TechNewsFirst

Tesla's Full Self-Driving package has never had a stable price, and the direction it moved in tells the real story. It launched at $3,000 in 2016, rose to $6,000 in 2019 and $8,000 in 2020, then Tesla added a subscription option in 2021 ($199 a month, or $10,000 outright). The one-time price kept climbing from there: $12,000 in January 2022, then $15,000 that September, after Musk announced the increase himself on social media on August 21, 2022, effective September 5, according to TechCrunch's reporting at the time. That $15,000 mark was the all-time peak. Tesla walked it back to $12,000 in September 2023 as FSD's take rate among buyers reportedly slid, then cut it to $8,000 in April 2024, the same month it dropped the subscription price from $199 to $99 a month. On February 14, 2026, Tesla removed the outright-purchase option entirely in the US, Canada and Europe. What had briefly been a $15,000 asset you could own forever is now, for anyone buying today, a $99-a-month rental you can never fully purchase.

> Here's the number worth sitting with. In July 2026, Tesla's cheapest Model 3 lease, the base Rear-Wheel Drive trim, reset to $379 a month for 36 months, according to Carscoops and CarsDirect's reporting at the time, up from $329 a month the week before, not counting $4,075 due at signing that pushes the effective rate closer to $481 a month once it's spread across the lease. Add the $99-a-month FSD subscription on top of that $379 lease payment and someone renting both the car and its self-driving software is paying roughly $478 a month, before fees, for a car they will never own and a feature they can never buy outright either. The lease number has kept moving since, cut to $299 a month at one point, back up toward $419 and higher by September, so treat $379 as a specific July 2026 snapshot, not a fixed price. What hasn't moved is the shape of it: two products from the same company, a car and a piece of software, both sold now as payments that never stop instead of a price you pay once. Think about that.

It's worth sitting with how Tesla originally sold that rising price, because the pitch was the opposite of a subscription. On Lex Fridman's podcast in April 2019, Musk said, "if you buy a Tesla today, I believe you are buying an appreciating asset, not a depreciating asset," tying that claim directly to FSD: as the software improved, he argued, Tesla would keep raising its price, and every car already carrying the FSD computer would gain value rather than lose it. He repeated a version of the claim on a 2021 earnings call. It didn't hold up. Tesla has cut FSD's price three times since the 2022 peak instead of raising it, and used Teslas are now reported to be depreciating faster than the market average, not slower, a gap Electrek laid out in a March 2025 piece bluntly titled "Elon Musk's biggest lie." By April 2025, a car Musk had described as an appreciating asset years earlier was, according to a widely cited teardown by YouTuber JerryRigEverything, worth only about a fifth of its original price. The honest read: the appreciating-asset pitch was real marketing, made in earnest, and it turned out to be wrong. What replaced it, the flat-rate subscription, isn't a continuation of that story. It's Tesla abandoning it for a model that doesn't depend on FSD ever getting good enough to justify a rising price tag, because a subscription's economics work even when the underlying product never gets dramatically better at all.

That repricing didn't happen on its own. On the Q4 2025 earnings call, January 28, 2026, Musk announced Tesla was ending production of the Model S and Model X entirely, converting the Fremont factory's line to build Optimus humanoid robots instead, a shift he called giving the two vehicles "an honorable discharge." Electrek reported the last S/X units, roughly 600 of them, were still working through inventory as of April 2026. Read next to the FSD pricing history, these are two halves of one strategy: stop building low-volume, high-margin flagship hardware, and stop selling high-margin software up front too. Sell more cars, cheaper, and make the recurring money on the subscription instead of the sticker price. Tesla's own numbers support the motive: CFO Vaibhav Taneja told investors in October 2025 that only about 12% of Tesla owners had ever paid for FSD in any form, meaning the case for a much lower monthly price, to convert some of that other 88%, is straightforward math, not just theory.

The comparison worth taking seriously isn't to other car companies. It's to every industry that already made this exact switch, and the receipts are more concrete than "software went subscription" as a vague gesture. Adobe killed Creative Suite, its roughly $2,500 one-time bundle, in May 2013, replacing it with Creative Cloud at $49.99 a month; the backlash included a Change.org petition past 50,000 signatures, and the payoff was Adobe's revenue growing from $4.4 billion to more than $15 billion in the years that followed. Autodesk ran the same play more slowly with AutoCAD, ending new perpetual licenses on January 31, 2016 and fully retiring support for old perpetual licenses by 2021.

Gaming did it too, and the numbers are bigger than most people realize. Take-Two's own SEC filings show that in fiscal year 2023, what the company calls "recurrent consumer spending," mostly in-game purchases inside Grand Theft Auto Online and Grand Theft Auto V, made up 78.1% of $5.35 billion in net revenue, north of $4 billion from microtransactions alone, more than a decade after GTA V's original 2013 release. A single blockbuster sale used to mean one large payment and then a cliff. GTA Online turned that same game into a revenue stream that hasn't stopped. A separate 2025 data breach at Rockstar's parent company, reported by outlets including RockstarIntel, leaked internal figures suggesting GTA Online's "Shark Card" microtransactions had generated roughly $5 billion since launch, with a small fraction of players, around 4%, accounting for most of that spending. That figure comes from a leak rather than an official disclosure, so it's offered here as directionally consistent with, not equivalent in reliability to, Take-Two's own audited number. The wider mobile-app economy tells the same story at even bigger scale: industry trackers compiled by BusinessOfApps and similar aggregators estimate mobile games alone generated roughly $82 billion from in-app purchases in 2025, with subscriptions now making up close to 44% of all iOS App Store revenue. Those are compiled industry estimates, not company-disclosed figures the way Take-Two's SEC filing is, but they point the same direction: recurring, incremental payments have become the dominant way software, games and apps make money, replacing the one-time sale almost everywhere it used to exist.

The part of this that isn't about software at all, and that explains why this pricing model is so durable once it's in place, is what happens to the price after the switch. Amazon Prime launched at $79 a year in 2005. It went to $99 in 2014, $119 in 2018, and $139 in 2022, a roughly 17% jump CNBC covered as it happened. Nobody canceled Prime en masse at any of those steps; each increase came years apart, bundled with new perks, and the absolute dollar amount per hike stayed small enough not to register as a real decision. The self-storage industry has turned the same idea into a named, formal practice: what the trade press calls ECRI, Existing Customer Rate Increase, where an operator advertises a low rate to win a new customer, then raises that specific customer's rent afterward in small steps, often twice a year rather than once, because staying under what pricing researchers call the "just noticeable difference" threshold generates 10 to 20% more revenue over time than one larger annual increase, according to Inside Self Storage, a trade publication that discusses the practice openly. That's the mechanism this piece's original framing described as a lobster in slowly heating water, and it's a real, named, well-documented industry strategy, not just a metaphor: a small enough increase, spaced out enough, doesn't feel like a decision to the person paying it.

Here's the part of this argument that's specific to Tesla, not just a repeat of everyone else's playbook. Adobe's competitors could copy Adobe's subscription model once it worked, because making creative software doesn't require anything Adobe alone had. A legacy automaker cannot copy what Tesla is doing on any comparable timeline, because Tesla's advantage isn't the pricing decision, it's the decade of over-the-air software infrastructure, in-house chips and sensors, and fleet data that makes a $99-a-month self-driving subscription technically deliverable at all. Ford or Toyota deciding tomorrow to sell a driving-assistance subscription doesn't have the underlying software business to rent out.

None of this works, though, if the thing being rented is still bad, and that's the part that has visibly changed. By June 2026, Tesla had pushed Full Self-Driving through a run of updates, versions 14.3.2 through 14.3.4, that unified what used to be separate systems for highway driving, city-street driving and unsupervised Robotaxi driving into one. Forbes contributor Brooke Crothers, who tests driver-assistance systems across multiple car brands as a regular part of his coverage, wrote on June 14, 2026 that his latest FSD drive through Los Angeles freeway traffic and downtown Burbank required no intervention at all, and that the software now "feels more like a Level 4 driverless ADAS" than earlier versions, meaning a system built to operate without a supervising driver, even though Tesla still requires one in the car. Tesla's actual driverless service backs that up with scale, not just one reviewer's drive: on June 3, 2026, Tesla switched on unsupervised Robotaxi rides across the entire Austin metro area, a roughly 245-square-mile zone reaching the suburbs of Pflugerville and Manor, stretches of I-35 and the airport, and by July 2026 the service had expanded into Houston, Dallas, Miami, Orlando and Tampa. On its Q2 2026 earnings call in early August, Tesla disclosed the unsupervised fleet had logged more than 380,000 cumulative miles across six cities in two states with, in its own words, zero notable incidents, and that nearly 1.5 million customers globally were now paying for FSD in some form, with new subscriptions outgrowing new one-time purchases for the first time. None of that means the technology is finished, or that Tesla's marketing about it has gotten more honest: Electrek's own count found the Austin service running at only about 19% availability eight months after safety-monitor removal began, with a crash rate the outlet put at nine times worse than human drivers, and NHTSA filings reviewed by other outlets show Tesla's Austin fleet logging 14 crashes since its June 2025 launch versus 51 for Waymo's Austin fleet, which is roughly five times larger and has been running fully driverless for longer. Set against FSD's decade of price history, though, this is the first stretch where the underlying product has visibly, measurably improved rather than the company simply changing what it charges for the same thing, and that's the case for why $99 a month now buys something closer to what Musk promised in 2019 than it did in 2022, even with real gaps still showing up in the numbers.

That improvement is also why Wall Street has started arguing about Tesla's value in a completely different vocabulary than "car company." Morgan Stanley formally redefined Tesla as an AI platform company in a March 18, 2026 report, assigning roughly $47 a share to the core vehicle business and about $330 a share, seven times as much, to Robotaxi, network services and Optimus combined, while projecting Tesla's combined supervised-and-autonomous robotaxi fleet growing from around 1,500 vehicles by the end of 2026 to roughly 30,000 by 2030. Bank of America has separately put the robotaxi business alone at up to $750 billion, close to half of Tesla's entire market capitalization, in a note built on similar logic. Wedbush's Dan Ives, Tesla's most consistently bullish major-bank analyst, told clients in December 2025 that he expects the stock to reach a $2 trillion market cap by the end of 2026, with a bull case as high as $3 trillion, on the strength of the robotaxi rollout, an "AI chapter" he argues the market hasn't finished pricing in. That's the structural version of this piece's subscription argument: if the software behind the $99-a-month FSD subscription is the same software scaling into robotaxi network revenue, a business no legacy automaker currently has any version of, then the subscription price is a leading indicator, not a footnote, for a business some analysts already argue is worth more than the cars themselves. Optimus belongs in that same bet, not next to it: Tesla's own framing for the robot business, converting Model S/X manufacturing lines to build it, is the same playbook as FSD, get the hardware into the field cheaply, even at a loss, and monetize it afterward through software and service revenue rather than the upfront sale. Nobody has priced Optimus as precisely as Morgan Stanley has priced Robotaxi, but the mechanism analysts are pointing to for both is identical: lower the price of getting the hardware into someone's hands, then charge recurring revenue for what the hardware does.

The other side of that argument is just as real and belongs in the same paragraph, not a caveat at the bottom. Tesla's 2025 revenue fell to $94.8 billion, its first annual revenue decline as a public company, as deliveries dropped 8.6% to 1.64 million units and BYD outsold it globally in pure-electric vehicles for the first full calendar year on record, 2.26 million to Tesla's 1.64 million, though Tesla reclaimed the global lead in the first quarter of 2026. By one widely cited measure Tesla trades at roughly 330 times trailing earnings, a multiple GLJ Research's Gordon Johnson cited in reiterating a $25.28 price target and Sell rating as recently as March 2026, more than 90% below where the stock traded at the time; Wall Street's actual consensus sits closer to a $407 average target with a Hold rating, meaning even mainstream analysts are, on balance, far less bullish than Ives or Morgan Stanley's reframing. Robotaxi and FSD together are still a small fraction of Tesla's overall revenue today, and nobody serious is claiming otherwise. The bull case for a market cap in the trillions rests on that revenue scaling from a promising trend line into an actual business over the next several years, which is exactly the gap this piece's own $10 trillion prediction, below, is betting will close.

One small piece of firsthand color, since this outlet's whole reason for existing is coverage from people who've actually used the tools: this outlet has owned a Tesla Model X for years, and it remains, Falcon-wing doors and all, one of the best cars we've ever driven, well before any of this pricing math entered the picture. FSD itself, across that ownership, has never been the standout feature the marketing, including the appreciating-asset pitch, suggested it would be. The bet in this repricing isn't that FSD has to get dramatically better to make the subscription work. It's that $99 a month is cheap enough that most owners keep paying regardless, the same way most Prime subscribers don't cancel over a $20 annual bump.

One claim in this piece is a prediction, not a fact, and it's being labeled as one on purpose: that this pricing shift will be a primary driver behind Tesla reaching a $10 trillion market cap within ten years of today, September 19, 2026, meaning by roughly September 2036. Tesla's market cap as of this writing is a small fraction of that figure. This is a specific, dated, checkable thesis about where the stock goes, built on a real and verifiable change in Tesla's business model, not an assertion that it has already happened or is guaranteed to. Readers, and this outlet, should hold Tesla and this piece to that number, on that timeline, rather than let the prediction quietly disappear if it doesn't pan out.