Tesla FSD Is Closing In on 9 Billion Miles — And Elon Musk Says 10X Safer Than Humans Is the Target

In back-to-back posts this morning, Elon Musk made his strongest public statement yet about Tesla's Full Self-Driving capability: the AI will ultimately be more than 10 times safer than human driving. That's not a vague aspiration — Tesla's own safety data is already closing in on that benchmark, and the fleet's real-world mileage is about to cross a milestone that makes the claim harder to dismiss.

📊 Key Figures

Metric	Value	Context
FSD Supervised miles (as of Mar 5, 2026)	8.5 billion	9B milestone imminent
Daily accumulation rate	~20M miles/day	~1B miles every 50 days
Major collision rate — FSD Supervised	1 per 5,300,676 miles	Feb 2026 data
Major collision rate — U.S. average	1 per 660,164 miles	National benchmark
Current implied safety multiplier	~8X	Musk's target: >10X
Projected 10B mile date	Mid-2026	At current accumulation rate

Where the 9 Billion Mile Number Comes From

Tesla community analyst @wholemars flagged this morning that the FSD fleet is on the cusp of 9 billion cumulative miles — and suggested the threshold may have already been crossed by the time the post went live.

The context behind that number: Tesla's FSD Supervised fleet surpassed 8 billion miles on February 18, 2026, and had reached 8.5 billion by March 5. At the current accumulation rate of approximately 20 million miles per day, the fleet is adding roughly 1 billion miles every 50 days. That puts 9 billion squarely in the rearview mirror — and 10 billion on track for mid-2026.

These aren't simulated miles. They're real-world trips on public roads, in varied weather, across every type of driving condition — the kind of data that no competitor has come close to matching at scale.

What Musk's "10X Safer" Claim Actually Means

The statement sounds bold, but Tesla's own published safety statistics show the system is already in the vicinity. According to Tesla's February 2026 safety report, FSD (Supervised) recorded one major collision every 5,300,676 miles. The U.S. national average is one major collision every 660,164 miles. That's approximately an 8X improvement — already remarkable, and closing in on the 10X target Musk cited this morning.

It's worth being precise about what these numbers measure. The comparison is between FSD Supervised — where a human driver is present and legally responsible — and the general U.S. driving population. Critics argue this introduces selection bias: FSD users may be more attentive drivers on average, and Tesla's hardware skews toward newer vehicles with more safety equipment. Tesla disputes this framing, pointing to the raw mileage scale as statistically meaningful regardless.

The Regulatory Counterweight

Musk's confidence lands against a complicated regulatory backdrop. NHTSA escalated its FSD investigation (PE25012) to a full engineering analysis on March 20, 2026 — just nine days ago — expanding the scope to 3.2 million Tesla vehicles. The probe focuses on whether FSD may fail to detect or adequately warn drivers under poor visibility conditions, including concerns about traffic violations such as red light running and wrong-way driving.

That escalation doesn't invalidate the safety statistics, but it does signal that regulators want independent verification — not just Tesla's own data — before accepting the 10X narrative at face value. The outcome of that engineering analysis will carry significant weight for FSD's regulatory future, particularly as Tesla moves toward unsupervised autonomy.

What's Coming Next for FSD

The version currently rolling out to Hardware 4 vehicles is FSD (Supervised) v14.2.2.5, which began deployment in mid-March 2026. FSD v14.3 is in active testing and is expected to reach wide release in early-to-mid April 2026. For owners still on Hardware 3 vehicles, FSD v14 Lite is anticipated for late June 2026.

On the subscription side: FSD is now exclusively available at $99/month following the discontinuation of the one-time purchase option on February 14, 2026. Musk has previously indicated the price will rise as capabilities improve — a trajectory that makes the 10X safety claim commercially important, not just a talking point. For more on our FSD coverage, see our full archive.

🔭 The BASENOR Take

📰 Deep Dive

The significance of the 9-billion-mile threshold goes beyond a round number. Each additional billion miles represents a statistically richer dataset — more edge cases, more adverse weather encounters, more complex urban intersections. At 20 million miles per day, Tesla is accumulating what would take a traditional automotive safety validation program years to replicate in controlled testing. This is the core of Tesla's competitive moat in autonomy: no other company is generating real-world supervised driving data at this scale.

The gap between the current ~8X safety figure and Musk's >10X target is also instructive. It suggests Tesla believes meaningful capability improvements are still ahead — likely tied to the v14.x release cycle and the ongoing shift from rule-based to end-to-end neural network architecture. Each version update isn't just a feature addition; it's a recalibration of the system's ability to handle the long tail of rare driving scenarios that cause most serious accidents.

For FSD subscribers, the practical implication is straightforward: the system you're using today is already statistically safer than average human driving by a wide margin, and it's improving with every update. The open question isn't whether FSD is safer — the data suggests it is — but whether it's safe enough, consistently enough, across all conditions, to satisfy regulators who are now actively scrutinizing its edge case behavior. That's the challenge v14.3 and beyond will need to answer.