The Secret Economic Boom Almost No One Is Talking About

There's a 1,500-acre stretch of empty land in the middle of Wyoming going for about $15,000 an acre. That's roughly 75% cheaper than the average acre in a U.S. metro. Metro land comes with restaurants, plumbing, roads, people. This is dirt in the middle of nowhere.

That Wyoming parcel is about to be worth more than almost anything else in the country, and it has nothing to do with data centers. AI-powered robotics and autonomous machines are attacking the cost of turning empty land into a place you can actually live. If that works, land we treat as nearly useless gets repriced very fast. New towns start showing up in places almost nobody would have picked. Don't laugh this off. Don't go buy random cheap dirt either. Both are bad moves. See the mechanism first. Then I'll make it real with a town of 500 people and 2,000 robots.

Two pieces, one thesis

First: land in underdeveloped places can skyrocket because the cost of robotics, energy, labor, and intelligence are all falling at once. Those are the inputs that make a place useful: roads, homes, power, farms, factories.

Second: we get a boom in companies that build towns from scratch. Give a company a fleet of useful robots and highly capable AI, and city-building starts to look less like a government project stuck in red tape and NIMBYism, and more like a product. Those two ideas are one thesis.

Land is not mysterious. The bundle around it is.

A parcel is worth some portion of what people can do with it, minus the cost and risk of making it useful. Take two pieces of land with the same views and the same weather. One has a road, power, clean water, internet, a legal title, and permission to build. The other has none of that. They can be 20 miles apart and economically they are different planets. The expensive part is the bundle of systems around the dirt. When several of those costs fall together, things that were impossible suddenly become possible.

Intelligence is already going through that process. Stanford's Institute for Human-Centered Artificial Intelligence found that the cost of using a model around the level of GPT-3.5 from 2022 fell more than 280-fold between November 2022 and October 2024. That intelligence can design a site, coordinate schedules, order materials, diagnose equipment, and run many of the services a small community needs.

Construction is the lagging industry. One Energy Department-backed modular housing project used automated steel framing and reported up to a 46% cut in construction time versus traditional wood framing. Fast-forward that factory five to ten years. Autonomous delivery. Robotic site prep. Then humanoids for the weird final tasks humans still crush today. A lot of people will say that's 30 years away.

The robots are already here. You parked next to one.

Fully autonomous robots already exist in the millions in the United States. One of them is probably sitting in your driveway. Tesla cars with Full Self-Driving already do most of what a fully autonomous physical system has to do in the real world. Nine cameras. An onboard AI computer. Coast to coast without touching the wheel or the pedals. Any road network. All kinds of weather. Parking lot to parking lot.

Tesla trained an end-to-end neural net on billions of miles from regular customers. That same stack is what they intend to put into Optimus. World models like that are being built across drones, heavy machinery, trucks, and purpose-built robots. The question is how you power a whole town of them.

The U.S. grid is a mess. The cost curve is not. The International Energy Agency reported that onshore wind and solar were the two lowest-cost sources of new electricity generation globally in 2024. Texas now has a larger solar buildout than any other U.S. state, including California. China is installing about five times more capacity per year than Texas. In 2025 China installed more solar in a single year than the United States has installed in its entire history.

A town of 500 people and 2,000 robots

Start with good land, water rights, legal permission to build, and a connection to the outside world. Solar and batteries supply much of the power. A local microgrid keeps the town running if the larger grid fails. Robotic equipment clears and grades the site. Automated factories produce wall panels, roof systems, pipes, utility modules. Construction robots assemble the repeatable pieces. Humans handle inspections, the nasty connections, and the problems that still need judgment.

Those first robots build housing for the people who operate the town, then storage, workshops, a clinic, a school, greenhouses, water treatment. Part of the fleet farms. Autonomous trucks bring in supplies. Another group maintains roads, solar, batteries, water, and buildings.

Those 2,000 robots do not all have to be humanoids. Most of them should not be. A machine built only to dig trenches will beat a humanoid holding a shovel any day of the week. Humanoids matter for stairs, doors, tools, shelves, ladders, kitchens, repair work, medicine.

Once enough of this is built, AI becomes the town's knowledge layer. It helps a nurse prep a case for a remote doctor. It gives a mechanic specialized diagnostics. It tutors students. It models water demand. It balances the microgrid. Humans still own the decisions. Self-sufficient does not mean isolated. It means a community can support far more daily life with far fewer people.

Small towns have always had a scale problem. How does a community of 500 people support a full-time expert in every medical field, every engineering specialty, every trade, every school subject? It can't. Large cities solve that with density. AI and robots weaken those reasons for concentration. Expertise arrives through a screen or a machine. Physical work arrives through machines. Distance matters less when transport is cheap and autonomous.

That does not mean megacities die. People like cities. Living in one just becomes a choice instead of the price of admission to modern life.

What stays scarce

Cheap land today — especially land with natural beauty and above-average weather — goes up a lot as developing it gets cheaper and faster. But there is a twist. If robots make roads, utilities, and buildings cheaper, they increase the total amount of land humans can use, which can push average land prices down. Physical structures get commoditized. The valuable land still needs something machines cannot make more of.

People jump to water. The U.S. Geological Survey says the country has enough water overall, but not in every place at every time. The same technology that makes buildings cheaper also makes desalination cheaper. Climate is a different story for the next couple of decades. Air conditioning made the American Sun Belt livable. Giant weather domes are a multi-decade problem.

Property rights matter. A World Bank study of a land title program in Ghana found it did not produce every promised benefit, but land values rose. A title alone is not enough. Permitting may become even more valuable as construction gets cheaper. If ten companies can manufacture buildings, but only one piece of land is legally allowed to add homes, the scarce asset is permission. The OECD calls land use regulation a key determinant of housing supply.

Energy is solvable with sun, wind, geothermal, and cheap batteries. Connectivity is solvable with Starlink. The most overlooked asset is still a functioning community people want to join. You can automate trash collection until you are blue in the face. You cannot mass-produce trust.

Four variables. All four.

The most valuable land in this future has four things at once. A scarce attribute robots and AI will not easily solve, like natural beauty or weather. A land price that is still low relative to the competition. A development problem that technology can actually solve. Institutions in that jurisdiction moving in the right direction.

History has done versions of this. Economists Dave Donaldson and Richard Hornbeck studied the U.S. railroad network from 1870 to 1890. Better market access showed up in agricultural land values. Highways did another round. Air conditioning made hot places livable. I live in Texas. The population here would be far smaller without it. Same story in Arizona and Nevada.

If I am right, this becomes one of the most interesting new company categories of the AI era. Some companies sell construction robots. Others build automated factories for walls, roofs, and entire rooms. Some operate robot fleets as a service. Power companies package solar, batteries, and microgrid software into one local system — Tesla is already in that lane. A few companies will combine the entire stack and sell towns: find the land, secure the rights, design the town, install the utilities, manufacture the buildings, deploy the machines, operate the shared services.

This is not investment advice. This is how I would interrogate a piece of land. What does it have that technology cannot reproduce? Why is it cheap today? Which cost curve changes this parcel from cheap to expensive? Name the machine, the job, or the cost it removes. Can people legally build there? Where do water and energy come from on the worst day? How does the community connect to the outside economy? And who actually wants to live there?

That is the boom. Not data centers. Not a land-flip lottery. A stack of falling costs — robots, energy, labor, intelligence — that turns empty dirt into a product. Most people will miss it until the towns are already standing.

Check the video here.