Q&A: Tesla Cybercab and Robotaxi: The Complete Story Explained

Why is Tesla building a two-seat robotaxi instead of a larger vehicle?
Ride-sharing data shows the vast majority of trips carry one or two passengers, so a two-seater covers the dominant use case while keeping cost per mile low. Peer-reviewed studies of Uber and Lyft trip data put the average ride at 1.3-1.4 passengers, and Tesla's head of design estimated 90-95% of ride-share trips carry one or two people.
The National Household Travel Survey from the Federal Highway Administration backs this up on the private-vehicle side, with the average personal vehicle trip in the US at around 1.6 occupants.
For anything bigger, Tesla plans to lean on the rest of the fleet, including the Model Y and the upcoming larger Model Y L. Trying to make one vehicle handle every possible ride results in a Model Y, which wasn't engineered with robotaxi economics fully in mind.
Why is self-driving technology finally working now?
Modern self-driving works because neural networks replaced rules-based systems, and compute and data finally caught up. The same AI revolution that enabled chatbots and image generators is what made real self-driving cars viable.
Early self-driving systems were rules-based, requiring engineers to hand-code logic for every possible situation, and they broke down the moment a car encountered something unanticipated, like a bag blowing across the road with a kid chasing it.
A neural-network approach learns to drive the way a human does, by being shown millions of examples and figuring out patterns on its own. Without modern GPUs and neural networks, none of the current robotaxi work would happen.
What is the CyberCab's design and specifications?
The CyberCab is a two-seat coupe roughly 14 feet long with butterfly doors that open automatically, no steering wheel or pedals, and a single 20.5-inch center display as the only control interface.
Under the body sits a 35 kWh battery pack, roughly the size of a base Chevy Bolt's, targeting about 200 miles of range at 5.5 miles per kilowatt-hour. Charging is inductive only, with no plug, using wireless charging at above 90% efficiency.
Body panels are molded polyurethane with the color baked into the plastic, which lets Tesla strip the paint shop out of the factory entirely. The car was engineered from day one to run on a robotaxi network 24/7 without a human to wipe a lens, with almost the entire frunk serving as a fluid reservoir for the camera washers.
What is the state of Tesla's commercial robotaxi service?
Tesla launched its robotaxi service in Austin on June 22, 2025 with modified Model Ys and began offering unsupervised paid rides on January 22nd, with no safety monitor in the car.
The unsupervised fleet is around 13 Model Ys and growing, with the full Austin robotaxi fleet at 37-44 vehicles depending on the day. The Austin geofence expanded from about 20 square miles at launch to over 245.
Tesla expanded to Dallas and Houston with some unsupervised rides and confirmed plans for Phoenix, Miami, Orlando, Tampa, and Las Vegas in the first half of 2026. San Francisco Bay-area rides still require a safety driver behind the wheel.
How does Tesla's camera-only approach compare to competitors' sensor stacks?
Tesla uses cameras only, processed by a neural network that predicts depth for every pixel, while competitors like Waymo combine cameras, radar, and LiDAR.
The rationale traces back to Andrej Karpathy's 2021 argument that humans drive with vision, so a neural net can process visual input to understand depth and velocity. Cameras are cheap to manufacture, while a single LiDAR unit can cost more than an entire base model EV.
The bigger strategic advantage is scaling: every Tesla rolling off the line already has the camera suite, so each becomes a potential robotaxi with a software update. Safety researchers still believe sensor redundancy is required since LiDAR gives ground-truth depth while a camera-based system has to infer it from pixels.
Who are Tesla's competitors in the robotaxi market?
Waymo is the undisputed leader by paid trips, delivering about 500,000 paid robotaxi rides weekly across 11 US cities, while Zoox is taking a purpose-built approach closer to Tesla's.
Waymo's vehicles are retrofitted Jaguar I-Paces loaded with LiDAR, radar, and camera stacks, with industry estimates putting each vehicle well north of $150,000. Zoox has operated commercially in Las Vegas since September 2025 and has carried over 350,000 passengers.
Several well-funded attempts wound down: GM shut down Cruise in December 2024 after losing more than $10 billion, Ford's Argo AI shut down in October 2022, and Apple cancelled Project Titan in early 2024 after burning a reported $10 billion.
What is the cost-per-mile economics of the CyberCab?
Tesla's stated target for the CyberCab at scale is 20 cents per mile, compared to about 77 cents per mile for owning a typical new vehicle and $1-2 for a ride-share trip.
The drop from 77 to 20 cents per mile is roughly three-quarters, the kind of number that could make living car-free economically rational for large parts of the population, not just dense urban centers.
Is FSD actually safer than human driving?
Tesla reports one major crash every 5.1 million miles driven on FSD supervised in North America, versus one every 2.3 million miles for manually driven Teslas and one every 660,000 miles for the US national average.
The comparison to the national average isn't apples-to-apples, since Tesla drivers skew toward newer vehicles, specific demographics, and certain road types that already correlate with lower accident rates. Safety researcher Phil Koopman has noted that Tesla controls the boundaries of what counts as a crash and how the figures are calculated.
The gap between FSD mileage and manual Tesla mileage is probably the most honest signal in the data, because both populations are Tesla drivers. If FSD is reducing serious accidents even by a fraction of what Tesla claims, the safety case for the technology is real.
Why did Tesla choose the robotaxi over the Model 2?
In the summer of 2022, Elon Musk decided Tesla would go all-in on a purpose-built robotaxi rather than a low-cost passenger EV, rejecting a steering-wheel backup option.
Franz von Holzhausen and others in the meeting tried to convince Musk to build a version with a steering wheel as a backup while regulators approved fully autonomous vehicles, but Musk rejected it, saying the vehicle must be designed as a clean robotaxi.
The logic behind the bet is Musk's first-principles approach: what ultimately matters in moving people is cost per mile, and the cheapest way to drop it is a vehicle purpose-built for the job with no wasted material or complexity.
Does AI4 have enough compute for unsupervised driving?
This remains unresolved. Tesla previously said the CyberCab would launch with AI5, but on the Q1 2026 earnings call Elon confirmed AI5 has slipped to mid-2027, so production CyberCabs will ship on AI4, and whether AI4 has enough compute for unsupervised driving at Tesla's target safety level is something to be watched closely over the next year.

Why is Tesla building a two-seat robotaxi instead of a larger vehicle?
Ride-sharing data shows the vast majority of trips carry one or two passengers, so a two-seater covers the dominant use case while keeping cost per mile low. Peer-reviewed studies of Uber and Lyft trip data put the average ride at 1.3-1.4 passengers, and Tesla's head of design estimated 90-95% of ride-share trips carry one or two people.
The National Household Travel Survey from the Federal Highway Administration backs this up on the private-vehicle side, with the average personal vehicle trip in the US at around 1.6 occupants.
For anything bigger, Tesla plans to lean on the rest of the fleet, including the Model Y and the upcoming larger Model Y L. Trying to make one vehicle handle every possible ride results in a Model Y, which wasn't engineered with robotaxi economics fully in mind.
Why is self-driving technology finally working now?
Modern self-driving works because neural networks replaced rules-based systems, and compute and data finally caught up. The same AI revolution that enabled chatbots and image generators is what made real self-driving cars viable.
Early self-driving systems were rules-based, requiring engineers to hand-code logic for every possible situation, and they broke down the moment a car encountered something unanticipated, like a bag blowing across the road with a kid chasing it.
A neural-network approach learns to drive the way a human does, by being shown millions of examples and figuring out patterns on its own. Without modern GPUs and neural networks, none of the current robotaxi work would happen.
What is the CyberCab's design and specifications?
The CyberCab is a two-seat coupe roughly 14 feet long with butterfly doors that open automatically, no steering wheel or pedals, and a single 20.5-inch center display as the only control interface.
Under the body sits a 35 kWh battery pack, roughly the size of a base Chevy Bolt's, targeting about 200 miles of range at 5.5 miles per kilowatt-hour. Charging is inductive only, with no plug, using wireless charging at above 90% efficiency.
Body panels are molded polyurethane with the color baked into the plastic, which lets Tesla strip the paint shop out of the factory entirely. The car was engineered from day one to run on a robotaxi network 24/7 without a human to wipe a lens, with almost the entire frunk serving as a fluid reservoir for the camera washers.
What is the state of Tesla's commercial robotaxi service?
Tesla launched its robotaxi service in Austin on June 22, 2025 with modified Model Ys and began offering unsupervised paid rides on January 22nd, with no safety monitor in the car.
The unsupervised fleet is around 13 Model Ys and growing, with the full Austin robotaxi fleet at 37-44 vehicles depending on the day. The Austin geofence expanded from about 20 square miles at launch to over 245.
Tesla expanded to Dallas and Houston with some unsupervised rides and confirmed plans for Phoenix, Miami, Orlando, Tampa, and Las Vegas in the first half of 2026. San Francisco Bay-area rides still require a safety driver behind the wheel.
How does Tesla's camera-only approach compare to competitors' sensor stacks?
Tesla uses cameras only, processed by a neural network that predicts depth for every pixel, while competitors like Waymo combine cameras, radar, and LiDAR.
The rationale traces back to Andrej Karpathy's 2021 argument that humans drive with vision, so a neural net can process visual input to understand depth and velocity. Cameras are cheap to manufacture, while a single LiDAR unit can cost more than an entire base model EV.
The bigger strategic advantage is scaling: every Tesla rolling off the line already has the camera suite, so each becomes a potential robotaxi with a software update. Safety researchers still believe sensor redundancy is required since LiDAR gives ground-truth depth while a camera-based system has to infer it from pixels.
Who are Tesla's competitors in the robotaxi market?
Waymo is the undisputed leader by paid trips, delivering about 500,000 paid robotaxi rides weekly across 11 US cities, while Zoox is taking a purpose-built approach closer to Tesla's.
Waymo's vehicles are retrofitted Jaguar I-Paces loaded with LiDAR, radar, and camera stacks, with industry estimates putting each vehicle well north of $150,000. Zoox has operated commercially in Las Vegas since September 2025 and has carried over 350,000 passengers.
Several well-funded attempts wound down: GM shut down Cruise in December 2024 after losing more than $10 billion, Ford's Argo AI shut down in October 2022, and Apple cancelled Project Titan in early 2024 after burning a reported $10 billion.
What is the cost-per-mile economics of the CyberCab?
Tesla's stated target for the CyberCab at scale is 20 cents per mile, compared to about 77 cents per mile for owning a typical new vehicle and $1-2 for a ride-share trip.
The drop from 77 to 20 cents per mile is roughly three-quarters, the kind of number that could make living car-free economically rational for large parts of the population, not just dense urban centers.
Is FSD actually safer than human driving?
Tesla reports one major crash every 5.1 million miles driven on FSD supervised in North America, versus one every 2.3 million miles for manually driven Teslas and one every 660,000 miles for the US national average.
The comparison to the national average isn't apples-to-apples, since Tesla drivers skew toward newer vehicles, specific demographics, and certain road types that already correlate with lower accident rates. Safety researcher Phil Koopman has noted that Tesla controls the boundaries of what counts as a crash and how the figures are calculated.
The gap between FSD mileage and manual Tesla mileage is probably the most honest signal in the data, because both populations are Tesla drivers. If FSD is reducing serious accidents even by a fraction of what Tesla claims, the safety case for the technology is real.
Why did Tesla choose the robotaxi over the Model 2?
In the summer of 2022, Elon Musk decided Tesla would go all-in on a purpose-built robotaxi rather than a low-cost passenger EV, rejecting a steering-wheel backup option.
Franz von Holzhausen and others in the meeting tried to convince Musk to build a version with a steering wheel as a backup while regulators approved fully autonomous vehicles, but Musk rejected it, saying the vehicle must be designed as a clean robotaxi.
The logic behind the bet is Musk's first-principles approach: what ultimately matters in moving people is cost per mile, and the cheapest way to drop it is a vehicle purpose-built for the job with no wasted material or complexity.
Does AI4 have enough compute for unsupervised driving?
This remains unresolved. Tesla previously said the CyberCab would launch with AI5, but on the Q1 2026 earnings call Elon confirmed AI5 has slipped to mid-2027, so production CyberCabs will ship on AI4, and whether AI4 has enough compute for unsupervised driving at Tesla's target safety level is something to be watched closely over the next year.

Q&A: Tesla Cybercab and Robotaxi: The Complete Story Explained

Why is Tesla building a two-seat robotaxi instead of a larger vehicle?

Why is self-driving technology finally working now?

What is the CyberCab's design and specifications?

What is the state of Tesla's commercial robotaxi service?

How does Tesla's camera-only approach compare to competitors' sensor stacks?

Who are Tesla's competitors in the robotaxi market?

What is the cost-per-mile economics of the CyberCab?

Is FSD actually safer than human driving?

Why did Tesla choose the robotaxi over the Model 2?

Does AI4 have enough compute for unsupervised driving?

Cybercab Cost Per Mile: Tesla's Lifetime Economics Explained

Will Uber Survive Tesla Robotaxi? Tesla, SpaceX, Terrafab

Tesla Cybercab Launch: Robotaxi Growth and Starlink Plans

Tesla Cyber Cab Launch: What It Means for TSLA Stock

Q&A: Tesla Cybercab and Robotaxi: The Complete Story Explained

Why is Tesla building a two-seat robotaxi instead of a larger vehicle?

Why is self-driving technology finally working now?

What is the CyberCab's design and specifications?

What is the state of Tesla's commercial robotaxi service?

How does Tesla's camera-only approach compare to competitors' sensor stacks?

Who are Tesla's competitors in the robotaxi market?

What is the cost-per-mile economics of the CyberCab?

Is FSD actually safer than human driving?

Why did Tesla choose the robotaxi over the Model 2?

Does AI4 have enough compute for unsupervised driving?

Cybercab Cost Per Mile: Tesla's Lifetime Economics Explained

Will Uber Survive Tesla Robotaxi? Tesla, SpaceX, Terrafab

Tesla Cybercab Launch: Robotaxi Growth and Starlink Plans

Tesla Cyber Cab Launch: What It Means for TSLA Stock