The State of Construction Money 2026

Across the US commercial construction sector, the median cash-cycle (sub invoice received to GC funds settled to all parties at the bottom of the chain) is between 41 and 58 days. There is no public clearinghouse-style benchmark; the figure is constructed by triangulating ENR Top 400 surveys, public surety annual reports, and the working-capital footnotes of the half-dozen publicly traded GCs.

Securities markets ran cash-cycles of similar magnitude in the 1960s, before the establishment of the Depository Trust Company and the broader move to T+1 settlement. Healthcare claims ran 60- to 90-day cycles before the 2010s push to electronic adjudication. Logistics ran weeks-long cycles before EDI 940 and 944 messages collapsed them into hours. In each case, the cycle compressed not because individual operators got faster, but because a shared infrastructure layer absorbed the reconciliation work.

Construction has not yet had its clearinghouse moment. The closest analog (the lien-waiver chain backed by state statute) is a paper-era artifact. The ACORD certificate-of-insurance form is a paper-era artifact. The AIA G702/G703 pay application is a paper-era artifact, digitized only as a PDF. None of these are infrastructure; they are conventions.

On a $20M commercial project, the capital stack is: 25 to 40 percent owner equity, 60 to 75 percent senior construction loan, 5 to 12 percent retainage held by the GC against subcontractor performance, and a working-capital line of credit at the GC level that absorbs the timing mismatch between sub payment and owner draw. The LOC is the slack variable; it expands and contracts to keep the chain flowing.

The cost of that LOC, at 2026 rates, is roughly 8 to 10 percent annually for investment-grade GCs and 12 to 16 percent for mid-market GCs without strong financial covenants. Every dollar of working capital trapped in draw float is a dollar paying that rate. On a typical $200M revenue mid-market GC carrying $14M of working capital, that is between $1.1M and $2.2M of pure interest expense per year, financed entirely by the GC.

The owner does not see this cost. The lender does not see this cost. The sub does not see this cost. It sits on the GC's P&L as a deductible expense and quietly compresses GC margins by 1 to 2 percentage points. In a sector with low single-digit operating margins, that is a meaningful share of profit going to the bank.

A clearinghouse, in the sense the securities and futures industries use the term, is a shared rail that nets obligations between counterparties so each party only settles a single net position rather than gross exposures. In construction, the equivalent is a shared ledger across GC, sub, owner, lender, and surety that reconciles pay applications, lien waivers, compliance gates, and disbursement instructions in real time.

The first-order effect is the cash-cycle compression we have built the platform to deliver: 52 days collapses toward the lower-teens within ninety days of onboarding. The second-order effects are larger.

Bonding capacity expands when the surety can see real-time payment readiness instead of a quarterly snapshot. Bonded capacity headroom of 20 to 30 percent is achievable on a clean book; we have seen it move 8 to 14 percent on the first-pass design-partner cohort.

Insurance pricing tightens when carriers can verify continuous coverage and matched endorsements rather than estimating from sampled certificates. Loss-control discounts of 4 to 8 percent on GL premiums are realistic for GCs that maintain Clearinghouse-grade compliance.

Tier-2 liability falls. Joint-check coordination, which currently fails silently in roughly 6 percent of supplier relationships, becomes traceable. Tier-2 lien claim frequency drops to a fraction of its current rate.

Three forces converge in 2026 to make the clearinghouse model viable. First, payment rails: Modern Treasury, Plaid, and the underlying ACH/RTP/wire infrastructure are mature enough to handle multi-party disbursement with seconds-to-minutes latency rather than days. Second, document AI: extraction from ACORD 25/28 forms and AIA G702/G703 pay applications is reliable enough at 95 percent-plus accuracy to remove the human bottleneck on every draw cycle. Third, regulatory tailwind: state lien-statute reform programs in roughly twelve states have standardized the conditional/unconditional waiver chain in ways that allow software to operate across jurisdictions without bespoke per-state logic.

None of these forces existed in combination before 2024. The sector has been waiting for them.

We expect the clearinghouse model to become the dominant operating pattern for commercial construction capital within ten years, the same way DTCC eligibility became table stakes for institutional equities trading by the early 1980s. The question is not whether the rail emerges, but who builds it.

Three operating patterns are competing right now. The first is the PMIS-led pattern, which extends a project-management platform downward into financial workflows. The second is the bank-led pattern, which extends construction-loan administration upward into draw orchestration. The third (the pattern we believe will win) is the clearinghouse pattern: a horizontal rail under every project, owned neither by the GC's PMIS vendor nor by the lender's loan-admin platform, but by a shared infrastructure operator.

By 2030, we expect three to five clearinghouse-pattern operators to handle the majority of US commercial construction capital flow. By 2035, we expect one to be a clear category leader in the way DTCC is in equity post-trade. We are building Subcontractor Audit to be a serious candidate for that role.

The cohort is private until the first public draw clears the rail, which we expect in the same quarter we publish the next edition of this report. We will not invent numbers. What we can say is that the operational shape of the early cohort matches the thesis: cash-cycles are compressing, bonding-capacity conversations with sureties are getting easier, and CFOs are choosing to extend the platform deeper into their books rather than bouncing back to spreadsheets.

We will publish the cohort numbers, with each design partner's voluntary participation, in the next edition. Until then, we resist the temptation to fabricate the proof.