Wall Street Isn't Doubting Tesla's Robotaxi. It's Pricing Everyone Else's.

On September 3, Tesla pulls the cover off its production Cybercab in Austin — a two-seat robotaxi with no steering wheel and no pedals, rolling off the line at Gigafactory Texas in the same city where autonomous Teslas have been carrying passengers since June 2025. The stock is already trading as if the date decides the decade. Bank of America figures the robotaxi business accounts for roughly half of Tesla's $1.4 trillion market value, more than double the slice it assigns to the car business that actually pays the bills. The familiar story says the skeptics keep doubting and the product keeps working, and one day the doubters capitulate.

The most instructive skeptic no longer argues the technology will fail. Gary Black, managing partner of the Future Fund, says Wall Street's hesitation is not about growth at all. He reads the analysts' models this way: they assume Tesla wins approval for unsupervised autonomy at about the same time as a handful of other players — and that the skill therefore "quickly becomes a commodity". They build a small robotaxi for Tesla not because they doubt the engineering, but because they expect the market to be shared. Market share, not capability, drives their price target.

That is the entire argument in one frame. The robotaxi debate was never really "does autonomy work." It was always "who still earns a margin after everyone can do it?" The two sides answer differently, and the answers are hundreds of billions of dollars apart.

Everyone Agrees the Technology Will Work

Start with what each side silently assumes. Musk has mused aloud that Tesla might take "99% market share or something ridiculous" of robotaxis. The credible bulls who build actual models are only a little less greedy: Ark Invest sees Tesla holding half the global robotaxi market by 2030, and Wolfe Research reaches the same 50 percent — just five years later. Those are near-monopoly numbers for a service that did not exist at scale a year ago.

Black's retort is that Wall Street has run this exact subtraction before. Half a decade ago, enthusiasts insisted Tesla would sell 20 million vehicles a year by 2030 — an implied share of roughly a fifth of the global car market. The street's eventual forecast came in near 3.7 million units, under 5 percent. The technology back then was real too. History, as he likes to say, does not repeat, but it often rhymes.

The Numerator Nobody Prices

Now look at the metric the bull case is built on. The numbers enthusiasts celebrate — vehicles on the road, cities mapped, miles of self-driving data — all measure supply. They describe how large the machine is, not how much it earns. And the machine is tiny. Regulatory records showed 59 robotaxis operating across three Texas cities in mid-June; by August the fleet was "stuck" around 90 to 100 vehicles, by Black's reading. Musk's own milestones slipped: the "half the U.S. population" coverage he promised for the end of 2025 never arrived, nor did his "500 or more" Austin-area vehicles. Early riders describe half-hour waits, and Cathie Wood — the stock's most devoted cheerleader — paid $75 for her celebrated ride. Tesla itself says the robotaxi service will not be profitable until at least 2027.

None of this proves the technology is fake. It proves the celebrated number is the numerator of a question nobody is asking: how many rides can this fleet sell, and at what price?

Whoever Owns the Rider Wins

The companies already answering that question own the other side of the transaction — the riders. Waymo delivers about 500,000 paid rides a week across ten U.S. cities and explicitly targets a million weekly rides by the end of 2026. It registers more than 600 automated vehicles in Texas alone, roughly ten times Tesla's entire announced fleet. Uber builds no self-driving technology at all, yet reaches about 183 million monthly active users across more than 15,000 cities and has committed over $10 billion to more than 30 autonomous-vehicle partners, from Waymo to Wayve to Rivian. Black calls that an "open source supplier model", and it is why he thinks Uber — not Tesla, not Waymo — is best positioned to bring robotaxis to the masses.

That comparison is the whole thesis wearing work gloves. If autonomy becomes what Wall Street assumes — a commodity — the car itself stops being the moat. A driverless mile from one vendor will cost about what a driverless mile from the next costs, because the software is converging and the hardware is plentiful. What does not commoditize is the demand layer: the app where 183 million people already decide which car to take, the utilization data that tells a fleet where to wait, the brand a nervous passenger opens at midnight. The owner of a commodity fleet has to buy riders one geofence at a time — which is precisely what Tesla is doing, mapping a sliver of Miami (West Miami to Doral and Sweetwater) while its Texas territory still cannot scale.

Priced Before the First Cybercab Is Sold

Then price the agreement. Tesla trades near 360 times trailing earnings, and its forward multiple is actually higher, above 430 times — an unusual tell that the underlying business is expected to shrink before the robotaxi saves it. The carmaker's revenue grows about 12 percent a year, its operating margin is near 5 percent, and its return on invested capital is in the low single digits: the profile of a mature business wearing the multiple of a monopoly about to be born. By Bank of America's math, that monopoly is already more than half the valuation. The market has decreed, before a single Cybercab is sold and while the active fleet is still in triple digits, that Tesla is mostly a robotaxi company. Against that, Uber trades at a little over 16 times earnings, after a quarter in which it grew operating income 30 percent to $1.89 billion. One of these is priced for outsized share. The other is priced for ordinary commodity economics.

The Test

The honest counterargument is that vertical integration might be exactly the edge that matters once software stops being scarce. Tesla pays no driver, splits no fare with a dispatch layer, and controls the car from camera to charger. If commodity autonomy pushes the price of a driverless mile toward cost, the producer with the lowest cost wins, and Tesla's unsupervised economics plausibly are the lowest in the industry. Bulls add more than eight billion miles of real driving data as a widening moat.

Black does not dispute that the engineering superiority is real. He disputes that it survives contact with a dozen funded rivals and a demand layer that already exists. The disagreement therefore reduces to a question of price and speed — and that is what makes it testable rather than a slogan. If the September 3 event is followed by a fleet that actually reaches the thousands, a 2027 profitability timeline that holds, and service areas that widen into real cities, the scarcity premium is real and the commodity frame was premature. If the fleet is still in the hundreds a year into Cybercab production while the market keeps capitalizing half the company on it, the commodity frame wins — and the correction does not need a crash. It needs only the slow quarterly subtraction of hope.

The robotaxi debate was always going to end one of two ways. Either Tesla is cheap enough to operate that it earns a fortune on an outsized share of the market, or autonomy becomes what it has become everywhere else: table stakes, available to anyone with capital, with the margin migrating to whoever owns the rider. Wall Street has quietly priced the second outcome. The stock's multiple is a bet on the first. Both cannot be true, and the people whose jobs are safest repeating the robotaxi story are not the ones who will be asked to pay the difference. That difference is the whole trade.