Tesla's fleet of Full Self-Driving vehicles has surpassed 10 billion miles, according to the automaker's updated safety page. That threshold was set by CEO Elon Musk earlier this year as the necessary data milestone for "safe unsupervised" driving.
See also: Tesla HW3: Millions of vehicles will not receive unsupervised FSD
The achievement represents a significant acceleration in data collection — the fleet was logging about 29 million miles per day by the end of April, up from 14 million miles per day at the start of the year. However, reaching this milestone doesn’t mean Tesla is ready to implement Level 4 autonomy.
In January 2026, after failing to deliver unsupervised FSD by the end of 2025 as promised, Musk said the company needed about 10 billion miles of real-world driving data to achieve safe unsupervised autonomous driving. This number was a revised goal, as he had previously said 6 billion miles would be enough.
Tesla's safety page now reports one major crash per 5.3 million miles under FSD (supervised), compared to one per 660,000 miles for the average American driver. The company presents this as evidence that the system is significantly safer than human driving.
But experts have long criticized Tesla’s safety methodology. Tesla uses different methods to count crashes than the NHTSA data it compares to, creating a misleading picture. Additionally, Tesla’s robo-taxi fleet in Austin, which logged about 800,000 miles through February 2026, reported 14 crashes to NHTSA — a rate about four times the average rate of crashes caused by humans in similar urban conditions.
The 10 billion mile figure is impressive, and collecting data on that scale is truly valuable for training neural networks. Tesla's fleet advantage — millions of vehicles collecting real-world driving scenarios — is something no competitor can replicate.
See also: Tesla sales recover slightly but challenges remain
But there's a fundamental question that an odometer can't answer: when is the system truly ready to take full responsibility for driving?
Musk has now pushed back unmanned FSD for Tesla consumer vehicles to the fourth quarter of 2026 at the earliest, and even that timeline comes with heavy caveats, as Tesla has missed all of Musk's deadlines over the past decade.
Meanwhile, Waymo, which operates at Level 4 without a human behind the wheel, has expanded to 10 cities with driverless operations and is aiming for 1 million weekly trips with its 6th-generation system. Waymo reports 90% fewer serious injury crashes and 82% fewer airbag deployments than regular drivers in 127 million miles of autonomous driving.
The difference is that Waymo takes legal responsibility for driving, while Tesla does not for 'Full Self-Driving'.
Data is critical for training autonomous systems, but not all data is created equal. Elon Musk set 10 billion miles as the necessary database for unsupervised FSD, but it wouldn’t be the first time he’s been wrong about Full Self-Driving. He’s been wrong about this more than any other topic. There’s no magic mile milestone that unlocks autonomy.
The key issue is that Tesla doesn’t know when it can achieve Level 4 at scale, because the end of the extreme cases is almost infinite. It’s not necessary to cover every possible scenario — some of the best human drivers haven’t encountered every possible situation, but they can still handle a wide range of them safely. The challenge is knowing when an automated system has reached that level of generalized capability.
See also: Musk: Tesla Model S and X production ends – ~600 left
FSD is undoubtedly improving. The system today is significantly better than it was a year ago. However, it’s doubtful that Tesla will take responsibility for it — meaning true Level 4, where the company is responsible and the driver can look away — anytime soon. Hitting 10 billion miles is a marketing milestone, while accepting responsibility for billions of miles of unattended driving is a whole different proposition.
