Four 'Valencias' or twelve 'Valladolids' to build 2 million homes

Building two million homes overnight is the thought experiment posed in the debate: imagine a benevolent dictator who wants to make housing affordable in Spain. At the average of two inhabitants per home, that stock would shelter four million people. The goal is not in question. The hard part is where, with what water, and how fast. The same exercise posed to two language models returns two plans that have nothing in common.

How many new cities does it take for 2 million homes?

One proposal bets on four Valencia-style megacities, never bigger, with 500,000 homes each, placed at intermediate points in empty Spain (the depopulated interior) to raise the country's productivity. Two would have to go to Extremadura and Castilla-La Mancha. The other model spreads the same target across twelve Valladolid-style cities of 150,000 homes and drops them anywhere on the map. The second model's argument is common sense: that size does not overwhelm transport or quality of life, and it gets built sooner.

That is where the real problem begins, because at the bottom of the issue there is a plumbing question more than a planning one. And whoever signed off on the four-megalopolis plan knows it.

Water and empty Spain: the bottleneck nobody solves

The most repeated criticism is not about size, but about location. The argument goes that the interior of the peninsula is mountainous and that large areas already suffer water stress, so moving hundreds of thousands of new residents there, with their floating population and industry, requires solving supply first. The very choice of Extremadura and Castilla-La Mancha, territories with uneven water capacity, is called into question by this camp.

The geographic objection clashes with the stated goal of those two models: to bring in infrastructure, services and industry before the cranes. Nobody has disputed that order. What is disputed is whether the political patience to stick to it exists.

Build upward before building into nowhere

Against the fever for new cities appears a less segarro alternative: densify what is already built. Some point out that entire neighborhoods of Madrid are still capped at three stories when they could hold five or six, and that this land already has transport and sewerage. Building vertically is cheaper than inventing a Valladolid in the middle of a wasteland, even if it makes no headlines.

The timeline math does not help either. The idea circulating is that any plan would start a decade or two late and that, once underway, it would not be sped up by bringing in industrialized machinery. Buying cement fast is easy; processing licenses and permits is not.

Is expanding supply enough, or must demand be touched?

The whole logic of the plan rests on supply and demand: more housing, lower prices. It is the starting premise and almost nobody denies it in the abstract. The disagreement comes with the distribution. Part of the analysis holds that the effect dilutes if demand keeps growing at the same pace, and argues for cutting certain subsidies and slowing population inflows so the market does not swallow the new stock.

In the middle sits an uncomfortable comment: those demanding housing could join in building it, though, in the writer's view, what they want is for it to be free. The hand asking for a house does not seem willing to lay bricks.

The Spanish precedent that dismantles both models

Spain has no tradition of new cities. The case cited as a reference is Tres Cantos, created in 1971 and home to some 54,000 inhabitants, a ridiculous figure next to the projects on the table. The rest of the country's purpose-built towns barely reach 7,000 residents. There is no recent experience in raising settlements of hundreds of thousands of souls from the dust.

With that track record, the reasonable prediction is that neither the four megalopolises nor the twelve medium-sized cities will see the first stone before a decade. And by then, rents will have done whatever they pleased. The safe bet is that some model works somewhere. Nobody will sign off on which.