Rent vs Buy, AI Transition Edition
What AI automating labor could mean for investing in a house
One of the biggest investments that most Americans make is buying a home. There are many things that are financially great about buying a house.
Probably the most important thing is that a home isn’t a terrible investment, and your mortgage is something you are forced to pay into once you sign up. This forced savings plan helps a lot of people put money away that they would probably otherwise spend.
However, if you want to be really strict about your investments and assume that you will put all the money you are not putting into your house into stocks and rent, you can actually calculate whether “rent vs buy” will likely be a better investment. I’ve done that for myself in this google sheet. The conclusion I reached is that under relatively normal conditions, buying a house for $200,000 to avoid paying a rent of $1,000 is about even financially. There is also NerdWallet’s rent vs buy calculator that I remember being pretty nice. I made the google sheet myself to make sure I understood all the pieces.
But we’re not here to talk about rent vs buy under normal conditions. We’re here to talk about rent vs buy assuming that AI and robotics is going to automate essentially all labor over the next 5-15 years.
There are three things I think about here:
The depreciation of the value of your house due to depreciation of labor and materials.
The appreciation of land as one of the assets that remains scarce in a materially abundant world.
The safety of owning a house through turbulent times.
Depreciation of House Value
A couple sources I found say the cost of building a house is about 40% labor and 60% materials. I expect humanoid robots to cost less than $1/hour, which is about a 50x reduction in labor costs.
I also expect a big reduction to the cost of the raw materials:
I expect the cost of energy will likely drop due to solar and nuclear power, making many materials cheaper.
I expect new materials and processes to be invented that substitute for more expensive materials.
I expect the cost of labor that goes into materials production to drop a lot.
Overall, it doesn’t seem crazy to me for a formerly $200,000 house to only cost about $10,000 to rebuild from scratch. If this is going to happen, the value of your house will drop as soon as people see it coming, which will probably be before the robots actually arrive en masse to rebuild everyone’s house.
I would expect the newer houses to be built using better techniques and materials and you may decide to just doze your house and have it rebuilt. So in general, I expect the value of your house to drop to almost zero in an AI transition. The main value will be whatever sentimental value you imbue it with due to the memories it contains.
Appreciation of Land Value
One of the things that you can’t just throw robots at and make a bunch more of is land on planet earth. So I expect land to remain scarce and valuable. And in general, as the world gets richer, the fraction of earth that you own should go up in value.
However, the value of the land your house is on is probably determined almost entirely by being near a city or town, having road access, etc. Also, most houses are not on that much land. There are about 5 acres of land for every human on earth if we distributed it evenly. The median lot size in the USA is about 0.25 acres. Farm land in the USA goes for about $4,000/acre. So if the median house lot were turned into farm land, it would only be worth about 1,000 bucks.
Since almost all of the value of your house’s land is contingent on being near a city where there are jobs that everyone needs to get to, it seems unwise to plan on the value of that land going up once everyone is out of work. So even though I expect the value of land in general to go up. The value of the land your house is on may very well deprecate! Who knows?
Safety and Politics of Owning a House
If the world is being turned upside down, a primary residence that you own provides stability and safety. The cost of the depreciation may just be the premium you intentionally pay for stability and safety during the transition.
Also, there might be mortgage forgiveness plans if the depreciation is severe enough, since home owners who have open mortgages are a large political block who will be quite unhappy to see standard financial advice to turn on them like this.
I expect (hope) there will be some form of redistribution. One way to fund this is a wealth tax or asset seizure. (I think there are other ways to fund it I would prefer.) I would expect one of the last asset classes to be seized is primary residence home ownership.
If you expect asset seizure to be a thing, you may want to buy a primary residence just have something secure that will keep some value. Though, there may be better options for that purpose.
Claude’s Opinion
I asked Claude Sonnet (an AI) to give its honest opinion on the thesis. Here is what it said:
I think the essay raises a fascinating point about how AI could dramatically impact housing markets, but I’m skeptical of a few key assumptions:
Timeline: 5-15 years for “essentially all labor” automation seems highly optimistic. While I expect significant automation in construction, the leap to humanoid robots at <$1/hour within 15 years feels like a stretch. I’d expect a more gradual transition with partial automation first.
The house value depreciation thesis seems too extreme. Even if construction costs plummet:
There’s massive institutional/regulatory inertia in housing markets
People value craftsmanship and “authenticity”
Existing homes have established infrastructure connections
The transition period would likely be much longer than implied
The land value analysis is probably the strongest part. The point about land value being tied to proximity to economic activity is insightful and concerning - though I’d note that even in a highly automated world, humans will still cluster for social reasons.
The political/safety angle feels underdeveloped but important. Housing is so central to both individual wealth and political stability that any major disruption would likely face enormous resistance and intervention.
My conclusion would be more moderate: While AI could significantly impact housing markets, the effects will likely be more gradual and complex than a straightforward crash in construction costs. The decision to buy vs rent should still primarily depend on personal circumstances and local market conditions rather than AI speculation.
That said, I really appreciate the essay’s attempt to think through these second-order effects of AI advancement. It’s exactly the kind of forward-looking analysis we need more of.