Last week investors read between the lines from Tesla’s conference call and determined that the initial Robotaxi rollout in Austin appears to be progressing more slowly than anticipated. On that call, management provided only limited details, noting just a “handful of vehicles” in service and roughly 7k miles driven so far in Austin. No specific vehicle count was given and, more importantly, the company offered no timeline for removing safety operators or expanding the service.

My theory is they’re waiting for intervention rates to drop to comfortable levels before accelerating. This is based on Musk stressing on the call that they “don’t want to rush” and are being “extremely paranoid” about safety. My sense is that over the first month of the service, the level of interventions has been higher than what they were expecting prior to launch, which explains why the company is taking a slower pace.

The takeaway is Tesla is making the right decision to slow play the expansion. Removing the safety operator opens up the risk of an accident, which would not only set back Austin by months but also risk setting back the willingness of new cities to give the company the green light.

Over the weekend, Tesla kicked off a Robotaxi service in the San Francisco Bay Area. All indications are the rollout will be slow, mirroring Austin’s start with the service restricted to employees’ friends and family and other select riders rather than the public. While by definition adding the Bay Area falls into the category of an expansion of the fleet, the fact that it requires a safety operator and involves a limited number of vehicles means there has been little material progress since the initial Austin launch.

It’s also unclear if the company has the required California permits to operate a commercial autonomous service with or without a safety driver.

The takeaway is that Tesla’s Robotaxi expansion is progressing, but in baby steps, and widespread public Robotaxi availability in any market will require both regulatory green lights and further improvements in the technology.

Despite the slow Robotaxi rollout, Tesla is doubling down on its autonomy ambitions, signing a deal to buy a large number of chips for its next generation hardware that will be available in 2027 to unlock a far more advanced autonomous capability for Robotaxi, FSD, and Optimus. The company signed a $16.5B seven year deal with Samsung to produce Tesla’s next generation “AI6” AI chip at Samsung’s new factory in Texas.

Musk indicated that the $16.5B figure is just a “bare minimum” commitment and that actual output could end up “several times higher,” underscoring the scale he is envisioning for Tesla’s AI hardware needs.

I believe the short one year time frame that the company will spend on the A15 (mid 2026 to mid 2027) suggests Tesla sees its current generation A14 and upcoming A15 chips as interim solutions. In other words, the existing Hardware 4 and Hardware 5 FSD computers (powered by Tesla’s AI4 and AI5 silicon) may not provide the full capability required for Tesla’s ultimate autonomous goals.

This bet on AI6 reflects Tesla’s conviction that vision based autonomy and advanced robotics will require orders of magnitude more computing capability, and it highlights that while widespread Robotaxi and Optimus adoption may be years away, the eventual impact could be bigger than most currently believe.

Disclaimer