A paid ride with a driver at the wheel is a useful test, not a driverless service. Tesla began offering those rides in the Bay Area in 2025. I drive a Model 3, so I am interested in the product, but the label “Robotaxi” needs a qualifier here: Tesla’s 2026 investor update still listed the SF Bay Area with a safety driver.
That distinction changes the business case. A supervised ride can test dispatch, routing, pickup points, and customer demand. It cannot demonstrate the labor savings of a driverless fleet while someone sits ready to intervene. Tesla’s later filing described Bay Area operation under a California TCP permit using FSD (Supervised), which requires active driver supervision.
I think Tesla should get credit for running a paid service and no credit yet for removing the driver in California. Waymo’s service map shows why the difference matters: a customer booking a car without anyone in the front seat is buying a different operating product.
The numbers I would ask Tesla for are paid rides by city, interventions, hours per vehicle, and how much human support each ride takes. Forecasting thousands of Bay Area robotaxis from a supervised launch skips the expensive work between those two stages. A small service that discloses its labor honestly is more informative than a large vehicle target.