Issue 068 - Public administration - Operational throughput
How many detention beds does an enforcement surge require?
AP reported in 2026 that immigration officials planned a major detention expansion, including regional processing centers and large detention centers. Consider a hypothetical operation with 50,000 people entering detention each month.
Public systems and economicsAbout 1 minute
Sources checked October 2, 2026. Figures and circumstances may have changed.
The problem
If 50,000 people enter detention each month, about how many detention beds are needed to sustain that pace?
Then ask what happens if processing delays double the average length of stay while admissions remain unchanged.
Because Fermi problems target an order of magnitude, I normally use no more than two significant digits and write most calculations in scientific notation; the Fermi reference explains both conventions.
Grounding facts
The central model is a stock-flow relationship: the number of occupied beds equals the admission rate times the average length of stay.
beds needed
~= admissions per day x average days detainedICE's detention data are published as downloadable tables, and independent trackers using those data show average lengths of stay often around several dozen days, with many recent monthly values around 45 to 55 days before a partial July 2026 value.
For facility scale, CRS summarized that ICE operated in more than 200 facilities and had roughly 70,000 beds, while AP reported a planned expansion to about 92,600 beds. AP also reported proposed regional processing centers of 1,000 to 1,500 people and large-scale detention centers of 7,000 to 10,000 people.
After checking sources
Checked answer and calculation
Turn the monthly inflow into a daily inflow:
admissions per day
~= 50,000 / 30
~= 1.7 x 10^3 people/dayUsing a rounded 50-day average detention stay:
beds needed
~= 1.7 x 10^3 people/day x 50 days
~= 8.5 x 10^4 beds
~= about 80,000 bedsIf the average stay doubled to 100 days while admissions stayed the same:
delayed-processing beds
~= 1.7 x 10^3 people/day x 100 days
~= 1.7 x 10^5 beds
~= about 170,000 bedsThat is an additional 80,000 to 90,000 beds. In facility terms, that extra capacity is roughly:
extra regional centers
~= 85,000 beds / 1,500 beds per center
~= 60 centersextra very large centers
~= 85,000 beds / 8,000 beds per center
~= 10 centersSo capacity depends on both admissions and case resolution, but length of stay is the multiplier people often miss. Double the average stay, and you double the required bed capacity even if the monthly enforcement pace does not change.
Before checking sources
Matt's original estimate
This is the unverified estimate Matt wrote before checking sources, not the checked answer.
I don't have a great intuition about how long it should take someone to be fully processed after detention, but I've heard stories in the 3-6 month range - those might be outliers, but I'm going to use 3 months as the average time period counting processing delays.
If it takes the average person 3 months to work through processing, that means we should expect the maximum capacity necessary to be around 150,000 beds.
I don't know what counts as a facility - when I was in the army, a barracks building might have 30 rooms per floor, 4 floors, and 2 soldiers per room, for about 240-250 soldiers per barracks building. If they're instead erecting soft cover structures and detainees are sleeping on cots in an open bay-style covered room, they might be able to contain 60-100 cots in a single covered dwelling structure.
If that's the available range for structures, capacity of 100-250 per structure, at max capacity it would require about 600 to 1,500 structures.
Reasoning score
Matt's reasoning score: 70 / 100
Higher is better: earn points for useful facts, a sound reasoning approach, correct math, and a final estimate close to the sourced answer. The owl meter shows percent full of it: 100 minus the reasoning score.
Useful facts: 10/30. The average-stay assumption was high by roughly a factor of two, and the facility-size intuition was much smaller than the large facilities described in recent expansion plans.
Reasoning approach: 30/30. The core stock-flow model was right: monthly entries multiplied by months or days in detention.
Math: 10/10. The arithmetic followed correctly from the stated assumptions.
Final estimate: 20/30. The 150,000-bed estimate was high but still within an order of magnitude, and it correctly showed that detention capacity is highly sensitive to processing delay.
Post-check reflection
Matt's reflection
Looks like the average stay is about 1/2 to 1/3 of what I guessed, close but not quite right. Also I hadn't considered large venues for housing as many as 1,000 people, which would drop the total number of required structures down considerably - but I don't think those can be considered the "rule" and are less common.
Ultimately, it is a pretty large investment to accomplish this, but not an unthinkable one.
Recommended memory peg
For detention, hospitals, shelters, and similar systems, remember: beds needed ~= daily entries x average length of stay. Length of stay is the hidden multiplier.
Reader results
Bars show how submitted estimates sort into the answer choices from the gut-check prompt.