Start with who was asked. The AGC put the question to contractors, not to owners: 63 percent said an owner postponed, scaled back or cancelled something in the previous six months, and the reasons on offer were funding uncertainty at 37 percent and financing that was unavailable or too expensive at 34 percent [1][2]. Both answers describe a decision taken in someone else's building. A contractor reporting that the money was not there is reporting a conclusion; a contractor reporting that its own price came in high enough to end the conversation is reporting a performance. The survey collects the first at no cost to the respondent, which is reason to treat the 71 points those two answers add up to as softer evidence than the number looks [12]. The counterweight sits in the same argument. Bloomberg reported in April that roughly half of the US data centres planned for 2026 were likely to be delayed or cancelled, in a market where capital was already queued behind the asset class [4]. That is a single sector, and the source does not say what else was holding those particular projects up. It does establish that money can be present and plentiful while projects still fail to start, which is enough to stop treating financing as a complete account. The case is made in Forbes by Shiva Dhawan, CEO and cofounder of Attentive.ai, the company behind Beam AI [8]. A skeptic's version is short: a preconstruction software vendor has located a preconstruction problem. The answer is that the mechanism does not require his category to be the remedy. Dhawan argues that firms without estimating capacity simply no-bid, and firms that bid without time to validate scope or vet supplier pricing push the residual risk into the number [10]. More people at the estimating desk, or wider bid windows, would relieve that as well as any tool would, except that AGC found 80 percent of firms struggling to fill salaried openings, the highest share in three years, with estimating squarely in that category [5]. Padding is the rational response rather than a lapse. Construction input prices were up nearly 10 percent year over year in May 2026 on Associated Builders and Contractors' reading of federal producer price data, and a contractor who misses escalation on that scale, or a gap in the drawings, inherits the loss [6]. Dhawan's description of what the owner then receives is a number that reads too high, or two bids where five were wanted, or figures far enough apart that none of them is trusted [14]. What the record does not contain is the join. No one has shown that the cancelled projects were the ones that drew thin bid lists, or that wide spreads predict abandonment; the survey reports the cancellations and the decline in bid opportunities separately, and the line between them is inference [1][3]. The analogy offered instead comes from health insurance, where a 2025 Imagine360 survey found 38 percent of insured Americans had skipped or postponed necessary care or medication over cost, against 27 percent in a separate 2023 study, with 42 percent of those who delayed saying their condition worsened [7]. Two different studies eleven points apart, a relative rise of about 41 percent, make a weaker trend line than one tracker would [13], and the comparison illustrates the mechanism rather than testing it on a jobsite. For an owner deciding this quarter, the practical question is whether a high bid is a verdict on the project or on the hours the bidder had, and that is checkable one job at a time: how many firms were invited, how many returned numbers, how long they were given, and how far apart they landed. The decade version is less comfortable.
