Our data study of the federal award record found that federal contract obligations rose 40% between early 2024 and early 2026. Underneath that total, two line items moved harder than anything else in the dataset, in opposite directions, and both of them have a counterparty outside the United States.
They are usually told as one story: the money came off foreign aid and went onto the border. The award record says that is wrong on the sequence. The aid stopped in year one. The wall did not start until year two. Eighteen months separate them, and nothing moved from one to the other.
Year one was not a slowdown. It was a stop
Three matched February–May windows, contract obligations, from USAspending:
| Feb–May 2024 | Feb–May 2025 | Feb–May 2026 | |
|---|---|---|---|
| Agency for International Development | $1,899M | $4M | $448M |
| Department of State | $3,526M | $2,494M | $2,375M |
| Department of Homeland Security | $7,401M | $5,983M | $26,198M |
Four million dollars. Not four billion — four million, across four months, at an agency that had obligated $1.9B in the same window a year earlier and whose median month across the two preceding years was $541M. That is not a budget cut. It is a full stop, and it lands inside the first hundred days.
The monthly series is blunter still. February 2025 came in at −$155M. A negative obligation is not a typo or a rounding artifact: it is money released back off contracts that already existed — descoped, terminated, deobligated. In February 2025 the agency handed back more than it committed. November 2025 was $0M exactly.
The obvious objection is that the work simply moved to the State Department, which formally absorbed what survived. In contract dollars, it did not. Take the two agencies together and the pair went from $5,425M to $2,823M — down 48% — over a period when total federal contracting rose 40%. There is no third agency where this reappears.
Then July 4th, and the money came back pointed somewhere else
Homeland Security is the mirror image, and it runs on a different clock. DHS obligations fell 19% in the first year, from $7.4B to $6.0B. Its construction line fell 42%, from $431M to $250M. Whatever was going to happen at the border had not started.
Then the One Big Beautiful Bill Act was signed on 4 July 2025, appropriating $46.5B for border-barrier construction, $45B for detention facilities and $24.2B for the Coast Guard. Compare the same month a year apart — 30 September is the federal year-end, and mixing it with any other month measures the calendar rather than the policy. September 2024: DHS obligated $4.8B. September 2025: $12.1B.
By the February–May 2026 window, DHS stood at $26.2B across all sectors, and its construction spending had gone from $431M to $15,254M. That figure now exceeds the Department of Defense's own construction total of $14,741M. For the first time, the largest builder in the federal government is not the Pentagon.
An appropriation did this, not an executive order — and appropriations run to fiscal 2029.
A barrier with a treaty problem
Five of the ten largest federal construction recipients in early 2026 appear nowhere in the hundred largest of early 2024. Two of them matter here: Spencer Construction at $2,099M and Cochrane USA at $641M, which between them hold roughly $1B for a water-borne barrier in the Rio Grande — floating buoys rather than steel bollards.
The Rio Grande is not a domestic construction site. It is an international boundary governed by treaty, and Mexico has filed formal complaints arguing that a floating barrier violates Article 17 of the 1944 Water Treaty and Article IV B.1 of the 1970 Boundary Treaty, both of which require the two governments to agree before either builds anything capable of diverting the river. DHS announced an expansion to more than 500 miles of the river in January 2026. Flooding has since pushed one-ton buoys downstream past Eagle Pass toward Laredo, closing international bridges.
Our data can tell you that the money was obligated and to whom. It cannot tell you how the treaty question resolves. What it does tell a contractor is that a line item this size now sits on top of an unresolved diplomatic dispute, which is a schedule risk with a name.
The missiles that were not in the top twenty
The other side of the international ledger is armament, and it moved harder than the border did.
The product-and-service code GUIDED MISSILES does not appear among the twenty largest codes of February–May 2024 — the twentieth was $1.9B, so it was smaller than that. In February–May 2026 it is $15,755M, the largest single code in the federal government. In industry terms, Guided Missile and Space Vehicle Manufacturing went from $6,679M to $23,942M, up 258%. Raytheon moved from 26th place government-wide to third; Electric Boat from 16th to fourth.
The obvious frame is Golden Dome, the $185B missile-defence architecture with a 2028 demonstration target and $3.2B of prototype agreements placed with twelve firms. We are not going to claim that attribution: much of that work runs through existing programme elements, and the award record does not label it. What is not in doubt is the direction, or the international context. NATO's combined allied defence spending passed $1.5 trillion for the first time in 2026, against a new target of 5% of GDP by 2035. The United States is not rearming alone, and contract records are where that becomes visible first.
Who is still allowed to bid
For a firm outside the United States, the two movements point in opposite directions.
At the top, the door is demonstrably open. FERROVIAL CONSTRUCCION PR — the Puerto Rico arm of a Spanish parent — moved from twelfth place among federal construction recipients at $151M to sixth at $1,079M, on the largest Army Corps civil-works contract in Puerto Rico's history: $1.08B to widen the Río Puerto Nuevo flood channel in San Juan. Puerto Rico's total federal obligations rose from $398M to $1,580M, up 297% — the sharpest percentage gain of any state or territory.
At the bottom, it is narrowing. Small-business set-aside dollars grew 8% while the total grew 40%, dropping the reserved share from 7.69% to 5.96%. Buy American domestic-content thresholds sit at 65% and step to 75% in 2029. And a non-US firm cannot bid a small-business set-aside directly at all: Canadian suppliers are waived out of Buy American for defence work under the Defence Production Sharing Agreement, but that waiver does not make anyone a US small business. The route in is a joint venture or a subcontract, and it always was.
What this does not show
The measure is federal prime contract obligations, award types A through D. Grants, loans and direct payments are excluded, which is why the 2021 infrastructure law is invisible here — and why the collapse in aid obligations is not the whole picture of what happened to American assistance abroad. A great deal of that money moved as grants, and none of it is in this ledger.
Every window here stops at May 2026, because Defense and the Army Corps withhold contract actions from the reporting system for 90 days. A recent month reads as almost no defence activity; that is an embargo, not a decline. An earlier version of this study ignored that and published four inverted findings — defence down 7% when it was up 46%, total obligations up 7% when they were up 40%. The windows are now bounded by what defence has actually released, and the build refuses to run past it.
Obligations are not contracts signed: money added to an award from an earlier year counts in the period it is obligated, not the period the contract was let. Place of performance is where the source says the work happens, not where the benefit lands. Recent months revise upward as modifications post, so every figure here is a floor.
And none of this explains anything. It describes an award record.
What a bidder should take from this
- Termination risk is no longer theoretical. USAID is the worked example: agency scale, four months, with a negative month in the middle. Price termination-for-convenience as a real line, not a boilerplate clause.
- The growth is narrow. Texas went from $16.8B to $50.2B and Arizona from $3.7B to $11.5B. Two states absorb most of the increase. "Federal spending is up 40%" is not a market signal.
- The volatile lines are the ones with a counterparty abroad — foreign aid, munitions, the border. If your pipeline touches any of them, it is more exposed to a single vote in Congress than it has been in years.
- If you are not a US firm, aim at the prime and subcontract tiers. The set-aside share is falling and was never open to you; the $1B civil-works award to a Spanish contractor was.
The underlying figures, with charts and full methodology, are in the data study. The practical read for bidders is in the companion post, and what is open right now is on the tender portal.
Sources
- USAspending.gov — all contract obligation figures, queried through the public aggregation API. Award types A/B/C/D, matched February–May windows, snapshot 2 September 2026.
- FPDS: DoD data availability — the 90-day hold on Defense and Army Corps contract data that bounds every window here.
- The White House, One Big Beautiful Bill — the July 2025 appropriation amounts for barrier, detention and Coast Guard.
- KFF, U.S. Foreign Aid Freeze and Dissolution of USAID — timeline of the award terminations behind the year-one figures.
- U.S. Customs and Border Protection, Smart Wall — barrier awards and mileage.
- Mexico's treaty complaints over the Rio Grande barrier — the 1944 Water Treaty and 1970 Boundary Treaty objections.
- U.S. Army Corps of Engineers / Ferrovial, Río Puerto Nuevo — the $1.08B San Juan flood-control award.
- NATO, funding and defence expenditure — allied spending totals and the 5%-of-GDP target.
- Canadian Commercial Corporation on the Buy American Act — domestic-content thresholds and the defence procurement waiver.
