The bid you get is not the budget you wrote.

There is a particular silence on a job when the bids come back. Somebody has to say the number out loud, and the number is not the one in the feasibility study that got the deal approved.

In 2026 there is a specific, documentable reason for that, and it is worth understanding before you send the drawings out.

Materials moved first. Bids are moving second.

Underneath that average, individual lines are far worse: copper up 36% year over year, aluminium up 45%, US hot-rolled steel up 27%. Tariffs of up to 50% are landing on structural steel, aluminium and other imported products.

Contractors absorbed some of this. They are not going to keep absorbing it — they have neither the margin nor the appetite, and bid prices in the back half of 2026 are expected to close the gap. The baseline for final project cost, materials and margin together, is running around 5% year over year.

A budget written in January and bid in October is not a budget. It is a hypothesis about margin compression that somebody else has to fund.

Labour is the constraint people underestimate

Construction wages are up over 4% year over year, and 9–11% in high-demand markets and specialised trades. More than 60% of metro markets report labour shortages. Data centre construction is actively pulling crews away from multifamily, healthcare and industrial work, because it pays better.

That matters beyond cost. A trade you cannot staff is a schedule you cannot hold, and schedule is where a development deal usually dies — not in the line items, but in the carry on an extra two quarters.

What we would actually check

• An escalation allowance sized to the gap above, not a flat 3% because that is what the last deal used.
• When the estimate was written and when the bid date is. Every month between them is risk somebody is carrying.
• Which packages are tariff-exposed — structural steel, aluminium, electrical gear — and whether they can be bought early or substituted.
• Whether the GC has the crews, not just the price. A low bid from a contractor who cannot staff it is the most expensive bid on the table.
• The contingency, tested against a 5% escalation and a two-quarter delay together, because they arrive together.

None of this argues against building. It argues for bidding earlier, buying long-lead packages sooner, and writing the escalation into the model at the number the market is actually printing.