Building Easier AI

The Custom Premium Is the Widest Since 2011

NAHB has measured the widest gap between custom and spec construction prices since 2011. Custom homes started in 2025 ran a median $171 per square foot of contract price; homes built for sale ran $152 with the lot excluded — a 12.6% premium, against a two-decade average near 9% (NAHB on Census data, 14 September 2026).

The tempting read is that custom builders finally have pricing power. But a premium is one of two things, and they behave nothing alike. Pricing power means the client pays more for the same work and the difference stays with you. A pass-through means your costs rose, the price followed, and the difference leaves for your supplier. From outside they look identical — the invoice is bigger either way.

Which of the two you are charging is a question every owner who builds to order faces. A fabricator quoting a short run, a service company charging for a same-day call — same question, thinner data. Construction publishes its numbers, so it is the worked example here.

And the premium is not national. In four of the nine census divisions, it does not exist at all.

Where the $19 comes from

Spec pricing has been flat four years — $152 for homes started in 2025, against $153 a year earlier — while custom went from $156 in 2021 to $171. The gap widened because one side moved. (NAHB computes the 12.6% from unrounded medians; the rounded $171 and $152 divide to 12.5%.)

Paul Emrath published NAHB's material cost results on 24 August 2026, from the July 2026 builder survey: median material-price inflation for the same house was 6.7%. The spread underneath is the story. The increase, in NAHB's words, "declines regularly with the size of the builder" — 9.1% for builders who started "5 or fewer homes in 2025," 1.8% for those with "100 or more starts." Same house, five times the inflation.

Rose Quint published NAHB's member census sixteen days later, on 9 September 2026: the median NAHB builder member started five housing units in 2025 — "fluctuated narrowly between five and six units since 2013" — with a median six employees and $3.7 million of revenue. The median member sits in the worst bucket of that spread: one more start a year and they would be out of it.

NAHB draws the bridge to custom itself, noting on the premium post that small builders — "most custom builders are in this group" — reported the higher hikes. The three explanations for the big builders escaping are Emrath's, on the material-cost post, and all three are purchasing rather than pricing: larger builders may have greater ability to "stockpile materials," may also have "longer-term contracts with suppliers, locking in current prices," and may be positioned to "negotiate deferred price increases."

Three labels before anyone leans on that. Material costs are the twelve months to early July 2026; square-foot prices cover homes started in 2025. The 9.1%-versus-1.8% split is cut by starts, not custom versus spec — the custom bridge is NAHB's own inference. And the census counts builders who joined NAHB and chose to answer its survey, not every builder in the country.

What custom buys is a different balance sheet

Robert Dietz's custom construction update of 19 August 2026 carries the difference between a custom builder's balance sheet and a spec builder's in one clause: custom is work "for which the builder does not hold tax basis in the structure during construction," and it is "almost universally undertaken by smaller, private home builders."

Look at what the other side holds: new single-family inventory stood at 488,000 homes in July 2026 — 9.6 months' supply, the highest that supply measure has been since January (NAHB on Census, 25 August 2026). Spec construction loans carried an effective rate of 11.82% in Q2 2026, against 11.67% on pre-sold work (13 August 2026) — NAHB's own measure, contract rate plus points over a short expected loan life, not an APR. That was the eighteenth straight quarter in which builders and developers reported tightening, and of those reporting it in Q2, 53% said lenders wanted personal guarantees or collateral unrelated to the project, "putting the entire business, rather than just an individual project, at risk."

Then the dirt. Median lot value under a spec single-family home started in 2025 was $59,000 (NAHB, 28 July 2026). That is the spec side's number — the custom client's own lot is never priced here at all, which is exactly why NAHB strips lot value out of the spec price, and why the custom builder's balance sheet never carries it. Real, and not margin.

The premium costs 4.3 months, and months are overhead

NAHB's construction time data, published 8 September 2026 for homes completed in 2025, puts permit-to-completion at 7.4 months for homes built for sale and 11.7 months for those built by hired contractors — "the larger portion of custom home building," in NAHB's own parenthetical. The subtraction is ours, off two figures on one page: 4.3 more months, 58% more calendar. The vintages differ — the premium covers homes started in 2025, the durations homes completed in that year.

The arithmetic is illustrative, run on national medians rather than any real job. $171 minus $152 is $19 per square foot. On a median new single-family floor area of 2,185 square feet in Q2 2026 — a different Census survey, reported 27 August 2026 — that is roughly $41,500 of extra contract value for 4.3 extra months, or about $9,650 a month, every dollar of it carrying overhead before it is profit.

Two public builders priced the margin-versus-volume trade this summer, in quarters a month apart. Neither is a custom builder — both build on land they control, not a client's lot. Toll Brothers — the closest public analogue to margin over volume, and "the nation's leading builder of luxury homes" in its own words, 18 August 2026, quarter ended 31 July — delivered 2,662 homes at an average $996,400 on a 23.9% gross margin, down from 25.6%, while its SG&A rose to 10.0% of home sales revenue from 8.8% as deliveries fell from 2,959. Lennar, reporting 16 September 2026 for the quarter ended 31 August, ran 1,713 active communities at 4.1 homes per community per month, with ASP $372,000, SG&A 9.2% and a 6.6% net margin — which Lennar prints and Toll does not. Inside Toll's own quarter, fewer deliveries did not buy lighter overhead: SG&A dollars rose 5% while deliveries fell 10%. Toll's 10.0% against Lennar's 9.2% points the same way, but that half of the comparison is the weaker evidence — two different models, quarters a month apart.

In four divisions the premium runs the other way

NAHB prints the counterweight on the premium post itself. By census division the premium runs 29% in the East South Central, 27% in what NAHB's page calls "the North Central," 9% in the West North Central, 7% in the South Atlantic — and in the West South Central the two prices are "nearly identical."

And in New England, the Middle Atlantic, the Mountain and the Pacific divisions, the spec median is higher than the custom median. The Pacific gap is the widest: NAHB puts the custom median in that division at $171 — the same number it prints for the national median — roughly a quarter below its spec median of $227.

NAHB's own conclusion: "Rather than representing a universal markup for custom construction, the 12.6% figure reflects the net result of very different regional housing markets and the characteristics of the homes being built in each market." On cause it goes no further than "presumably reflects differences in the types of homes being built." The organization holding the data hedges, so nobody reading it gets to do better.

Volume argues against the pricing-power read too. Custom starts were 49,000 in Q2 2026, 9% below a Q2 2025 that had itself set a three-year high; the trailing four quarters ran 183,000 against 184,000. NAHB's premium post says the custom market "remained stable" as spec contracted; Dietz's custom update prints that −9% quarter and says share expansion "stalled somewhat." Both are true: "stable" is the four-quarter total, −9% is the quarter. Custom's share of single-family starts is 20% on a one-year moving average, off 21% in early 2023. The premium widened. The segment did not.

Demand is thinning on the client's side too: mortgage applications fell a sixth straight month in August, purchase applications down 3.1% on the month (9 September 2026), into a 30-year fixed at 6.95% (PMMS, 17 September 2026) — applications for mortgages, not construction-to-permanent loans, but the climate your client walks in from.

Check it against your own numbers

A 12.6% premium on price per square foot is not a 12.6% premium on profit, and nothing published gets from one to the other. NAHB does split financial performance by builder type — inside its Cost of Doing Business Study, a 110-page book at $149.95, $79.95 for members (NAHB release, 9 April 2025). The one benchmark NAHB gives away is three years stale — fiscal 2023, average net margin 8.7% (11 March 2025) — and NAHB warns that incentives and price cuts "are likely to have caused this margin to shrink in 2024." The custom-versus-production split that circulates online is on neither NAHB page — check before repeating it.

So the answer comes off your own files. Three pulls, one afternoon:

Your materials bucket. Price one assembly you buy on every job — the lumber package, the window order, one mechanical sub — against the same assembly twelve months ago, from invoices, not memory. Near 9% puts you at the small-builder end of NAHB's spread, and your premium is a receipt. Then work NAHB's three reasons with a supplier: stockpile, longer contract, deferred increase.

Your calendar. Count permit to keys handed over, in months, on your last three contracts. Multiply the months beyond a for-sale schedule by your monthly overhead. That product is what the premium has to beat before a dollar of it is yours.

Your overhead per job. Divide last year's overhead by jobs closed. At the median NAHB member's five starts a year, each job carries a fifth of the office — and Toll's SG&A moved 1.2 points when deliveries fell 10%, from 2,959 to 2,662. One job out of five is twice that swing.

Run all three and you will know which premium you are charging: one you earned, or one your supplier already spent. The national medians will never tell you. Your own invoices will.