TechNewsFirst

Tesla used to let you buy its self-driving software, FSD, one time. That price kept going up, from $3,000 in 2016 all the way to $15,000 in September 2022. Then Tesla reversed course completely: it cut the price back down three separate times, and as of February 2026, you can't buy it outright at all anymore. The only option now is to rent it, for $99 a month, and Tesla can raise that whenever it wants.

Here's a number that makes this concrete. In July 2026, the cheapest way to lease a Tesla Model 3, according to reporting from Carscoops and CarsDirect at the time, cost $379 a month, plus thousands of dollars due upfront. Add the $99-a-month self-driving subscription and you're paying roughly $478 a month for a car and a piece of software, neither of which you will ever actually own. That lease price has bounced around since, down to $299 at one point, back up past $419 by September, so it's not a fixed number. But the pattern is the point: two separate things from the same company, both turned into payments that never end.

Here's the twist most people don't know: back in 2019, Elon Musk said buying a Tesla was buying an "appreciating asset," meaning it would be worth more over time as FSD got better and its price kept climbing. That didn't happen. Tesla cut the price instead, and used Teslas are now losing value faster than most other cars, not slower. The marketing was wrong. What replaced it, the subscription, doesn't depend on that marketing being right anymore.

At the same time, Tesla stopped making its priciest, lowest-selling cars, the Model S and Model X, and turned that factory over to building robots. Put it together: sell more cars, at lower prices, and make the real money on a monthly subscription instead of a big one-time price tag.

This same switch happened everywhere else first, with real numbers behind it. Adobe swapped a $2,500 one-time software purchase for a $50-a-month subscription in 2013 and roughly tripled its revenue in the years after. Take-Two, the company behind Grand Theft Auto, made over $4 billion in one year alone from in-game purchases inside a game that came out in 2013, according to its own financial filings, more than most companies make from selling anything at all. Amazon Prime went from $79 a year in 2005 to $139 by 2022, a few dollars at a time, and almost nobody canceled over any single increase. Self-storage companies have a name for this exact trick, ECRI, meaning they raise your specific rate in small steps after you've already moved in, because tiny increases don't feel like a decision worth fighting.

The part that makes Tesla different from all of those: any software company could copy Adobe's subscription idea once it worked. A regular car company can't easily copy what Tesla is doing, because they don't have Tesla's years of self-driving software and over-the-air update technology already built into every car.

There's a real reason to think the $99 might be worth it now: the software actually got a lot better. By mid-2026, Tesla had combined its separate driving systems, highway driving, city driving and fully driverless robotaxi driving, into one. A Forbes reviewer who tests these systems regularly rode through Los Angeles traffic in June 2026 without touching the wheel once and said it felt like a real self-driving car, not just fancy cruise control. Tesla's actual driverless taxi service, with nobody in the driver's seat at all, backs that up: in June 2026 it started covering the entire Austin area, about 245 square miles, and by July it had spread to five more cities. On its earnings call that summer, Tesla said the driverless fleet had driven more than 380,000 miles with no serious incidents, and nearly 1.5 million people worldwide were now paying for FSD in some form. It's still not perfect: one report found the Austin service only running about a fifth of the time it's supposed to, with a crash rate worse than human drivers, and fewer crashes in absolute terms than Waymo's much larger driverless fleet in the same city. But this is the first time in years the product itself, not just the price tag, has visibly moved.

That improvement is also why some Wall Street analysts have started talking about Tesla completely differently. One major bank said the actual car-selling business is only worth about $47 of Tesla's stock price, while the self-driving and robot business could be worth about $330, seven times more. Another bank values just the robotaxi part at up to $750 billion, close to half of Tesla's entire stock market value. One well-known analyst thinks Tesla's total value could double to $2 trillion by the end of 2026, or even triple to $3 trillion if things go really well. But another analyst, looking at the same company, has a price target more than 90% lower, arguing the stock is priced for a future that hasn't shown up yet, and the actual middle-of-the-road Wall Street view is far less excited than either extreme. Nobody agrees, which is exactly why this is a bet, not a fact.

One honest note: this piece includes a bold prediction, that this pricing move helps push Tesla to a $10 trillion stock market value within the next ten years, by around 2036. That's a real bet about the future, not something anyone can prove today. It's written down here, with today's date attached, specifically so it can be checked later, whichever way it turns out.