May Mobility plans to put more than 10,000 autonomous Volkswagen ID. Buzz vehicles on global robotaxi platforms through 203110. Tesla has eliminated its overnight service gap, running robotaxis around the clock in every active city6. And Jeremy Clarkson, writing in The Sunday Times, reports that a Cybercab in Texas ran red lights, braked for nonexistent obstacles, and would not stop at his destination5. The robotaxi sector is scaling capital commitments faster than it is closing the operational-safety questions those commitments depend on—and the gap between the two is widening in ways that matter for regulators, riders, and investors.
That thesis is disputable. The strongest counterargument is that fleet scale *is* the safety mechanism: more miles produce more edge-case data, faster iteration, and a tighter feedback loop. Palmer Luckey backed Tesla's robotaxi business model, arguing that autonomous vehicles could capture a significant share of a ride-hailing market worth more than $150 billion by eliminating driver costs11. If that logic holds, then lifting a nighttime curfew is evidence of rapid engineering success, not recklessness. Musk identified nighttime visibility as one of the biggest remaining challenges in early October, describing the difficulty of spotting grey kittens on grey tarmac in the dark. The 24/7 expansion arrived roughly five weeks later. But the scale-begets-safety argument requires a testable claim: that incident rates fall as miles accumulate. Neither Tesla nor May Mobility has published per-mile disengagement or incident-rate trends for their current fleets. Until that data exists, the argument is a hypothesis, not a demonstrated flywheel.
Start with May Mobility. The Ann Arbor company announced on November 12 that it targets deployment of more than 10,000 autonomous ID. Buzz vehicles through 2031, with the first production vehicles targeted for delivery in 2028 and U.S. on-road testing planned for early 2027. ECARX will handle system integration for May Mobility's Autonomous Development Kit, while Charlton Intelligence.ai has joined the partnership to assist with vehicle deployment. May Mobility is targeting all-in cost savings of more than 50% compared with its current-generation vehicle through the supply-chain collaboration2,9.
The numbers are large. A May 19, 2026 strategic framework agreement between ECARX and May Mobility estimated the project's total value at approximately $750 million over its entire duration, though that figure is an estimate of a conditional project's value, not a finalized order amount8. May Mobility went through a $1.4 billion acquisition in September. Separately, a $337 million IPO was recorded the same month. The company is capitalizing aggressively around a deployment target that remains, by its own disclosure, a target rather than a confirmed commitment. That distinction matters: a planning ceiling used to size a supply chain is normal corporate practice; what makes it analytically significant here is the ratio of capital deployed to operational proof.
May Mobility reports completing more than 600,000 commercial autonomous rides covering 1.1 million miles in the U.S. and Japan3. That base is roughly three orders of magnitude smaller than what 10,000 vehicles would generate annually. The target is not overreach because it is aspirational—targets should be—but because the capital structure is already priced to it while the operational validation is not.
The caution flags are embedded in the announcement itself. The 10,000-vehicle figure is May Mobility's deployment target, not a commitment confirmed by Volkswagen, which is not quoted in the announcement. May Mobility has not identified which companies would own and operate the ID. Buzz robotaxis, which cities would receive them, or what purchase agreements would be used. The cost-savings objective remains a development target for the planned vehicle rather than a realized saving from May Mobility's existing operations. The absence of named fleet partners, city commitments, or Volkswagen production confirmation leaves the deployment path largely undefined.
Tesla, meanwhile, is pushing in the opposite direction: not announcing future scale but expanding present operations. The company's official Robotaxi account on X announced that the service now runs 24/7 in every city where it is available. Elon Musk said a robotaxi works 168 hours a week. The shift caps a rapid schedule expansion: on October 29, Tesla listed robotaxi availability from 6 a.m. to 2 a.m., and thirteen days later the overnight gap was gone7. The service uses both Model Y vehicles and Cybercabs, though the Cybercabs are still only available in Austin.
Vector Wire reported on October 4 that Tesla's camera-only architecture had forced an earlier curfew because nighttime edge cases were unresolved. Musk identified nighttime visibility as one of the biggest remaining challenges in early October. The curfew lift arrived roughly five weeks after that acknowledgment. Whether Tesla solved the nighttime problem or accepted the residual risk would show up in per-mile incident data during overnight hours, which Tesla has not disclosed. Without it, the curfew lift is ambiguous: it could mark an engineering milestone or a commercial decision to prioritize utilization hours over demonstrated nighttime safety margins.
Which brings us to Clarkson. The broadcaster booked two Tesla robotaxi rides in Texas for his Sunday Times column. The first car stopped around the corner from his pickup point and sat there while its onboard assistant insisted it was already moving; a Tesla representative reportedly told him to cancel the ride and request a replacement vehicle. The second car reportedly ran red lights, braked for obstacles Clarkson says were not on the road, and kept driving once it reached his destination. Clarkson wrote that he was genuinely worried for about 10 minutes during the ride, unable to do anything but watch. The Cybercab has no steering wheel and no pedals.
ANALYSIS One columnist's anecdote is not a safety study. Clarkson is a television personality known for adversarial automotive commentary, not an engineer; his account is self-reported, not instrumented; and Tesla has not publicly responded to the specific claims. If NHTSA's standing general order data or Tesla's own disengagement reports show Cybercab performing within acceptable incident-rate bounds, the anecdote is noise. But the account matters for a narrower, structural reason: the Cybercab's design removes every manual override. In a Model Y robotaxi, a rider can grab the wheel. In a Cybercab, the rider's only option during a malfunction is to watch. That design choice does not make the Clarkson ride representative, but it does raise the consequence ceiling of any software failure that does occur—and it makes vivid public accounts of misbehavior disproportionately damaging to the regulatory confidence a no-steering-wheel vehicle requires, regardless of base-rate frequency.
State registration data cited in November show Tesla Robotaxi LLC with 1,244 registered vehicles in Texas, compared with Waymo's 1,259. Vehicle engineering VP Lars Moravy has said Cybercab production is outpacing deployment, limiting fleet growth. Waymo closed a $5 billion debt round in October. The Texas registration figures show near-parity between the two largest operators, and the fact that Texas regulators have permitted both fleets to reach that size is real evidence that the regulatory environment is, for now, permissive rather than restrictive. But permissive is not permanent. The competitive dynamic ahead is less about vehicle counts than about whether the public-trust and regulatory environment each company faces will remain cooperative as fleets scale from hundreds to thousands.
The deeper question is whether the industry's deployment ambitions are calibrated to the regulatory environment they will encounter. May Mobility's plan to deploy 10,000 vehicles across global markets by 2031 assumes regulatory access in cities it has not yet named. Tesla's 24/7 expansion assumes continued permission to operate without overnight restrictions. Both assumptions are testable, and both could break.
Uber's autonomous-mobility leadership is itself in flux. Sarfraz Maredia, Uber's global head of autonomous mobility and delivery, departed on November 16. Whether Uber reaffirms its AV-partnership roadmap under new leadership is now a material watch item, given that May Mobility plans to launch an Uber service in Arlington, Texas, in 2027.
What to watch: whether any city regulator responds to the Clarkson account or similar reports by imposing new operating restrictions on Cybercab specifically, distinct from Model Y robotaxis. Whether May Mobility names fleet-operator partners, target cities, or a Volkswagen production commitment before U.S. road testing begins in early 2027. Whether Tesla's overnight utilization data, if disclosed, shows meaningful rider demand during the hours previously blacked out. And whether Volkswagen itself confirms any production commitment to the ID. Buzz robotaxi variant, a step that has not yet occurred.