Important Warning: Unsupervised FSD Could Crush Car Values

I'm going to walk you through some math that will make you uncomfortable if you just bought a new car or you're about to. I'm not trying to be a jerk. I genuinely think most people in the United States have no idea what is about to hit the new and used car markets.

Start with the boring numbers. The average new car in the U.S. right now costs about $48,000. Round it down to $40,000 for a cleaner example, even though that is already generous for a lot of buyers. Four years from now, that car is worth maybe $16,000 if you are lucky. That is a 60% hit. You just lit $24,000 on fire. That is the optimistic case in a normal market.

The next few years are not going to be normal.

We are roughly 12 to 18 months away — maybe sooner — from unsupervised full self-driving getting approved in major U.S. markets. If you watch this channel a lot, you have heard versions of this before. Elon has said versions of this before. What is different now is the product and the rollout. I used to work at Tesla. I have been an investor since 2012. I drive FSD every day in my Cybertruck. With version 14, we are at a point where Elon himself said they are getting comfortable with people texting while the car drives. Think about that for a second. The technology is essentially there.

Tesla already released the robotaxi app. They are operating in Austin and the Bay Area. They got approved in Arizona. By the end of this year they are targeting 8 to 10 metro areas. Waymo is already doing driverless rides. This is not a forever promise. It is happening right now.

The autonomy depreciation thesis

When unsupervised FSD is truly unsupervised — you do not have to pay attention — the value of owning a personal car changes overnight.

Why pay $40,000 for a depreciating asset that sits in your driveway 95% of the time when you can hail a robotaxi that shows up in a few minutes, costs about a dollar a mile, takes you where you need to go, and you never worry about parking, insurance, or maintenance? That math stops working for a huge share of owners.

Even if you already own the car outright, ownership is expensive. The average cost to own a vehicle in the U.S. lands somewhere around 70 to 80 cents per mile once you add the payment, insurance, fuel or electricity, maintenance, and parking. Used cars can get closer to 50 cents a mile if you are driving a beater, but you still eat tires, repairs, gas, and insurance. Tesla's robotaxi economics point toward roughly 30 cents a mile, maybe closer to 25 cents when Cybercab scales. That is about a third of what many people pay to own.

When robotaxis launch at real scale — tens of thousands of units across major metros — demand for private cars in those cities is going to crater. Why would someone spend $30,000, $40,000, or $50,000 on a car they can summon for a fraction of the cost? They will not. When demand for used cars drops, prices drop. That is basic supply and demand, and it is going to be brutal for people who just bought new metal thinking they could sell it for something reasonable in a few years.

Two scenarios

Scenario one: unsupervised FSD gets approved widely by 2027. Texas looks likely. California is pushing. Arizona already approved robotaxi operation. If robotaxis are running in 15, 20, even 30 major metros by then, that $40,000 car that might have been worth $16,000 under a normal curve could be closer to $8,000 — maybe less. Demand for a 2025 Camry evaporates when people can summon a Tesla robotaxi for a dollar a mile.

Scenario two: I am wrong on timing and this takes until 2030. You still take the normal ~60% depreciation over that window. You are still out about $24,000. You just avoid the extra robotaxi cliff. That is the case that frustrates me most — people treating a new car like a safe store of value when even the "slow" path is a big cash burn.

Then layer opportunity cost. This is not financial advice and I am not telling anyone to YOLO their life savings. Just think. That $40,000 parked in a car is money that could sit in assets positioned for this transition — Tesla, Nvidia, companies building AI infrastructure, robotics, autonomy. If that capital compounds at even 15% a year for a few years as AI takes hold, you are looking at something like $80,000 in four years while the car falls from $40,000 toward $8,000. That is a roughly $72,000 swing. For a lot of households, that is life-changing money — and the only move was not buying a depreciating asset right before the disruption.

Who should still buy a car

Be fair. Some people still need one. If you live rural and robotaxis will not reach you for a decade, you need a car. If you are a tradesperson who needs a work truck, you need the truck. If you have a big family and you need a specific vehicle that robotaxi service will not replace, buy what you need. There will always be exceptions.

But if you live in or near a metro area, and you are buying mainly for commuting, errands, and weekend trips, you need to think hard. Leasing can make more sense than buying right now for a lot of people who still need wheels for a few years. You transfer depreciation risk to the bank. If the used market craters because of robotaxis, that is their problem. You hand the keys back. Leasing has downsides — you pay for not owning, mileage caps, limited customization — but in a world where depreciation could run worse than history, paying a bit more to dodge that risk can be the smart play.

Uber math versus ownership

Say you spend about $15,000 a year owning a car — payment, insurance, fuel, maintenance, parking, tolls. For a lot of people that number is conservative. If you skip ownership and use Uber or Lyft for everything for three years at roughly $20 a trip and 500 trips a year, that is about $10,000 a year. You save roughly $5,000 a year versus owning, and that is with today's driver-paid pricing.

When robotaxis hit around a dollar a mile without a driver to pay, transportation cost can fall again. You could be looking at something like $5,000 a year for mobility versus $15,000 to own. That is $10,000 a year kept in your pocket. Over four years that is $40,000 — the sticker price of the car you almost bought.

This hits cities first

Robotaxi disruption starts in places like Austin, San Francisco, Los Angeles, Phoenix, and Miami — metros with approval and density. Rural Wyoming or Idaho may not feel this for a decade or more. Overseas markets will move on their own clocks. If your suburb sits 40 miles outside a robotaxi geofence, you still need a car for a while. The impact will be uneven.

Here is the catch: about 85% of Americans live in urban or suburban areas. Most car buyers live where robotaxis should show up within a few years. For most people watching this in the U.S., this warning applies.

What I am doing

We have a 2021 Model Y with over 80,000 miles. I am not upgrading it. It already has FSD. Why spend another $40,000 or $50,000 on a new car when robotaxis are about to be everywhere and the car I have already drives itself? The math does not work. I also have a Cybertruck because I like triangles and I love the thing, and it has FSD too. If I am advising a friend on a pure money decision, I would tell them to drive what they have into the ground, keep it maintained, and do not buy a depreciating asset right before the biggest disruption in auto history.

Should you buy a new car right now?

Ask yourself:

1. Do you live in or near a major metro where robotaxis are likely within two to three years? If yes, do not buy.
2. Can you delay 18 to 24 months and watch the rollout? If yes, probably do not buy.
3. Do you have Uber, Lyft, transit, or carpooling that can cover you for a couple years? If yes, probably do not buy.
4. Is your current vehicle still safe and functional? If yes, keep driving it.

The people who should still buy are people who need a specific vehicle for work or lifestyle that nothing else solves, people in rural areas far from any geofence, or people who already did this math and are making a luxury purchase on purpose.

The canary

Watch for unsupervised FSD approval in Texas within the next year or so. Then watch used car prices in Austin. That is the canary in the coal mine. If robotaxis scale and used prices there fall faster than normal depreciation, the thesis is playing out. Once it happens in one market, it cascades to every other market that approves robotaxis. You do not want to be holding $40,000 of metal when that wave hits.

This is the warning. The car market is about to get repriced around transportation as a service. Act like it.

Check the video here.