**By Nicholas Thomas · Saturday, September 5, 2026 — TrendyVest Editorial. Opinion, with receipts, fact-checked before publication. Not investment advice.** --- Every mayor in America is about to get the same pitch, and I'd like you to have the numbers before they do.
Here's how it will go. Tesla started building the Cybercab at Giga Texas this year — a two-seat pod with no steering wheel, no pedals, gull-wing doors and a 22-inch screen — with capacity, per Axios, for **125,000 a year**, and a stated ambition to make them "by the millions." The company has "previously discussed a target price near $25,000 to $30,000 per vehicle." On Thursday it put up a form titled *"Help Us Build Our Robotaxi Network,"* inviting businesses to express interest in "Cybercab fleet vehicle purchasing" and "mobility hubs and infrastructure." The first forty-five of these pods are registered in Texas, out of 420 Tesla robotaxis statewide, and they're taking paid rides in Austin at what Not a Tesla App infers from the app's pricing to be about **$3 plus $1.40 a mile.** One sample trip came to $7.77, against $12.10 in a Model Y robotaxi — roughly 35 percent cheaper. So here's the idea, and it isn't crazy.
A city buys a thousand of them for $30 million, runs them as a public fleet, and keeps the fares. No medallions to sell. No Uber to tax.
No transit subsidy to defend at budget time. A robotaxi utility that funds itself, and maybe the buses too. I spent the weekend trying to make it work on paper.
It does — right up until it doesn't. Let me show you both halves. **THE HALF THAT WORKS.** Use the only real benchmark we have. Waymo, per Sacra's February figures, was running about **450,000 paid rides a week on roughly 2,500 vehicles**, at **$15 to $17 a ride**, for an annualized **$355 million** in revenue.
That works out to 180 rides per vehicle per week — about 26 a day — and about **$142,000 of revenue per vehicle per year**, on a car that costs "$80,000 or more" once it's fully sensored. Waymo's fleet pays for its own hardware in roughly seven months of gross revenue. Now run a Cybercab through the same utilization at Cybercab prices.
Twenty-six rides a day at the Austin sample fare of $7.77 is about **$200 a day, or $73,000 a year per cab.** That's this desk's arithmetic, and I'll label it as such every time. Against a $30,000 vehicle, that's a hardware payback of five months. Operating cost is where the argument lives.
Musk has claimed "around 20 cents a mile" at scale; an independent build I read (Taha Abbasi, February) puts a *mature* robotaxi at **30 to 50 cents a mile** all-in — depreciation, energy, maintenance, insurance, remote monitoring, software, cleaning — against fares of $1.00 to $1.50 a mile, for gross margins of 50 to 70 percent. Call it 150 miles a day per cab, empty miles included, at 40 cents: **$60 a day, about $22,000 a year.** Net, roughly **$50,000 a year per cab.** A thousand cabs: **$50 million a year** on a $30 million purchase. If those numbers held, a mid-sized city could replace its parking-meter revenue with a fleet that also cut fares for its residents.
That's why the pitch will land. And that's why you should read the second half before you vote. **HOLE ONE — FEDERAL. The car a city would be buying is under audit, and its maker hasn't asked for the permit that would make it legal at scale.** This week NHTSA opened Audit Query AQ26002 on roughly 1,000 Cybercabs, to examine "the process and technical data on which Tesla relied when certifying" a vehicle with no steering wheel to standards that assume one.
The agency's own words: "Until [the rulemaking] is completed... existing standards remain in force." And, per Axios, Tesla "has not filed an exemption request for the Cybercab." The precedent is Zoox, which self-certified the same way in 2022, spent about two years in a standoff, and ended up with an exemption capped at **2,500 vehicles a year.** A city that buys a thousand Cybercabs today is buying a thousand vehicles whose legal basis is an open federal question, from a company betting it can avoid the cap. I don't know a bond counsel who signs that. The federal hole closes only when Tesla either wins the audit or takes the exemption — and if it takes the exemption, the whole national supply is 2,500 a year, and your city gets a slice of it. **HOLE TWO — CONTRACTUAL.
You'd own the car. You would not own the driver.** A Cybercab has no value without the software that drives it, and that software is a subscription Tesla has never priced for a third party. Here's what we know.
At Autonomy Day in 2019, Tesla said it would "take 25–30% for operating the network" from owners who put cars on it, on the way to promising them "as much as $30,000 per year in profit" — a promise that, per Electrek, no private owner has collected in the year Tesla has run robotaxis commercially. Here's what we don't know: anything current. The interest form discloses "no information about vehicle costs, revenue-sharing arrangements, or licensing fees." Rerun the math with a 30 percent platform cut and the city's $73,000 of revenue per cab becomes $51,000, its $50,000 of margin becomes about $29,000, and the "utility that pays for itself" is now a customer of Tesla's that happens to hold the depreciation.
That isn't a knock on Tesla. It's the business. Whoever owns the driver sets the price, and the driver is a server in Austin.
The city-owned fleet is the *hardware* half of a business whose margin lives in the *software* half — and the form, as TechCrunch read it, describes "outside operators participating in infrastructure Tesla controls." **HOLE THREE — HISTORICAL. Cities have monetized the taxi before. Ask a medallion owner how it went.** New York capped its yellow-cab medallions at 13,587 and let the market price the scarcity.
The price peaked at **$1.3 million in 2013** and fell to **$241,000 by March 2017**, by which point Uber had 51,000 affiliated vehicles in the city and lenders had foreclosed on 39 medallions in a year, more than triple the year before. The drivers who'd borrowed against the asset were ruined; in one owner's words, the city "abdicated its responsibility." The lesson isn't that cities should stay out of the taxi business. It's that a city which owns a *depreciating technology asset* — and a self-driving pod in 2026 is about as fast-depreciating as an asset gets, because next year's model has better cameras and last year's may have a recall — is exposed exactly the way medallion owners were: to a better technology arriving from outside.
The medallion was worth $1.3 million until a smartphone app existed. A city fleet is worth $50,000 a cab until a Waymo, a Zoox, an Apollo Go or a cheaper Cybercab pulls up to the same curb at a lower fare — two of those four already operate in U.S. cities, and a third runs in Dubai. **WHAT CITIES DO NOW, AND WHY IT'S HALF RIGHT.** The current model is to tax the flow. New York charges **$2.75 per ride-hail trip** below 96th Street, with another $1.50 coming below 60th, plus a 2.5 percent Black Car Fund surcharge and sales tax — "about $4.25 plus" on a typical Manhattan trip.
Chicago levies a 60-cent ground-transportation tax on every ride and a downtown-zone surcharge of **$2.75 a ride, all day, since January 2025**, with $5 for special-zone pickups. San Francisco takes 3.25 percent; Seattle about 85 cents in stacked fees; Massachusetts a flat 20 cents it hasn't touched since 2016. Lewis Lehe, who catalogued them all, notes that "only Chicago and NYC raise a lot of money." Now apply that to a $7.77 Cybercab ride.
A $2.75 surcharge is a **35 percent tax** — precisely the discount that makes the service worth using. Cities that tax the flow at today's rates will tax the robotaxi out of the very ridership they want to monetize, while Tesla or Waymo keep the software margin. That's the worst of both worlds: no asset, no upside, and a fare that has stopped being cheap. **THE MODEL THAT WORKS: OWN THE CURB.** Here's what I'd put in front of a council instead.
It's a model cities already know, because it's how they run airports. *One — own the ground, not the machine.* Tesla's own form lists "mobility hubs and infrastructure" right beside "fleet purchasing." That's the tell. A robotaxi fleet needs depots, chargers, cleaning bays, curb space and staging lanes, and those are real estate — the one asset in this story that appreciates while the pods depreciate. A city that owns the hubs and leases them to Tesla, Waymo, Zoox and whoever comes next collects rent from all of them, bears no certification risk, and never owns a car a recall can strand. *Two — charge by the mile, not by the ride.* A per-ride surcharge punishes short, cheap trips, which are the whole point of a $7.77 cab.
A per-mile road-use fee on autonomous miles — including the empty ones, which is where the congestion lives — scales with the thing the city actually sells, which is road. Tesla's 2019 model assumed **50 percent empty miles.** Whoever generates those should pay for them, and a per-mile fee is the only instrument that prices them. *Three — take the data as rent.* Every autonomous mile is a survey of the city. The permit condition that matters isn't a fee; it's the feed — trip origins, curb dwell times, incidents — because that's what lets a city run its buses and its signals around the fleet instead of against it. *Four — if you must own vehicles, lease them, and size the exposure to the exemption.* Dubai's RTA runs the most aggressive government robotaxi program on earth — 4 million kilometers, 144 vehicles, a target of 25 percent of all trips autonomous by 2030 — and it doesn't build robotaxis.
It partners: Baidu's Apollo Go, WeRide, Pony.ai, Uber's platform, a local fleet operator. The government sets the target and the rules and licenses the operators. That's the orchestrator model, and it's the only one that survives a technology cycle. **WHAT IT MEANS FOR THE STOCK, BECAUSE THIS IS STILL A MARKETS PAGE.** Tesla closed Friday at $353.89, down 5.97 percent on the launch and the audit.
The fleet-sales form is a business-model pivot wearing a marketing page: it moves Tesla from *operating* a network at 50-to-70-percent gross margins toward *selling hardware* at automotive margins plus an undisclosed software toll. Electrek reads that as outsourcing infrastructure "reflecting financial constraints rather than technological readiness," and points out that Tesla abandoned its own San Francisco charging-hub project. I'd read it more neutrally.
A company that can build 125,000 pods a year can't operate 125,000 pods a year, and it needs balance sheets that aren't its own — fleet operators, and yes, cities — to hold the depreciation. The investor question is the take rate. If Tesla licenses the driver at something like its 2019 number, fleet sales are a margin *upgrade*: automotive gross profit up front and a 25-to-30-percent annuity on every fare forever.
If regulators or competitors force the toll down, Tesla becomes a very good bus vendor. Nobody outside the company knows which, and the form doesn't say. **THE ROWS, IN THE SAME FONT AS ALWAYS.** *One:* **60% that no U.S. city or public transit agency announces a Cybercab purchase order by June 30, 2027.** Mechanism: an open federal audit, no exemption on file, and municipal procurement cycles measured in fiscal years. Falsifier: a Texas or Florida city — Tesla's launch states — announcing one, most likely dressed up as a "pilot." *Two:* **65% that Audit Query AQ26002 is still open on December 31, 2026.** Mechanism: the Zoox precedent took roughly two years, and audit queries don't close in a quarter.
Falsifier: Tesla filing an exemption request, which would moot the audit and cap the supply. *Three:* **60% that Tesla publishes fleet-customer pricing or a revenue share for the Cybercab by the end of the first quarter of 2027.** Mechanism: an interest form is a pipeline, and pipelines need a price list. Falsifier: Tesla keeping the network closed to third parties for another year, which is what it has done to its own FSD owners so far. *Watch, no odds:* the first city to convert a per-ride surcharge into a per-mile AV fee. That's the first council that read the math. **THE HONEST FINE PRINT.** Every fleet figure above is this desk's arithmetic on sourced inputs — Waymo's rides, fleet and revenue per Sacra; the Austin fare structure as inferred by Not a Tesla App from the Robotaxi app; the cost-per-mile range per Abbasi; the vehicle price per Tesla's prior statements as reported by Teslarati — and it's labeled arithmetic, not forecast.
"Zero notable incidents" over 380,000 unsupervised miles is Tesla's claim, as reported. The NHTSA query is an inquiry, not a finding. No city has proposed the model I'm critiquing; I'm critiquing it before one does.
This is opinion, not investment advice. Do your own research. *Fact-check notes on the first draft, printed because that's the house rule:* it attributed the "infrastructure Tesla controls" line to "one analyst" — it was TechCrunch's characterization, now credited properly. It said three of four robotaxi competitors already run in U.S. cities; Waymo and Zoox do, and Apollo Go runs in Dubai, so it now says two, and a third abroad.
It called the pod "the fastest-depreciating asset on earth," which nobody has measured; it now says about as fast as an asset gets. It described Chicago's downtown surcharge as $2.75 all day *and* $3 at peak; the source describes the $2.75 all-day rate as replacing the peak-hours structure in January 2025, so it now gives the current rate. And it stated the Austin fare structure as fact; it's inferred by the outlet that reported it, and now says so. **ONE LINE BEFORE THE WEEKEND ENDS.** The medallion taught New York that a city which sells scarcity gets destroyed by abundance.
The Cybercab is abundance — 125,000 a year, by the millions, at $30,000 a pod. A city that buys the abundance will own a warehouse of last year's robots. A city that owns the curb the abundance has to park on will collect rent from every one of them, whoever builds them, for as long as there are streets.
Don't buy the cab. Own the curb. — Nicholas **Tickers in play:** TSLA · UBER · GOOGL · AMZN · BIDU · LYFT --- *Sources: Axios, Sept 4, 2026 (45 Cybercabs registered in Texas of 420 robotaxis; 125,000/year capacity; "by the millions"; the fleet-interest webpage; NHTSA "has not filed an exemption request"); Teslarati, Sept 3 (the "Help Us Build Our Robotaxi Network" form and its four options; "target price near $25,000 to $30,000"; "dozens" of Cybercabs testing in Austin); Yahoo Finance/TechCrunch, Sept 3 (form categories; "not definitive proof"; "outside operators participating in infrastructure Tesla controls"); Electrek, Sept 3 (businesses-only eligibility; no pricing, revenue split or licensing fees disclosed; FSD owners' promised income never paid; the abandoned San Francisco hub; "financial constraints rather than technological readiness"); Not a Tesla App, Sept 4 (Cybercab ~$3 base + $1.40/mile and Model Y ~$7 + $1.40, inferred; the $7.77 vs. $12.10 sample; ~264-square-mile Austin geofence; Musk's "around 20 cents a mile"; no onboard monitors); Not a Tesla App's Q2 2026 call summary (Cybercab production commenced at Giga Texas; seven metros; ~2.5 million cumulative paid miles; "0 notable incidents" over 380,000 unsupervised miles; ~10%/week growth); Smart Cities Dive, Jul 23 (Austin, Miami, Orlando, Tampa); Electrek, Sept 4, and Axios, Sept 4 (NHTSA Audit Query AQ26002 opened this week; ~1,000 vehicles; the "process and technical data" and "existing standards remain in force" quotes); TechCrunch, Sept 4 (Zoox's 2022 self-certification; Part 555 exemption; 2,500 vehicles per year); Forbes/Brad Templeton, Apr 23, 2019 (Tesla's "25–30% for operating the network," $30,000/year owner profit, 18-cent incremental cost, $1/mile fare, 50% empty miles, million-mile life); Sacra (Waymo ~450,000 weekly paid rides Dec 2025; ~2,500 vehicles; $355M annualized revenue Feb 2026; $15–17/ride; ~$80,000+ per vehicle; 1 million weekly-ride target); Taha Abbasi, Feb 2026 (mature robotaxi cost $0.30–0.50/mile by component; fares $1.00–1.50/mile; 50–70% gross margins); Critical Density/Lewis Lehe (NYC $2.75 congestion surcharge, $1.50 pending, 2.5% Black Car Fund, ~$4.25 typical; Chicago 60¢ GTT, $2.75 all-day downtown surcharge from Jan 2025, $5 special zones; SF 3.25%; Seattle fees; Massachusetts 20¢; DC 10–25¢; "only Chicago and NYC raise a lot of money"); CBS News (NYC medallions capped at 13,587; $1.3 million peak in 2013; $241,000 in March 2017; 39 foreclosures, more than triple 2015; Uber's 51,000 vehicles; the "abdicated its responsibility" quote); Dubai Media Office, Aug 19, 2026 (4 million km; 7,613 trips; 48 active of 144 vehicles; Apollo Go, WeRide, Pony.ai, Uber, New Horizon; 25% autonomous by 2030); stockanalysis.com (TSLA $353.89, −5.97%, Sept 4 close). Markets.
Tech. The Edge. Research with receipts.*