Tesla Is Betting Its Entire Future on Robotaxis

A Car Company the Market Stopped Valuing as a Car Company

Tesla stock is up about 60 percent so far in 2026 and here's the strange part. Almost none of that rally is about cars. Vehicle sales which is still what Tesla really does for a living have been under pressure from tougher competition and weaker demand for most of the year. What has investors excited are robotaxis. The market has decided that Tesla's future is not selling cars to people. It is operating fleets.of autonomous vehicles that make money 24 hours a day and the shares are priced as if that future is close to being assured.I want to spend this article breaking down that bet because most of the discussion about Tesla happens at the level of vibes car company vs. AI company rather than at the level of what actually has to be true for the current price to make sense

What Tesla Has Actually Launched

Progress is real and at an earlier stage than the stock price suggests. Tesla launched a paid robotaxi service in Austin in late 2025. Through 2026 it expanded that service to about a dozen cities offering driverless rides to the public in places like Dallas and Houston with more cities announced. Tesla has also begun producing the Cybercab a specially designed robotaxi without a steering wheel or pedals at its Texas factory. It is a type of vehiclegenuinely new not a Model 3 with extra sensors bolted on. After years of promised deadlines coming and going there are now paying customers traveling in cars with no one sitting in the driver's seat. I don't think that fact should be ignored. It's a true technical and regulatory achievement. The question of interest to this article is independent of whether the technology works today. It's about whether the size of the achievement so far justifies the size of the bet the stock is making

The Gap Between the Story and the Numbers

Here's the tension. Despite all the hype the robotaxi business is still small. The revenue generated by it is irrelevant for a company the size of Tesla in 2026 and significant revenue specifically from Cybercab is unlikely to be made before 2027 at the earliest. Sit with that for a second because it's the whole game. You're not paying the current price for what the robotaxi business makes right now which is next to nothing. You're paying for what it makes.You could win years from now if a long chain of things happens between here and there

That distinction present claim versus future claim is why a single earnings call a single regulatory approval or a single bad headline about a self-driving accident can move stocks five to ten percent in a day. None of those events change this quarter's earnings much. All of them change the likelihood that investors think it's the future

The One Thing It All Depends On

The whole bet is on the software. Specifically Tesla is finishing a version of its Full Self Driving system reliable enough to work with. zero human supervisionTesla has indicated that unattended autonomous driving for customers is probably a Q4 event and that word probably carries something like a trillion dollars of market expectations behind it. If the software arrives and actually works at that level of generality the robotaxi math could be extraordinary for reasons I'll explain with the numbers below. If it fails again as the self-driving schedules in this industry have goneDelayed again and again for a decade the gap between story and numbers becomes very difficult to justify. I don't think there's a way to resolve this plot intelligently. It's resolved with a working car or a broken promise and until either of those happen the action is a bet on which comes first

Why the Prize Is So Big

The reason investors endure all this uncertainty is the size of what's on the table. Global transportation is a market measured in trillions of dollars and a genuinely autonomous fleet eliminates its biggest cost: the human driver. Right now a typical car sits parked about 95 percent of the time. It's a depreciating asset that for the most part just sits there. A car that can drive itself and pick up strangers can in theory run most of the day instead of sitting around.in the driveway turning that same depreciating asset into something more akin to a cash-generating machine. If Tesla really manages to solve this at scale the economics will be genuinely transformative not incrementally better. yes You're doing a tremendous amount of work on that sentence and I want to spend the rest of this article showing you exactly how much work

Splitting Tesla Into Two Businesses

That's how I really think about separating this and it's a method you can use in any company that has a boring cash-generating business coupled with a speculative story-based business. This is a classic sum of the parts approach. You value the boring part on its own terms using the kind of multiple that the market offers comparable boring businesses. You then subtract that value from the market capitalization of the entire company. What's left is what the market is implicitly paying for the interesting part whether it's stated in some press release or not

For Tesla the boring part is the car business: designing manufacturing and selling electric vehicles which is a real profitable and competitive industry with decades of history and well-understood valuation multiples. The exciting part is everything related to the robotaxi the Cybercab the Full Self Driving software the fleet operations the whole story behind the 60 percent rally. Splitting the two is not just an academic exercise. It tells you in dollar terms how much of your investment depends on a car companythat you can finance with a spreadsheet and how much depends on a bet that is ultimately delivered with unsupervised software

A Worked Example: How Much of the Price Tag Is Robotaxi

I'm not going to use Tesla's actual delivery numbers or its actual market capitalization here. Both change every week and neither is something I can pin down for this article. Instead here's the mechanism with round clearly labeled illustrative numbers so you can run through the exact same arithmetic again once you look up the actual numbers

Suppose a car company delivers 1.8 million vehicles a year at an average selling price of $45,000. That's 1.8 million times 45,000 or $81 billion of illustrative annual automotive revenue. Given a net margin of 6 percent which is healthy for an automaker you get a net income of 81 billion times 0.06 or4.86 billion dollars

Now let's value that auto business the same way the market values regular automakers based on an earnings multiple. Traditional automakers often trade in the high and low single digits based on price to earnings. I'll be generous and use 15 times earnings above what most traditional automakers actually trade. That's 4.86 billion times 15 or about $73 billion. Let's call itto the value of the car business alone valued as a car company

Let's now assume that the entire company cars plus robotaxi has a market capitalization of 1 trillion dollars an even trillion again purely illustrative and chosen to be round rather than exact. Subtract the 73 billion value of the cars from that trillion and you are left with 927 billion dollars. That is the amount the market is implicitly allocating to robotaxis and everything adjacent to them with none of them generating significant revenue yet. As a percentage of theTotal 927 divided by 1,000 equals approximately 93 percent

That's the number I want you to sit with. Even after being generous with the car multiple more than nine-tenths of the illustrative market cap in this example belongs to a company that earns close to zero today. Do this again with Tesla's actual delivery numbers the actual market cap and whatever car multiple you think is fair and I bet you'll land somewhere in the same neighborhood. The vast majority of the stock's value is a bet on softwarewhich does not yet exist in its final form

Case Study: Waymo's Slower, Already Operating Bet

If you want to see what the other strategy looks like look at Waymo Alphabet's self-driving unit and a direct descendant of Google's self-driving car project which started in 2009. Waymo took an almost opposite approach to Tesla on almost every ax that matters here

As Tesla tries to solve autonomous driving with cameras alone a bet that vision plus enough computing power can eventually match or surpass a full suite of sensors Waymo built its stack around lidarThat approach is more expensive per vehicle. It is also by most technical accounts more conservative and easier to validate as safe because the car doesn't try to infer depth and distance from flat camera images like the human eye does

Waymo also moved slower and more locally. Instead of launching in a dozen cities in one fell swoop it built out fully driverless no-humans-in-the-seat operations city by city starting in Phoenix years before Tesla launched in Austin and then carefully expanding to places like San Francisco and Los Angeles as it gained local regulatory approval in each. It's not trying to sell you a car. Alphabet doesn't need Waymo to make and sell millions of vehicles to consumers like Tesla does. WaymoIt just has to make the transportation part of the business work and it's been generating real paid fully driverless rides for real customers for a while now even if the fleet is still much smaller than a mass-market operation would need to be

Contrast is the point of including Waymo in this. Tesla is betting that a cheaper sensor approach once the software is good enough wins because it scales faster and costs less per car. Waymo is betting that going slower market by market with more expensive but easier to validate hardware is the surest path to something that actually works and won't be out of commission after a serious incident. Neither company has yet proven that its approach is the right one. But Waymo is aProof that a slower lidar-based jurisdiction-by-jurisdiction strategy can already put paying passengers in a driverless car which matters when weighing how much credit to give to Tesla's faster cheaper camera-only bet

The Bull Case and the Bear Case, Argued Straight

I think both sides of this argument are more serious than internet discussions typically allow so let me make each one as strong as I can

The bull case starts with vertical integration. Tesla designs the car builds the factory writes the self-driving software and will operate the fleet all under one roof. That's unusual. Most robotaxi efforts are a software company installing sensors in someone else's car or an automaker clumsily partnering with a software company that doesn't control manufacturing. Tesla controls the entire stack meaning it can optimize the vehicle for the software's needs and not just bysetback.Add to that the advantage of fleet data.Millions of Teslas already on the roads are running supervised versions of full self-driving and sending actual driving data to the company a training data pipeline that most competitors simply don't have at that scale.Then there's the cost per mile.A single-camera sensor array is dramatically cheaper to manufacture than a heavy lidar one and if software ever closes the capability gap Tesla could operate at a lower cost per vehicle than anyone else.use expensive laser sensors. Put those three elements together vertical integration a data moat and a structural cost advantage and the bull case is not far-fetched. All three come together if the software works

The bear case is equally serious.Start with the timeline.Tesla has been approaching unattended autonomous driving for years and the goal has repeatedly missed. A fourth-quarter event with different words tied to different quarters has probably been said before before and a repeated sliding pattern should also update your confidence about the next promised date.Then there's regulation which doesn't move in a single step at the national level.Each state and often each city sets itsWaymo described above already has paying passengers in fully autonomous cars today using a more conservative and arguably safer sensor approach with a multi-year head start in operational experience in the cities it serves. Yesregulators insurers or the public end up trusting lidar-based systems more because they have a longer track record Tesla's bet on the camera could win the cost argument and still lose the trust argument

How I'd Actually Watch This Stock

My honest read: I find this really difficult to model and I would be suspicious of anyone who tells you they have the exact price. What I really do and this is how I think of it personally not as advice on what someone else should do with their own money is look at a small number of concrete signals rather than looking at the stock price itself

The first is whether unattended human-in-the-seat rides expand to new cities as planned or if the promised dates continue to move forward as they have before. That single pattern tells me more than any analyst note. The second is regulatory approvals tracked one jurisdiction at a time because that's really how this plays out city by city and state by state not in a national announcement. The third is what's happening specifically with the Cybercab production numbers aside from production.ordinary Model Y or Model 3 since that is the specially designed vehicle the entire commitment of the robotaxi depends on its scale manufacturing

The way I would actually use the example above is not to plug in Tesla's actual numbers spit out a fair value and then trade it. It's to remind myself every time the stock jumps or falls sharply on a headline that the move is almost exclusively due to the robotaxi's share of the number not the auto business which is the duller and more stable of the two businesses that trade under a single symbol. When I catch myself reacting to Tesla news as ifWhen it comes to auto sales it's usually a sign that I've lost track of what's actually being priced. I myself was wrong the first time I read a Tesla delivery report and expected the stock to move like the stock of a regular automaker would

The Bottom Line

Tesla in 2026 is actually two companies with the same symbol a pressured automaker and a robotaxi startup to the moon and the stock price is overwhelmingly close to the latter. The example above which uses illustrative numbers instead of Tesla's actual figures shows why. Even a generous valuation of the auto business in ordinary terms of the automakers leaves the vast majority of a plausible market capitalization to be explained only by robotaxis a business that today barely earnsNothing. Waymo shows that a slower more expensive and more conservative path can already put paying passengers in a self-driving car which is the clearest evidence that Tesla's faster cheaper approach is a real bet with a real alternative not a foregone conclusion. Whether this recovery turns out to be visionary or premature comes down to a single question that will not be resolved no matter how much it is argued: whether unsupervised software ultimately works everywhere as intended. Until then Tesla remains one of the bets.purest information on the market about a future that has not yet arrived