Tesla FSD Hits 10 Billion Miles, Still No Unsupervised Driving

Tesla has officially crossed the 10-billion-mile mark with its Full Self-Driving (Supervised) fleet — a milestone that CEO Elon Musk previously identified as a key threshold for enabling safe unsupervised autonomous driving. Yet despite reaching this self-imposed benchmark, the company has not flipped the switch to let vehicles take full control from human drivers.

The updated data, published on Tesla's safety page, confirms the cumulative mileage milestone. It raises fresh questions about the gap between Musk's ambitious timelines and the complex regulatory, technical, and safety realities of deploying truly driverless vehicles at scale.

Key Takeaways at a Glance

• Tesla's FSD (Supervised) fleet has surpassed 10 billion cumulative miles driven
• Elon Musk earlier this year cited 10 billion miles as the data threshold for 'safe unsupervised self-driving'
• Despite hitting the milestone, Tesla has not enabled unsupervised autonomous driving
• The FSD system still requires an attentive human driver behind the wheel at all times
• Competitors like Waymo already operate fully driverless robotaxis in multiple U.S. cities
• Regulatory approval remains a major hurdle for Tesla's unsupervised driving ambitions

Musk's Moving Goalposts on Autonomous Driving

Elon Musk has a long history of setting — and missing — ambitious deadlines for Tesla's self-driving capabilities. As far back as 2016, Musk claimed that a Tesla would be able to drive itself from Los Angeles to New York without any human intervention. That cross-country demo never materialized.

In early 2025, Musk pointed to the 10-billion-mile cumulative driving figure as a critical data benchmark. The logic was straightforward: the more real-world driving data Tesla collects, the better its neural networks can learn to handle edge cases, unusual road conditions, and the infinite variability of human driving behavior. Reaching 10 billion miles, Musk suggested, would provide enough statistical confidence to prove the system's safety for unsupervised operation.

Now that the number has been reached, the silence is conspicuous. Tesla has not announced any timeline for removing the 'Supervised' label from FSD or for launching an unsupervised driving mode. This pattern of setting data-driven milestones only to move past them without delivering on promises has become a recurring theme in Tesla's autonomy narrative.

What 10 Billion Miles Actually Means — and Doesn't Mean

The sheer volume of 10 billion miles is staggering. To put it in perspective, Waymo, Alphabet's autonomous driving subsidiary, announced in late 2024 that its vehicles had completed over 25 million fully autonomous miles on public roads — a fraction of Tesla's total. However, the comparison is not apples-to-apples.

Tesla's 10 billion miles were driven with a human supervisor behind the wheel, ready to intervene at any moment. The driver monitoring system uses cabin cameras to ensure the person remains attentive. Waymo's miles, by contrast, are genuinely driverless — no human safety driver is required in its operational domains.

• Tesla FSD (Supervised): 10+ billion miles with human oversight required
• Waymo: 25+ million fully autonomous miles with no human driver
• Cruise (GM): Operations paused after safety incidents in late 2023, recently restructured
• Baidu Apollo Go: Over 7 million robotaxi rides completed in China as of early 2025

The distinction matters enormously. Accumulating supervised miles demonstrates that the software can handle most driving scenarios, but it does not prove the system can safely manage every scenario without human backup. The long tail of rare, unpredictable events — a child darting into the street, unusual construction zones, emergency vehicles approaching from unexpected angles — is where autonomous systems face their greatest challenges.

The Technical Gap Between Supervised and Unsupervised Driving

Tesla's approach to autonomous driving differs fundamentally from most competitors. The company relies on a vision-only system, using cameras as the primary sensor modality. Earlier versions of FSD also used radar and ultrasonic sensors, but Tesla stripped those away in favor of a pure camera-based architecture powered by deep learning.

The current FSD stack runs on Tesla's custom HW4 (Hardware 4) computer, which delivers significantly more compute power than the previous HW3 platform. The system processes input from 8 cameras positioned around the vehicle, constructing a real-time 3D model of the driving environment using neural networks.

Moving from supervised to unsupervised driving requires more than just accumulating miles. It demands:

• Redundancy: Backup systems for steering, braking, and compute in case of hardware failure
• Fail-safe behavior: The vehicle must be able to pull over safely if the system encounters a situation it cannot handle
• Validation: Statistical proof that the system is safer than a human driver across all conditions, not just on average
• Connectivity: Remote monitoring and intervention capabilities for edge cases
• Regulatory compliance: Meeting federal and state-level requirements for driverless operation

Tesla's current vehicle hardware may lack some of these redundancy features. Unlike Waymo's purpose-built robotaxis, which include redundant steering, braking, and compute systems, Tesla's consumer vehicles were designed primarily for human-driven operation with driver-assist features layered on top.

Regulatory Roadblocks Loom Large

Even if Tesla's technology were ready for unsupervised driving tomorrow, regulatory approval would remain a significant barrier. In the United States, autonomous vehicle regulation is a patchwork of federal guidelines and state-level rules.

Waymo currently operates its robotaxi service in Phoenix, San Francisco, Los Angeles, and has expanded to additional cities with specific permits. Each new city requires extensive engagement with local regulators, safety demonstrations, and sometimes public hearings. Tesla would face similar — or potentially even more stringent — scrutiny given the scale at which it would deploy unsupervised driving across its existing fleet of millions of vehicles.

The National Highway Traffic Safety Administration (NHTSA) has opened multiple investigations into Tesla's Autopilot and FSD systems over the years, examining crashes and near-misses involving the technology. Any move toward unsupervised driving would likely trigger heightened regulatory attention.

In Europe, the situation is even more restrictive. The UN Economic Commission for Europe (UNECE) regulations and the EU's vehicle type-approval framework currently impose strict limits on automated driving systems. Level 3 autonomy — where the vehicle drives itself in limited conditions but can hand control back to the human — has only recently been approved for specific use cases like highway driving at low speeds, as seen with Mercedes-Benz's DRIVE PILOT system.

Tesla's Robotaxi Ambitions Hang in the Balance

The 10-billion-mile milestone is particularly significant in the context of Tesla's robotaxi plans. Musk has repeatedly described the upcoming 'Cybercab' — a purpose-built autonomous vehicle without a steering wheel or pedals — as a transformative product that could make Tesla the most valuable company in the world.

Tesla unveiled the Cybercab concept in October 2024 at its 'We, Robot' event, projecting production to begin in 2026. The vehicle's entire business model depends on achieving unsupervised autonomous driving. Without it, a car with no steering wheel is simply undrivable.

The company has also discussed plans to launch a robotaxi service in Austin, Texas, using existing Model 3 and Model Y vehicles equipped with FSD. This service would initially operate with supervised driving and a human behind the wheel, gradually transitioning to unsupervised operation as regulatory and technical readiness align.

However, the timeline remains murky. Musk has acknowledged that regulatory approval could vary significantly by jurisdiction, and the company has not provided concrete dates for when unsupervised FSD might be available to consumers.

What This Means for the Broader AV Industry

Tesla's 10-billion-mile achievement, while impressive in raw numbers, highlights a fundamental tension in the autonomous vehicle industry: data quantity does not automatically equal deployment readiness.

The industry is increasingly bifurcating into 2 camps. On one side are companies like Waymo and Baidu that have chosen a geofenced, carefully controlled rollout strategy — launching in specific cities with detailed maps and gradually expanding. On the other side is Tesla, pursuing a generalized approach that aims to work everywhere from day one.

Both strategies have merits. Tesla's massive fleet provides unparalleled data diversity, capturing driving scenarios from snowy Minnesota highways to congested Miami streets. Waymo's focused approach allows for deeper validation in specific environments but scales more slowly.

For consumers, developers, and investors watching this space, the key lesson is clear: reaching a mileage milestone is a necessary but insufficient condition for safe autonomous driving. The real test is not how many miles the fleet has driven with human supervision — it is how the system performs in the rare, critical moments when human-like judgment is most needed and no human is there to provide it.

Looking Ahead: What Comes Next for Tesla FSD

Tesla is expected to continue iterating on its FSD software through over-the-air updates, with version FSD v13 reportedly bringing significant improvements to the system's handling of complex intersections and unprotected left turns. The company's AI team, led by developments in its custom Dojo supercomputer and partnerships with NVIDIA for training compute, continues to push the boundaries of what vision-based autonomy can achieve.

Several key milestones to watch in the coming months:

• Whether Tesla begins its Austin robotaxi pilot as planned in 2025
• NHTSA's ongoing investigations and any new regulatory frameworks for unsupervised driving
• Progress on Cybercab production timelines and pre-production testing
• Competitive moves from Waymo, which continues to expand its service area aggressively
• Tesla's next safety report and whether intervention rates show meaningful improvement

The 10-billion-mile milestone is a testament to the scale of Tesla's data collection apparatus. But as the company itself implicitly acknowledges by keeping the 'Supervised' label firmly in place, miles driven are just one piece of the autonomous driving puzzle. The hardest miles — the ones driven without any human ready to grab the wheel — are still ahead.