Who builds the homes

New housing is the half of property most people care about and the half that is hardest to count. There is no world league table of housebuilders, only three countries that publish one — and in the countries that do not, the firms building the most homes are precisely the ones with no obligation to say so. This report is what could be established, and where the counting breaks.

Canada: no league table, and two cities that accidentally publish one

Canada publishes no ranking of housebuilders by units. Nothing equivalent to the American Pro Builder Top 200 or Australia's HIA Housing 100 exists, and most of the largest Canadian builders are privately held, so there is no filing to read either. The usual workaround — quoting each company's own "homes built" figure — produces a table where every row is counted on a different basis: lifetime totals beside annual closings, apartments under management beside homes sold.

Two cities escape this, and almost nobody cites them.

Ontario's regulator publishes the numbers. The Home Construction Regulatory Authority licenses roughly 5,000 builders and vendors and publishes, for each, its total possessions — homes actually handed to buyers. That is a units-based league table from the regulator's own database rather than a magazine's survey. On it, Mattamy Homes records 35,829 (27,040 freehold and 8,789 condominium), well clear of Minto at 14,262, Tridel at 12,101 — all condominium, the purest high-rise builder in the table — Menkes at 9,990 and Daniels at 7,927. What stands out is the ownership: every firm near the top is privately held, and the only entity in the upper reaches owned by an alternative asset manager, Brookfield Residential, records 2,405.

Calgary publishes its permits, applicant by applicant. The City's Building Permit Builders Report names the applicant on every permit and states the housing units it creates, back to 1999. Permits since 1 January 2024 total 59,026 new units. Truman Homes leads on 4,480 units from 562 permits; Jayman BUILT records 1,323 units from 1,268 permits — almost exactly one unit per permit, the signature of a single-family volume builder — while Deveraux Developments records 1,519 units from 15 permits, a hundred units at a time, a purely apartment-scale developer that would be invisible in any single-family ranking. The same measure shows Brookfield Residential's Alberta arm behaving almost identically to the local family firms.

Elsewhere in Canada the picture is family firms in their second and third generation, and one large pension-owned builder. In Metro Vancouver: Onni Group on 15,000-plus homes, Concert Properties on 12,000-plus — owned by 49 pension funds and institutional investors, which makes it the rare builder whose shareholders are named — Bosa Properties on 10,000 and Wesgroup on 7,000-plus. In Montreal, Prével on 11,000-plus units and Le Groupe Maurice on 12,192 apartments, a seniors-housing operator-developer rather than a homebuilder in the ordinary sense. Every one of those numbers is the company's own, on the company's own basis.

The United States: the one market that is properly counted

The US is the exception. Pro Builder's Top 200 is a real published ranking, though it ranks by revenue rather than units and the magazine has moved the goalposts — the old Builder 100 URL now 404s. More usefully, the largest builders are listed companies that file, so the units are audited:

Two of the five shrank year on year — Horton's closings fell 5% and Pulte's fell from 31,219 — which is the kind of fact a filing gives you and a league table does not.

At city level the counting fails again, in an instructive way. New York cannot be given a builder ranking honestly: title to nearly every new building sits in a single-asset LLC, and the city's permit data will not rescue it. What New York can show is who holds the finished housing, and the answer is unlike anywhere else — the New York City Housing Authority on 177,155 units, then two limited-equity housing cooperatives, Co-op City on 10,914 units in one tax lot and Rochdale Village on 5,860. Blackstone appears on 13,481 units, almost all of it Stuyvesant Town and Peter Cooper Village. Columbia University holds 4,790, NYU 3,450: universities are among the largest residential owners in the city.

Dallas shows the other failure mode. No Dallas league table is published any more, but the appraisal roll names owners, so builder land positions can be counted directly: Bloomfield Homes on 577 accounts, far ahead of Lennar on 133 and Megatel on 106, with D.R. Horton — the largest homebuilder in the country, headquartered in the same metroplex — holding only 60. That is a measure of land banked in one county, not of homes built, and it should not be read as the latter.

Internationally: ownership classes instead of builders

Outside the countries with league tables, the honest unit of analysis is often not the builder at all.

Amsterdam publishes no builder ranking, but it publishes something better: the whole dwelling stock by owner class. Regulated housing corporations hold 191,010 dwellings — 39.2% of every home in the city — and are the principal commissioners of new construction. Owner-occupiers hold 148,400, up 18,525 since 2021, the largest gain of any class. Companies hold 61,210, up 10,725. And the only class that shrank is individual private landlords, down 1,480. That is a clearer account of who builds and holds Dutch housing than any list of contractors would be.

Mexico City publishes no metro housing starts or completions at all — nothing comparable to CMHC or the US permits survey exists. The listed builders can be read directly: Vinte, having absorbed Javer, reports 15,243 homes deeded on a pro-forma 2025 basis. Beyond that the market runs through private developers that publish no unit counts, and through the FIBRAs, which build apartments inside mixed-use schemes rather than subdivisions.

What to distrust in this report

Counts on different bases should not be added. A lifetime total (Tridel's 90,000-plus homes over 90 years) is not an annual figure (Horton's 84,863 in one year), and a land position (Bloomfield's 577 Dallas accounts) is neither. Where a company publishes only a cumulative number, the row says so.

The private-equity question Jim asked — how much of homebuilding sits inside PE platforms — does not yet have an answer good enough to publish. On the two Canadian measures where builders are counted properly, the PE-owned entrant is present but not dominant: Brookfield Residential records 2,405 possessions in Ontario against Mattamy's 35,829, and in Calgary it behaves like the family firms around it. That is two cities, and two cities is not a finding. Establishing it properly means reading ownership behind the top twenty builders in each of the larger markets, which is the next pass.