Musk: Cybercab's Cost Per Mile Underestimated, ARK Invest Estimates Operating Cost Could Drop to About $0.20 per Mile
Tesla CEO Elon Musk posted on X, stating, "Cybercab has been optimized to an extreme degree in terms of 'total cost per mile', and few truly understand it," reiterating the operational cost advantages of its autonomous taxi, Cybercab; this statement was made after Cybercab officially began passenger operations in Austin, Texas on September 3.
According to ARK Invest's estimates, by 2030, Cybercab's operating cost per mile is expected to drop to about $0.20, roughly half of Waymo's sixth-generation model (about $0.40 per mile); Morgan Stanley previously estimated Tesla's cost at about $0.81 per mile, and Waymo's at about $1.36 to $1.43 per mile. In comparison, the average cost per mile for human-driven ride-hailing in the U.S. in 2025 is about $2.80, while the global average for robot taxis is about $0.25 per mile (lowered by China's pricing of about $0.50 per mile). Musk had also previously responded in an interview that Cybercab's cost per mile being below $0.20 by 2030 is "very likely true".
Regarding vehicle acquisition costs, ARK Invest estimates Cybercab's production cost at about $18,000, while Morgan Stanley estimates the cost of Waymo's Ojai model at about $125,000; with a capital of $1 million, approximately 55 Cybercabs can be purchased, nearly seven times the fleet size of Waymo's equivalent investment (about 8 vehicles). Musk also mentioned that this cost comparison "doesn't even account for Cybercab's operational efficiency," citing data that Waymo's modified I-Pace model consumes about 104.4 kWh per 100 km, more than six times the estimated consumption of Cybercab.
Cybercab is a two-seat model without a steering wheel or pedals, designed specifically for autonomous driving. Currently, it only provides passenger services in certain areas of Austin, and the official fare schedule has not yet been announced. Company executives previously stated that they would adopt "dynamic pricing", aiming for "coach cabin prices to enjoy first-class experiences". Tesla had also stated that the initial mass production of Cybercab would follow an "S-curve", with early progress being "painfully slow", but ultimately becoming "very fast", with mass production set to start in April 2026.
This cost claim has also faced external skepticism: the National Highway Traffic Safety Administration (NHTSA) has stated that it is "in communication with Tesla and assessing the situation", and California (considered a key market by Tesla) has yet to approve its autonomous passenger services without a safety driver; Reuters testing has also found issues with Cybercab, such as long wait times and some orders being unfulfilled, with instances in Dallas where vehicles failed to reach their destinations, requiring passengers to walk about 15 minutes. As of now, Tesla has approximately 420 registered autonomous vehicles in Texas, fewer than Waymo's 988.
Musk repeatedly emphasizes the "cost per mile" metric, with the core goal of shifting the capital market's valuation logic for Cybercab from "can the vehicle be sold" to "can the fleet operate at a lower cost to generate more rides"; if the cost estimates from ARK and other institutions are accepted by the market, investors would be more willing to assign a valuation premium to Tesla's robot taxi business similar to platform-type businesses with stronger economies of scale, rather than simply valuing it as a vehicle manufacturer. Beneficiaries are Tesla bulls holding related cost model narratives and institutions like ARK betting on its autonomous driving business, while those under pressure include competitors like Waymo, which operate at higher per-vehicle costs and have slower expansion rates, as well as analysts and regulatory bodies who have yet to see large-scale stable operational data from Cybercab and are skeptical of the aforementioned cost predictions.
Source: Public Information
ABAB AI Insight
Musk has repeatedly used metrics like "cost per mile" in Tesla's earnings calls and product launches to set the tone for new businesses that have not yet been widely validated—ranging from Autopilot and FSD subscriptions to now Cybercab. He consistently employs the narrative of "currently expensive, but will become exponentially cheaper in the future" through the S-curve, attempting to get the market to price in non-existent economies of scale ahead of time.
The cost narrative for Cybercab essentially shifts capital from a "vehicle sales" logic to a "fleet operation" logic: Musk emphasizes that with a single vehicle cost of around $18,000, Tesla can achieve a fleet density far exceeding Waymo's and lower unit energy consumption, aiming to convince investors that the same $1 million capital investment can yield nearly seven times the operational capacity compared to Waymo, thereby supporting a larger potential revenue base in the capital-intensive autonomous ride-hailing sector with lower fleet capital expenditures.
This is similar to how Uber and Didi persuaded the capital market to tolerate long-term losses in exchange for market share through a "decreasing marginal cost" narrative; however, this time the core variable has shifted from "subsidies to acquire users" to "hardware and energy cost advantages to scale fleet size". In contrast, Waymo has chosen to steadily expand using modified third-party vehicles with lidar, while Tesla bets on a pure vision solution and dedicated vehicles to lower per-vehicle costs. Currently, Cybercab is still in the early stage of limited passenger operations and has not entered large-scale commercial validation, with cost advantages remaining largely at the level of institutional estimates and Musk's statements.
This is essentially a narrative competition over the "pricing power" in the autonomous driving sector—whoever can convince the market that their unit economic model represents the true cost curve of the industry will be able to secure a capital valuation that matches that narrative before large-scale revenue validation occurs; the mechanism is that the true profitability of autonomous ride-hailing heavily relies on unverified economies of scale and regulatory progress. In the absence of publicly available, auditable fleet operation data, "cost per mile" is more of a public relations phrase that can be repeatedly cited to influence market expectations rather than an independently verifiable financial metric, which is also the root of NHTSA's investigation and analysts' ongoing skepticism.
ABAB News · Law of Cognition
- Cost advantages without scale validation are essentially a form of expected pricing.
- The first half of the S-curve is always easier to narrate than the second half.
- The louder the unit economic metrics are shouted, the more it indicates that economies of scale have not yet materialized.