet wind-down, or pre-refill pause).
- ND — Burleigh-Morton / Grand Forks: steady ICE populations far below jail bed counts.
Why It Matters
- Capacity ≠ population. The buildout (see national-detention-buildout-strategy-2026) is adding contracted beds faster than ICE can fill them under current legal constraints — so a facility’s contracted capacity overstates how many people are actually being held.
- Judicial checks are materially shrinking detention in pro-bond circuits even where facilities exist — the population data is downstream of the bond-hearing-circuit-split-2026. In Sixth-Circuit and Zadvydas states, releases visibly depress headcounts; in Fifth/Eighth-Circuit states they don’t.
- The revenue model is fragile. Counties that financed jail expansions on projected ICE per-diems (Woodbury, Henderson NV earlier) face shortfalls when populations run under target — a pressure point that can reopen contract fights.
- Caution for the heatmap: a high facility/IGSA signal does not by itself mean a facility is full or even active; pair bed-count signals with population/contract- status data where available.
Open Questions
- Is the Adams County MS collapse a transfer redistribution (to where?), a wind-down, or a pause before re-fill? Highest-value unknown.
- Track whether under-capacity facilities become the destinations for transfers OUT of pro-bond circuits (the forum-shopping mechanism in bond-hearing-circuit-split-2026).
Sources
- kansas-lungstrum-zadvydas-habeas-releases (KS — habeas releases depressing Leavenworth)
- miami-correctional-facility-in (IN — under-capacity + payment dispute)
- woodbury-county-lec-ia (IA — revenue model underperforming)
- adams-county-correctional-center-ms (MS — population collapse)