Today, September 3, 2026, Tesla is holding an invite-only, livestreamed launch event at Gigafactory Texas to formally unveil the production version of the Cybercab — a two-seat robotaxi built with no steering wheel and no pedals. It’s the most visible milestone yet in Tesla’s bet that Full Self-Driving software can carry it from a driver-assist feature into a purpose-built autonomous vehicle business. But the timing also puts Tesla’s approach in sharp contrast with Waymo, which spent the week before the event arguing that Tesla is climbing the wrong mountain entirely.
What’s launching
The Cybercab event is tightly controlled: guests must be 21 or older, face ID checks at the door, and the invite list skews toward riders who logged the most trips on Tesla’s existing Robotaxi service, supplemented by mail-in entrants. Everyone else gets the livestream. It’s worth being precise about what today’s event is — and isn’t. This is a product unveiling: production specs, pricing signals, and a clearer picture of how Cybercab slots into Tesla’s existing Robotaxi network. It is not a regulatory approval, and public access to Cybercab rides will remain limited well beyond today.
Where Tesla’s robotaxi program actually stands
Tesla’s Robotaxi service has been running in limited form in Austin since June 22, 2025, using modified Model Y vehicles with safety monitors rather than the driverless Cybercab. As of late August 2026, Tesla’s registered autonomous fleet in Texas had grown to roughly 37% the size of Waymo’s entire Texas fleet — real growth, but still a clear gap against the incumbent.
On the software side, Tesla’s FSD v14.1.x update produced the largest sequential jump in four years of independently tracked disengagement data, with one widely cited tracker showing miles-per-critical-disengagement climbing more than 20x versus the prior version. That number is genuinely notable, but it’s also a reminder of how noisy early-release data can be: as more real-world miles accumulated on v14, the same tracker’s rolling average settled back down substantially. The trend line is positive; the exact size of the jump is still being worked out in public, mile by mile.
Waymo’s rebuttal: “a false summit”
Days before Tesla’s event, Waymo published a blog post laying out “10 AI lessons” from its own program — without naming Tesla, but clearly aimed at it. The pointed line: “Simply improving a driver-assist system [Level 2] for full autonomy is a false summit.” Waymo’s argument is that true Level 4 autonomy has to be engineered and validated as a driverless system from the start — closed-course testing, redundant sensing, driverless-only refinement — rather than reached by incrementally upgrading a consumer driver-assist feature that still expects a human behind the wheel.
Waymo backed the argument with scale: over 500,000 paid, fully driverless rides per week (with a stated goal of one million by year-end), roughly 200 million cumulative autonomous miles, and operations across more than ten major cities and freeway networks. By contrast, Tesla’s unsupervised (no safety monitor) driving has totaled around 380,000 miles in total — a figure Waymo’s fleet now covers in less than a day. Waymo also took aim at two specific Tesla design choices: its camera-only sensor suite, versus Waymo’s cameras-plus-lidar-plus-radar approach, and Tesla’s end-to-end neural network architecture, which Waymo characterized as a “black box” that’s harder to build public trust around.
What it means for the evolution of FSD
Strip away the sniping and the two companies are making a genuine bet on different paths to the same destination. Waymo is scaling a system that was built driverless-first, with heavier sensing and more conservative, geofenced expansion — slower, but with a long track record of paid, fully autonomous rides to point to. Tesla is trying to scale a much larger existing fleet and a camera-based, end-to-end neural network that improves through software updates across millions of consumer cars, then use that same stack to power a purpose-built vehicle like Cybercab. It’s a bet on manufacturing scale and data volume closing the gap that Waymo’s approach currently leads on.
Today’s Cybercab unveiling doesn’t settle that argument. But it does mark the point where Tesla’s strategy stops being purely theoretical and starts producing an actual vehicle with no wheel and no pedals — which is exactly the kind of purpose-built commitment Waymo says has been missing from Tesla’s approach until now. Whether that’s enough to close a fleet gap and a data gap this large is the story to watch over the next year.
Sources: Motor1, Electrek — Tesla Cybercab, Electrek — Waymo “false summit”, Yahoo Finance — Piper Sandler on unsupervised FSD
