Tesla Robotaxi vs Waymo: what Cybercab could earn in Europe, and where Optimus fits

The sensible friend and the friend who says it's coming Thursday#

Waymo and Tesla are both trying to build the same thing: a car that picks you up with nobody in the driver's seat. They are going about it in very different ways.

Waymo is the friend with a spreadsheet who has already booked the table. It maps a city, validates it for months, launches, and expands the geofence a few miles at a time. Tesla is the friend who says the thing is coming Thursday, and it has been Thursday for about ten years. It also happens to own a car factory and millions of cars on the road.

I wrote about the long arc of this in How Tesla could become the world's number one company. This post is the narrower, more practical version. Where is Robotaxi actually at in the US as of 1 November 2026? How does it compare with Waymo? What could Robotaxi and Cybercab be worth in Europe if the door ever opens? And does Optimus really have a job in all this, or is it a keynote prop?

Not financial advice. I have a Tesla referral on the home page, I like the product, and I am trying to be fair to both sides.

Waymo today: the boring numbers are the impressive ones#

Waymo's numbers are the benchmark everything else gets measured against:

• Rides. More than 500,000 paid rides a week, up from about 50,000 a week in 2024.
• Cities. Public service in about 15 US cities, including Phoenix, San Francisco, Los Angeles, Austin, Atlanta, Miami, Dallas, Houston, Denver and Nashville. Austin and Atlanta are on the Uber app, the rest on Waymo's own.
• Fleet. About 4,000 vehicles, with around 80% in California and Texas. Semafor reported a plan for 8,000 by the end of 2026 and 20,000 by the end of 2027. Waymo has not confirmed those numbers.
• Miles. 271.3 million rider-only miles through June 2026.
• Money. A $16 billion round at a $110 billion valuation in February, and a $5 billion loan this month to fund the fleet and international growth.

Revenue is the number Waymo does not publish. Sacra estimates an annualised run rate of about $355 million in February 2026. Another back-of-envelope calculation of 500,000 rides a week at an average Bay Area fare of $19.69 gets to roughly $510 million a year. Either way, it is a business with a very large valuation and a fairly small revenue line. The bet is on what the fleet looks like in three years, not what it earns today.

Do the sums on a single car and it is interesting. 500,000 rides a week across about 4,000 cars is roughly 125 rides a car a week, or around 18 a day. At a $19.69 fare that is somewhere around $90,000 to $130,000 of gross fares per car per year. The numbers are rough, because the fleet has grown during the year, and it is revenue, not profit. But that is the shape of the business: an expensive car that earns a lot if you keep it busy.

On safety, Waymo's own data claims 82% fewer injury-causing crashes and 95% fewer serious-injury crashes than human drivers over the same distance in the same cities. That is Waymo's own comparison, but an independent IIHS study found a 68% lower crash involvement rate and 81% fewer injury crashes per mile. Over 270 million miles is not a small sample.

The weak spot is cost. Morgan Stanley estimates the old Jaguar I-PACE robotaxi costs roughly $200,000, and the new Zeekr-built "Ojai" about $125,000. Ojai's base vehicle is about $38,000 before heavy tariffs. Waymo is getting cheaper fast. It is still an expensive way to build a taxi.

Tesla Robotaxi today: smaller, newer, and moving quickly#

Tesla's numbers are a different order of magnitude, and the direction of travel is steep.

• Cities. Driverless rides in six cities: Austin, Dallas, Houston, Miami, Orlando and Tampa. Las Vegas, Jacksonville, Phoenix and San Antonio are listed as "coming soon", with no dates.
• Miles. Tesla's first Robotaxi Safety Report covers about 3.1 million unsupervised miles between 22 January and 18 October. Tesla's own Q3 deck says cumulative unsupervised miles grew eightfold quarter on quarter to nearly two million at the end of the quarter, with a live counter now past four million. That is about 1% of Waymo's total.
• Fleet. The Texas DMV shows 362 Model Y Robotaxis and 449 Cybercabs authorised for driverless operation in Texas. Authorised is not the same as on the road at any given moment, and Tesla does not publish a fleet number.
• Paid rides. Tesla does not publish them. A chart of cumulative paid miles in the Q2 letter dropped 36% from Q1, and the Q3 letter changed it to unsupervised miles, which count whether or not anyone is paying. That is the kind of thing that makes me want the real number.
• Revenue. Musk said on the Q3 call that the driverless fleet will not move the needle this quarter, should become noticeable in Q2 2027 and be "a big deal by Q3 or Q4 of next year".

Cybercab: the car with no steering wheel#

Cybercab is the part that makes Tesla a different kind of competitor. It is a two-seater with no steering wheel or pedals, built at Giga Texas, where installed capacity is listed above 125,000 a year. It started carrying public riders in Austin on 3 September, at a muted, invite-only event with 45 cars authorised that day. A month later, Musk said the Austin Cybercab fleet had grown about tenfold, and Tesla's AI lead said Dallas is next month.

The pricing is early and rough. One MotorTrend writer's five Austin rides averaged $2.93 a mile, which is a tiny sample. The Cybercab also comes with Grok built in, which Tesla's Q3 letter calls "a second living room".

Tesla's cost story is where it gets interesting:

• Tesla's target is about $0.30 per paid mile. On the Q3 call, CFO Vaibhav Taneja was asked about it directly: "Are we at our goal of $0.30 today? No. But do we see a path to get to that? Very - yes, we do."
• Musk has said it is "probably true" that Cybercab could get below $0.20 a mile fully loaded by 2030, against about $0.40 for Waymo's sixth-generation car in ARK Invest's model.
• For comparison, analysts put today's cost at above $0.60 a mile for Tesla's Model Y robotaxi and above $1.00 for Waymo, while one analyst's April estimate was $0.81 for Tesla and $1.43 for Waymo. Those are models, and nobody outside the companies has invoices.
• Goldman reckons that if Tesla gets Cybercab to a $20,000 to $30,000 unit cost it would have a $0.05 to $0.30 per mile advantage over rivals whose cars cost $50,000 to $100,000.

The honest read is that Tesla's advantage is the factory, not the software. Tesla can build the car cheaply and in volume. Waymo can drive it better, today.

The safety question#

This is where I want to be fair, because both sides tell the story they like.

Tesla's safety report says zero injuries, zero Robotaxi airbag deployments and zero pedestrians or cyclists hit across those 3.1 million miles. It compares that with Texas police data, which suggests humans would have caused about 9.1 injury crashes over the same distance.

The caveats are real. The report only counts miles with no safety monitor on board and leaves out the earlier, monitored miles. It does not publish a total collision count. And earlier NHTSA data showed 14 crashes in roughly 800,000 Austin miles through February, about one every 57,000 miles, which Fortune put at several times Tesla's own human benchmark. NHTSA has also opened an audit query into whether Tesla's self-certification of Cybercab as meeting all federal safety standards was justified, and a review of Robotaxis in school zones.

Tesla's latest numbers look good and are far too small to settle anything. 3.1 million miles is about one-hundredth of Waymo's record. Ask me again at 100 million.

What Tesla says is next in the US#

• Dallas Cybercab launch in November, then more cities by the end of 2026.
• 24/7 service by the end of 2026. Austin currently runs 6am to midnight.
• Selling Cybercabs to fleet operators by the end of the year, though Musk says the "vast majority" will be Tesla-owned.
• FSD (Unsupervised) on some customer cars around Q2 2027, which is the 2016 "your car earns money while you sleep" idea finally getting a date.
• Wireless charging. An Austin Robotaxi hub with a planned 80 wireless charging spots, and FCC approval for Cybercab wireless charging.

The pit stop: charging, cleaning, and where Optimus fits#

A robotaxi only earns money when it is carrying someone. Every minute it sits waiting for a plug or a person with a vacuum is a minute it is losing money. Waymo currently uses human fleet teams for this. Tesla's plan is to use robots, and it is the part of the story I find the most fun.

Charging: Optimus as a stopgap#

Cybercab has a hidden NACS charge port in its rear bumper and, for now, cannot charge from AC. It can drive itself to a Supercharger, but today a human has to plug it in. There is footage of a Tesla employee at an Austin Supercharger plugging in Cybercabs and Model Y Robotaxis as they arrive, and even moving cars that parked badly.

In a Munro Live engineering interview on 26 October, Tesla's VP of vehicle engineering Lars Moravy said Optimus is meant to plug the cars in until a more automated system is ready. The quote was "Hopefully, Optimus will help us with that shortly." That is a stopgap, with no timeline, and Moravy is clear that the real fix is wireless charging.

When asked if Optimus could do more, the answer was that it "might service the car, might add in the delivery space".

Cleaning: a robot arm that is already there#

The cleaning side is further along, and it is not Optimus. A Texas filing for Tesla's Austin Robotaxi hub lists a "Cleaning Robot" alongside Supercharger cabinets and an equipment inspection system. Tesla first showed it in January 2025 with the caption "This robot sucks", which is a good caption.

It is a large robotic arm with swappable tools. One vacuums the cabin, one picks up trash and one wipes surfaces, including a microfiber tool for the 21-inch centre screen. It can grab a bottle, a can or a forgotten backpack and sort it into a lost-and-found bin. Tesla has also talked about automatic cleaning fees for riders who leave too much of a mess, drive-through robotaxi car washes and a patent for UV sanitisation between rides.

So what is Optimus actually for?#

My honest take is that Optimus is the general-purpose glue around a system that is otherwise mostly fixed machinery. A fixed arm cleans the cabin better and cheaper than a humanoid could. Wireless pads remove the plug entirely. What a humanoid adds is flexibility: plugging in at a regular Supercharger that is not at the hub, wiping a camera, fixing the odd spill the arm cannot reach, or handling a car that has parked badly. It is a bet that you would rather buy one flexible robot than design a machine for every edge case.

The reality check is that Optimus has only just started early production in Fremont. Musk says external deliveries come around the end of 2027, and that the robot's hands, durability and AI still need work. The latest version has not been shown publicly. 20,000 a week is a target for around the end of 2027 or later. So do not expect a humanoid at your local Cybercab depot in 2027. A European version of this picture is even further out.

Europe: the prize with a locked door#

Now the part I actually wanted to write: what could this look like in Europe?

Where Tesla is in Europe today#

Tesla's Q3 letter says FSD (Supervised) is now available in nine EU countries and has been used by more than 100,000 customers for over 160 million kilometres. It started when the Dutch regulator RDW granted a provisional approval in April. The Q3 letter adds Croatia, Slovenia, the Czech Republic and Slovakia since the end of Q2. Musk said on the call he hopes for an EU-wide approval within a few months and would be surprised if it took beyond Q1.

That is a real milestone, and it is also not a robotaxi. FSD (Supervised) is Level 2 driver assistance. The driver is still the driver and still liable. A robotaxi has nobody in the car, and that is a separate legal category entirely.

What a European robotaxi needs#

Here is the stack of things that have to happen, as I understand it:

• Type approval, not self-certification. In the US, Tesla self-certified that Cybercab meets federal standards, and NHTSA is checking after the fact. The EU does not allow that. A technical service tests the vehicle and an approval authority signs it off before it can go on a public road.
• A cap of 1,500. The route that exists today for fully driverless vehicles is the small-series approval, which allows 1,500 vehicles per type per year across the whole EU. That is not a ban, but it is a very tight ceiling for a company that talks about millions of cars. It is also, I suspect, part of why Waymo chose a single city, Munich, for its first EU launch.
• A new global rulebook is arriving. In June, UNECE adopted the first international regulation for fully driverless systems. The European Commission plans to fold it into EU law in Q4 2026, but practical implementation may not come until January 2027. A vehicle rule is also not an operating permit.
• National rules on top. The EU regulates the vehicle, not the driving. Each member state decides how and where it can run. Germany wants an approved operating area, a technical supervisor and a car that can reach a safe stop on its own. Spain runs a testing framework through its traffic agency, the DGT.
• Local politics. Europe's taxi market is regulated and heavily unionised. Lyft bought FREENOW partly because of its relationships with taxi unions and regulators. Whoever wants to replace drivers needs those people on side.
• The UK is its own thing. It is not in the EU, but London matters. The UK's driverless permit scheme opened on 15 May, and as of late September nobody had applied for a driverless permit. London's first commercial robotaxi service launched in September, but with a safety driver on board.

Where Waymo is in Europe#

Waymo is already queuing up. Its plans: London (targeted for Q4 2026, but the rollout is reported to be delayed), Munich with public rides toward the end of 2027, and Madrid in 2027. In Munich, mapping and supervised testing started in the autumn of 2026, with a public launch targeted about 15 months later.

As far as I can find, Tesla has not announced a European robotaxi city. If it follows anything like Waymo's phased process, a Tesla European launch is unlikely before the back half of 2028. That is my guess, not something Tesla has said.

Tesla's cards in Europe#

Tesla is not starting from nothing. It has Giga Berlin, which set a production record in Q3. It says the Model Y was the best-selling EV in Europe last quarter. It also has a large and growing fleet of European cars collecting supervised driving data. If unsupervised FSD ever gets approved, those customer cars could in principle join a network, which is a different and bigger revenue shape from a company-owned fleet. That last bit is speculative, and Tesla is clear that Cybercab will be mostly Tesla-owned for now.

What Robotaxi and Cybercab could earn in Europe#

Here is the back-of-the-envelope part. Treat all of it as my own illustration, not a forecast, and not anything Tesla has said.

The market#

Estimates of the European taxi and ride-hailing market vary a lot by definition. One industry figure puts it at about €85 billion today, growing to more than €140 billion by 2032, with over half still traditional taxis. That figure comes from a ride-hailing software company, not a neutral source. A narrower consulting estimate in Lyft's FREENOW deal materials puts the European market at €41 billion in 2024. I use the larger numbers below, but if the real answer is closer to the smaller one, scale everything down by half.

The assumption that matters most#

How much does one Cybercab earn in a year? My base case is €80,000 of fares per cab per year. That comes from about 80,000 paid kilometres at about €1.00 a kilometre.

For context, Waymo's implied US figure is roughly $90,000 to $130,000 a car, and my number is deliberately a bit lower to allow for lower European fares, a ramp-up period and cars that sit idle. If fares have to fall to €0.60 a kilometre to win customers, that drops to about €48,000 a cab, and you need over 60% more cabs for the same revenue. Fares and utilisation are the two numbers to argue about.

The scenarios#

All figures are gross fares per year at €80,000 a cab, which is Tesla revenue if Tesla owns the fleet.

• The cap bites: 1,500 cabs, about €120 million a year. One year of the small-series allowance for a single vehicle type. After three years of the cap that is about 4,500 cabs and €360 million a year, which is roughly where Waymo's estimated revenue run rate is today.
• A few flagship cities: 10,000 cabs, about €800 million a year. Around 1% of the €85 billion market. London, Paris, Berlin, Madrid and Milan are plausible places to start. It needs rules to change, or more than one approved vehicle type.
• Real traction: 50,000 cabs, about €4 billion a year. About 5% of today's market, or 3% of the 2032 number. That is roughly 4% of Tesla's current annualised revenue, based on Q3's $28.0 billion.
• The taxi market gets disrupted: 175,000 cabs, about €14 billion a year. That is 10% of the 2032 market and roughly 14% of Tesla's current annualised revenue. It is also only about 1.4 years of the Texas factory's listed Cybercab capacity.

That last point is the one worth noticing. Tesla's manufacturing is not the constraint on a European robotaxi business. Law, politics and trust are.

Revenue is not profit#

Gross fares are the easy part. The more interesting question is what is left after running costs. The numbers below use the three per-mile costs I mentioned earlier, converted to kilometres at a rough $1 = €0.90, applied to 80,000 paid kilometres. Definitions of "cost per mile" differ between sources, so treat this as a guide to the shape, not a measurement.

• Tesla hits its $0.30 per mile target: about €0.17 a kilometre, around €13,000 a cab a year, leaving roughly €67,000. That would be a margin above 80%.
• Tesla stays at around $0.81 per mile, the sort of number analysts quote for Tesla today: about €0.45 a kilometre, around €36,000 a cab, leaving roughly €44,000. A margin of around 55%.
• Costs look like today's Waymo, at about $1.43 per mile: about €0.80 a kilometre, around €64,000 a cab, leaving roughly €16,000. A margin of around 20%.

That is why the cost per mile matters more than the fleet size. The same fleet could be a very good business or a thin one, depending on a number that Tesla itself says it has not hit yet. And Europe adds its own costs: higher labour for the people who do remain, insurance, depot space, local licensing, and VAT on fares. I would expect European margins to land below these US-based figures, at least at first.

Who wins, and what I would watch#

On 1 November 2026, Waymo is ahead. It has more cars, more cities, more miles, a longer safety record, real revenue and a European plan. Tesla has a much cheaper car, a factory with room to build 125,000 a year, and no shortage of ambition. That is a real asymmetry, and I do not think either side can be written off.

Here is what I would watch:

1. Does Dallas work? Tesla says Cybercab will scale with the same software. Dallas in November is the first real test outside Austin.
2. Does Tesla publish paid rides? Unsupervised miles tell me how far the cars drive. Paid rides and revenue tell me whether there is a business.
3. Does the safety record hold at scale? Zero injuries at 3 million miles is a good start. The test is tens of millions of miles, with a regulator watching.
4. Does NHTSA's Cybercab audit end quietly? A bad outcome could slow everything, in the US and in Europe.
5. Does the EU Commission adopt the UNECE driverless rules in Q4? That is the first real step toward a legal path for something like Cybercab in Europe.
6. Does Tesla name a European robotaxi city? It has not yet. Waymo named Munich more than a year before launch.
7. Does Optimus show up with a plug? A video is promised in a month or two, and the useful test is whether it is doing a real task rather than a rehearsed demo.

I think the most likely version of Europe is slower than the keynotes and faster than the sceptics. The cap and the national rules mean it starts small, a handful of cities and hundreds or low thousands of cars. If the safety data holds up and the rules catch up, the maths above says the upside is large. The first European Cybercab ride is the thing I want to see, and I would rather see it on my own street than in a slide deck.

References#

• Tesla, Robotaxi Safety Report (live counter, cities, methodology)
• Tesla Q3 2026 quarterly update deck
• CNBC, Tesla Q3 2026 earnings (Texas fleet authorisations, Musk on Robotaxi revenue timing)
• The Motley Fool, Tesla Q3 2026 earnings call transcript (cost target, Optimus timing, EU approval comments)
• TechCrunch, Tesla is prepping a Cybercab onslaught
• EVwire, Cybercab heads to Dallas next month
• The Verge, The unusually muted Tesla Cybercab launch
• MotorTrend, Tesla Cybercab first ride
• Fortune, Tesla robotaxi crash rate from NHTSA data
• NHTSA, audit query into Tesla Cybercab self-certification
• Teslarati, Musk on Cybercab cost per mile
• ARK Invest, Newsletter 496 (Robotaxi cost comparison)
• Autonomix Investor, robotaxi cost per mile 2026: fare vs operator cost
• Dr. Web, why Cybercab needs no exemption in the US (German rules summary)
• Electric Vehicles, Cybercab and Waymo cost per mile estimates
• Investing.com, Goldman on Cybercab cost advantage
• Not a Tesla App, Optimus to plug in Robotaxis
• Tesla North, Moravy on Optimus and Cybercab charging
• Not a Tesla App, Robotaxi hubs start installing cleaning robots
• Teslarati, Tesla's Cybercab and Robotaxi cleaning service
• TechCrunch, Tesla starts Optimus production in California
• The Verge, Optimus is in production, but is it ready
• TechCrunch, Waymo is scaling fast: what the fleet data shows
• Reuters via StreetInsider, Waymo plans to quintuple its fleet to 20,000
• Waymo, Safety Impact data hub
• IIHS, Waymo's driverless cars crash less often than people
• Sacra, Waymo revenue, valuation and funding
• Axis Intelligence, Waymo statistics 2026
• The Next Web, Waymo's Ojai robotaxi
• Forbes, Rides in Waymo's Ojai
• Waymo, Servus Munchen: Waymo is coming to Germany
• Reuters via Investing.com, Waymo to launch in Madrid
• Highways Magazine, London robotaxi rollout delayed
• Global Policy Watch, EU and UK autonomous vehicle regulatory developments
• Mobility Global, how Europe is preparing for autonomous vehicles (UNECE driverless regulation)
• EVSHIFT, Cybercab self-certified, Europe caps it at 1,500
• RDW, European type approval for Tesla FSD Supervised, provisional in the Netherlands
• Times of Malta, European taxi and ride-hailing market size
• Quartr, Lyft M&A presentation on FREENOW and the European mobility market