Skip to content
AI-Daily-Builder

2026-06-18 views

Physical AI Unit Economics — Waymo Revenue Per Mile vs Tesla Cybercab Projected Payback: The Robotaxi Profitability Race

Waymo earns est. $2.50–$5.00/mile but stays unprofitable. Cybercab projects ~3-month payback if the $30K cost target and minimal remote-ops claim both hold.

Overview

Unit economics — revenue per mile versus cost per mile — is the financial foundation that separates a real business from an expensive technology demonstration. In the robotaxi sector, two companies have disclosed enough information to benchmark: Waymo, which already operates a commercial service, and Tesla, which has published projections for its Cybercab. This article benchmarks their economics, stress-tests the key assumptions, and identifies the decisive variables that will determine whether Physical AI becomes the most profitable transportation business in history — or an expensive detour.

This is article 158 in the Physical AI Benchmark Series.


Section 1 — Waymo Current Unit Economics (Mid-2026 Estimates)

Waymo operates commercial robotaxi service in San Francisco, Phoenix, Los Angeles, and Austin. Alphabet has confirmed Waymo is pre-revenue-breakeven and describes it as a long-term investment. Waymo does not publish per-ride economics. The following table is constructed from public pricing signals, analyst estimates, and comparable rideshare data. All figures are estimates (est.) unless otherwise noted.

Economic Line ItemEstimateBasis / Notes
Average fare per ride$15–$25 (est.)Comparable to Uber/Lyft pricing in Phoenix and SF; Waymo has not disclosed average fare
Average ride distance4–8 miles (est.)Urban robotaxi trips tend to be shorter than rideshare average; Waymo SF/Phoenix typical trip est.
Revenue per mile$2.50–$5.00 (est.)Derived from fare/distance estimate above
Vehicle cost (Gen 6, fully equipped, est.)$30,000–$50,000+ (est.)Zeekr RT base + AV sensor stack + US import + tariff exposure; Waymo has not disclosed
Vehicle depreciation per mile (est.)$0.15–$0.40/mileAssumes 200,000–400,000 mile vehicle lifespan; 100,000 annual miles; est.
Remote operations cost per mile (est.)$0.30–$0.80/mileWaymo employs remote assistance operators who can intervene; ratio of operators to vehicles is key lever; est.
Maintenance per mile (est.)$0.15–$0.30/mileEV maintenance lower than ICE; lidar cleaning/calibration adds cost vs camera-only vehicle; est.
Insurance per mile (est.)$0.20–$0.50/mileCommercial AV insurance is expensive and evolving; Waymo likely self-insures partially; est.
HD mapping per mile (est.)$0.05–$0.15/mileOngoing map refresh required; cost amortized over fleet miles in mapped area; est.
Depot/charging infrastructure per mile (est.)$0.10–$0.25/milePer-city depot capex amortized over vehicle fleet and annual miles; est.
Total cost per mile (est.)$0.95–$2.40/mileSum of above operating cost components (est.)
Gross margin per mile at current scale (est.)Negative to breakeven (est.)Revenue est. $2.50–$5.00/mile vs cost est. $0.95–$2.40/mile; at low fleet density, fixed costs dominate; at high fleet density and high utilization, positive gross margin possible (est.)
Current profitability statusNot profitable (confirmed by Alphabet)Alphabet does not break out Waymo financials; Waymo described as pre-revenue-breakeven investment
Key lever to profitabilityRemote operations ratioEach doubling of vehicles-per-operator roughly halves the remote-ops cost line; AI improvement drives this ratio

Key insight: Waymo’s revenue-per-mile range ($2.50–$5.00 est.) already exceeds its estimated operating cost-per-mile floor ($0.95 est.) in optimistic scenarios. The gap is the question of scale: at current fleet size and utilization, fixed infrastructure and remote-ops costs prevent profitability. The business becomes attractive if — and only if — the remote operations cost per mile falls materially as the fleet scales and AI matures.


Section 2 — Tesla Cybercab Projected Unit Economics (Scenarios)

Tesla has not published official unit economics for the Cybercab. The following scenarios are constructed from Tesla’s stated cost targets (below $30,000 vehicle cost, published at investor events), industry benchmarks for robotaxi utilization, and analyst estimates. All figures labeled (est.).

ScenarioVehicle CostAnnual MilesRevenue/MileCost/Mile (est.)Annual Gross Profit/VehiclePayback Period
Conservative (2027, limited scale)$35,000 (est.)60,000 miles/yr$2.00/mile$1.50/mile (est.)$30,000/yr (est.)~1.2 years (est.)
Base case (2028, volume ramp)$28,000 (est.)80,000 miles/yr$2.50/mile$1.20/mile (est.)$104,000/yr (est.)~3.5 months (est.)
Optimistic (2029+, scale)$25,000 (est.)100,000 miles/yr$3.00/mile$1.00/mile (est.)$200,000/yr (est.)~6 weeks (est.)
Pessimistic (delays, competition)$40,000 (est.)40,000 miles/yr$1.50/mile$2.00/mile (est.)Negative (est.)Not profitable in this scenario

Notes on all scenarios:


Section 3 — Key Unit Economics Drivers: Waymo vs Tesla Cybercab

DriverWaymoTesla CybercabWho Benefits
Vehicle cost$30,000–$50,000+ (est.); Zeekr RT + AV stack + tariffBelow $30,000 target (Tesla stated)Tesla (if target achieved)
Vehicle lifespanPurpose-built for AV duty cycle; est. 200,000–500,000 milesPurpose-built; similar duty cycle target; est.Even
Annual utilizationCurrent: est. 50,000–80,000 miles/vehicle/year in commercial serviceProjected: est. 80,000–100,000+ miles/year if driverless permits achievedTesla (higher if driverless permits come through)
Remote operations ratioCurrently requires significant remote ops staff per vehicle; key cost leverTesla claims FSD will require minimal human oversight; unproven at driverless scaleTesla (if FSD self-sufficiency claim holds)
Insurance costHigher per vehicle: established AV insurer relationships; safety data advantageHigher per vehicle initially; Tesla’s safety record could lower premiums over timeWaymo (established insurer relationships)
Maintenance: lidar vs cameraLidar requires cleaning, calibration, periodic replacement; adds cost vs camera-onlyCamera only: simpler, cheaper maintenance; no lidar cleaningTesla
Scale economicsFleet of 2,500–10,000 (est. 2026-2028); never reaches consumer-product manufacturing efficiencyIf Cybercab reaches 100,000+/year production: consumer-EV manufacturing efficiencyTesla (at scale)
Geographic densityConcentrated in 4 cities: high ride density per city; efficient utilization in mapped areasWould deploy nationally if permits achieved; risk of low utilization in new markets initiallyWaymo (current density advantage)

Section 4 — Competitive Robotaxi Pricing Dynamics

DimensionWaymo PricingTesla Cybercab Pricing (est.)Uber/Lyft ComparisonImplication
Current pricing$15–$25 per ride (est.); similar to Uber/Lyft; not discountingN/A (not yet in driverless commercial service)Similar pricing; 25–30% platform take rateWaymo matching market rate; not yet competing on price
Long-run cost potentialIf remote ops costs collapse: cost per mile could fall to $0.50–$0.80/mile (est.)Below $30K vehicle + minimal remote ops: could reach $0.50–$1.00/mile (est.)Human drivers: structural cost floor ~$0.80–$1.20/mile for driver shareBoth AV operators have structural long-run cost advantage over human-driver rideshare
Price war scenarioWaymo could price below Uber/Lyft to gain share once utilization is highTesla could use Cybercab margins to price aggressively if unit economics holdUber/Lyft lose if AV operators achieve scale and cut prices; Uber has hedged via AV partnershipsLong-run winner takes market share through price
Network flywheelMore rides → more data → better routing → lower remote ops ratio → lower cost → more ridesMore rides → more FSD training data → better autonomy → lower oversight cost → more ridesBoth flywheels spin; Waymo’s is real today; Tesla’s is projected
Dynamic pricing leverWaymo can implement surge pricing; high utilization periods more profitableSame capabilityUber/Lyft use surge pricing heavilyDynamic pricing is a unit economics lever neither has fully disclosed plans for

Section 5 — Unit Economics Benchmark Scorecard

DimensionWaymoTesla CybercabEdge2028 Outlook
Vehicle cost$30,000–$50,000+ (est.)Below $30,000 targetTesla (if target achieved)Tesla decisive if Cybercab hits cost target
Remote operations costSignificant current burden; key risk and opportunityMinimal claimed; unproven at scaleUncertainIf FSD achieves driverless self-sufficiency, Tesla decisive
Current profitabilityNot profitable (Alphabet confirmed)Pre-commercial; no Cybercab revenueEven (both pre-profit)Race to first profitable robotaxi fleet
Revenue per mile todayReal revenue: est. $2.50–$5.00/mile from commercial ridesZero: Cybercab not in driverless commercial serviceWaymo (real revenue today)Tesla first revenue requires driverless permit ramp
Long-run cost potential$0.50–$0.80/mile if remote ops collapse (est.)$0.50–$1.00/mile at scale with minimal oversight (est.)Even at long-run scaleBoth converge to similar floor if AI matures
Payback period at scale (est.)Analysts estimate 5–8 years at current vehicle cost and scale (est.)Base case: ~3.5 months per vehicle at $28K cost + $2.50/mile + 80K miles/year (est.)Tesla decisive (if projections hold)Tesla’s stated unit economics are the most compelling in the industry

Overall verdict: Waymo generates real revenue today and has the most advanced commercial robotaxi economics in operation — but the business is not yet profitable, and Alphabet funds it as a long-term investment. Tesla’s Cybercab unit economics at stated cost and scale are significantly more attractive — but they are projections, not demonstrated outcomes. The decisive question: can Tesla achieve a below-$30,000 manufacturing cost, minimal remote-ops overhead, and 80,000 or more annual miles per vehicle, simultaneously? If all three hold, Cybercab’s unit economics would be among the most compelling commercial vehicle businesses in transportation history. That is a very high bar. The probability that all three materialize as projected, on schedule, is the central uncertainty in Tesla’s Physical AI investment thesis.


All estimates in this article are labeled (est.) and are derived from public company disclosures, analyst estimates, and industry benchmarks. Neither Waymo nor Tesla has published official per-mile unit economics. This article is part of the Physical AI Benchmark Series — article 158.


Sources

Tags

Tip