Tesla's Cybercab is entering volume production. The harder part isn't done.

Tesla's Cybercab production line at Giga Texas has been running since February, and continuous volume production is supposedly starting this month. Elon says consumers will be able to buy one for under $30,000 and add it to Tesla's ride-hailing fleet. That part sounds exciting.

Here's the part that doesn't get talked about enough.

**The Cybercab has no steering wheel. That changes everything about what "production" means.**

This isn't a Model 3 that you can hand to someone and let them drive with FSD as a backup. The Cybercab has no pedals and no steering wheel. If unsupervised Full Self-Driving isn't solved, it doesn't matter how many roll off the line — they can't move under their own operation. The $30,000 consumer purchase and fleet monetization model only works if Level 4 autonomy is actually solved first.

Tesla currently has about 8 billion miles of FSD data. Elon has said Tesla needs roughly 10 billion miles to achieve "safe unsupervised self-driving." At the rate they've been accumulating, that's a finite gap — but it's also a gap that has been "almost closed" for about a decade.

**Let's review the actual track record:**

- 2015: "Complete autonomy in approximately two years"
- 2016: Fully autonomous drive from LA to New York, no human touch
- 2017: Drivers will be sleeping in Teslas within two years
- 2019: Robotaxis by 2020
- 2020: Software "feature complete" by year's end
- 2023: Full autonomy coming "later this year"
- 2024: "Shocked" if unsupervised FSD doesn't arrive in 2025
- 2025: Didn't arrive

A California judge ruled in December 2025 that Tesla made "actually, unambiguously false and counterfactual" claims about its autonomous capabilities. Tesla's FSD stagnated through most of 2025.

**The current robotaxi operation tells you where things actually stand.**

Tesla's Austin robotaxi service launched in June 2025. Nine months later:

- About 35 vehicles operating in Austin
- Every single one has an in-car safety supervisor with a finger on a kill switch
- One unsupervised vehicle, operating in a very limited part of the service area
- Fleet crash rate: roughly one incident every 57,000 miles. The US average for human drivers: one crash every 500,000 miles
- Tesla redacts crash narratives from its NHTSA filings, unlike Waymo, Zoox, and Aurora which publish full details

For context: Waymo operates over 2,500 fully driverless vehicles across multiple US cities, no safety monitor in any car, 127 million autonomous miles logged, and an 85% lower crash rate than human drivers in injury-causing incidents.

**The talent exodus is a real problem**

The OTA update infrastructure director who built Tesla's robotaxi backend left in March. The Cybercab program manager left in February, weeks before volume production was supposed to begin. These aren't replaceable positions — the knowledge of how the system's software stack actually works walked out the door at the worst possible time.

**The honest picture**

The Cybercab is entering volume production. That's real. But the thing that makes it actually valuable — unsupervised Level 4 FSD — isn't solved yet, and the historical pattern suggests treating the timeline as certain is a mistake. The consumer purchase model, the fleet monetization, the whole "own a Cybercab and let it earn for you" narrative — all of it depends on solving the hard part first.

It's worth watching closely. But the gap between "production started" and "the business model works" is where Tesla still has to prove something it's been promising to prove for ten years.