Tesla delays driverless driving to mid-2027, limiting it to specific city zones
Owners must integrate their vehicles into Tesla’s managed Robotaxi network to access unsupervised driving, a shift from the original promise of autonomous capability anywhere.
By Benjamin Bernard ·

Tesla has officially set a target for the second quarter of 2027 to launch unsupervised Full Self-Driving capabilities in vehicles owned by customers. This announcement, made during the company’s Q3 2026 earnings call, represents the latest iteration of a timeline that CEO Elon Musk has adjusted repeatedly over the past decade. While the date is specific, the conditions under which this feature will operate are significantly more restrictive than many buyers originally anticipated.
The primary takeaway from the call is that this is not a simple software update that grants owners free rein to drive anywhere. AI chief Ashok Elluswamy clarified that the initial rollout involves owners adding their personal vehicles to Tesla’s managed Robotaxi fleet. In this model, Tesla operates the vehicle, meaning the car can only drive itself unsupervised within the specific geographic areas where Tesla’s Robotaxi service is already active.
This limitation is significant because the current Robotaxi zones are relatively small. The service currently operates in six US metropolitan areas, but the active geofences cover only small slices of those cities. For example, the Miami zone focuses on West Miami and a corridor toward Doral and Sweetwater, while Orlando and Tampa launched with similarly confined boundaries in July. Within these zones, the vehicles stick to low-speed routes, generally avoid highways, and steer clear of complex urban environments.
Regulatory hurdles further complicate the picture, particularly in California. In the Bay Area, Tesla’s service is not yet unsupervised because state rules require a safety driver in every vehicle. Tesla has chosen not to release certain data to the California DMV, a step required for operations beyond Level 2 autonomy. This regulatory standoff highlights the gap between Tesla’s internal readiness and external approval processes.
The scale of the current operation remains modest compared to broader industry expectations. By the end of Q3, Tesla’s Robotaxi fleet had accumulated nearly 2 million unsupervised miles, a milestone reached just a month after the Cybercab joined the Austin fleet. However, third-party tracking from earlier this year suggested that the number of fully driverless Teslas on the road was in the dozens, far short of the hundreds of thousands Musk has previously discussed. Competitor Waymo, which entered Orlando and Tampa before Tesla, has already logged over 3 million miles.
This new structure contrasts sharply with the original marketing pitch for Full Self-Driving. Since 2016, the promise to buyers has been a car that would eventually drive itself anywhere via software updates. In July 2025, Musk stated that unsupervised personal use would be available in certain geographies by the end of that year. He had previously suggested in 2024 that he would be surprised if the technology wasn’t ready the following year. Now, the target has slipped to Q2 2027, and the product is narrower: the car joins Tesla’s fleet, operates in Tesla’s zones, and runs on Tesla’s terms.
Complicating matters for some users, owners of the HW3 hardware have already been informed that their vehicles cannot run unsupervised FSD at all, with no clear upgrade path provided. Elluswamy reiterated that safety remains the primary bottleneck, which is why the current system limits itself to low-speed roads and avoids difficult environments. He estimates these safety issues will be resolved by next year, though he acknowledged the uncertainty surrounding that timeline. For now, the goalpost has shifted from a car that drives you anywhere to a car that can work as a Robotaxi inside a small, controlled geofence.
Source: Electrek · original report published





