Ten billion miles is an almost incomprehensible number. To put it in some perspective, that's roughly 40,000 trips to the Moon and back. Tesla dropped that figure this week as a milestone for its Full Self-Driving program, framing it as proof that the technology is maturing, accumulating real-world data at a scale no competitor can match. It's the kind of stat designed to silence critics and reassure investors.
Then a California jury awarded a Tesla owner $10,000 after finding that the company had made misleading claims about FSD's autonomy capabilities. Not a class action. Not a regulatory fine. A single buyer, in court, winning because Tesla oversold what the product actually does.
Those two things happening in the same week tells you almost everything about where Tesla's autonomy program stands right now.
What 10 Billion Miles Actually Means
Tesla has been collecting FSD mileage data from its customer fleet for years, and the 10 billion figure represents cumulative miles driven with the system engaged across all vehicles using the software. That's a genuinely significant dataset. Machine learning systems improve with data, and Tesla's approach of using real-world customer driving as a training pipeline has always been central to its strategy.
Waymo, by comparison, has logged somewhere around 20 million fully autonomous miles. The numbers aren't directly comparable because Waymo's vehicles operate without any human backup driver in geofenced urban environments, while Tesla's FSD still requires hands on the wheel and driver attention at all times. But the scale difference illustrates why Tesla's fleet-based data collection approach has always been seen as a potential long-term advantage.
The problem is that miles driven with FSD engaged are not the same as miles driven autonomously. Every one of those 10 billion miles had a human in the driver's seat, legally responsible for the vehicle, ready to intervene. The system has improved considerably since its early versions, especially with the v12 neural net rewrite that shifted FSD away from explicit rule-based code toward end-to-end learned behavior. Journalists and owners who've tested recent versions generally agree it handles common highway scenarios and many urban situations with noticeably more fluency than earlier builds.
But fluency is not the same as reliability, and reliability is not the same as autonomy.
The Jury Verdict and What It Signals
The $10,000 verdict sounds minor in the context of Tesla's market cap, and financially it is. What matters is what the jury was actually deciding. The plaintiff argued that Tesla's marketing, Elon Musk's public statements, and the product's branding created a false impression about the system's capabilities. The jury agreed, at least partially.
Tesla has been selling "Full Self-Driving" as an option since 2016. The name alone has been a source of ongoing controversy. SAE autonomy levels classify vehicles into tiers from Level 0 (no automation) to Level 5 (full autonomy with no human needed). FSD, despite its name, is a Level 2 system. That means the driver must remain engaged and responsible at all times. Tesla's own disclaimer language makes this clear, but the marketing language and the product's name have consistently implied something far more capable.
Musk has compounded this with years of public predictions that haven't aged well. In 2019 he said Tesla would have one million robotaxis on the road by 2020. In 2020 he said FSD would reach Level 5 capability by the end of that year. Neither happened. The gap between the promises and the product has accumulated into a credibility problem that a 10 billion mile announcement can't simply erase.
Consumer advocacy groups have flagged this pattern for years. The National Highway Traffic Safety Administration has opened multiple investigations into FSD and Autopilot incidents. The verdict, even at $10,000, gives individual buyers a legal precedent to point to when arguing they were misled. That precedent matters more than the dollar amount.
The Safety Picture Is Complicated
Tesla regularly publishes quarterly safety reports comparing crash rates per million miles for vehicles using Autopilot or FSD versus national averages for all drivers. The company's numbers consistently show its assisted driving modes performing better than the average human driver. Critics point out that these comparisons have significant methodological problems, including the fact that FSD is engaged most often on highways and in favorable conditions where crash rates are lower regardless of who's driving.
NHTSA's own data paints a more complicated picture. The agency's Standing General Order requires automakers to report crashes involving advanced driver assistance systems, and Tesla has consistently appeared near the top of those reports by volume. Tesla attributes this partly to its much larger ADAS-equipped fleet compared to competitors. That's a fair point, but it also means Tesla is the most consequential player in this space from a public safety standpoint.
There have been fatal crashes involving FSD or Autopilot where investigators found the system behaved unexpectedly or the driver was not sufficiently attentive. Each of those incidents reopens questions about whether the product's name and marketing encourages drivers to over-trust the system. That's not a hypothetical concern; it's a human factors problem with documented real-world consequences.
What Buyers Are Actually Paying For
FSD currently costs $8,000 as a one-time purchase, or $99 per month as a subscription. For that price, buyers get a system that can navigate highway on-ramps and off-ramps, handle stop-and-go traffic, recognize and respond to traffic signals, and navigate some urban environments without driver input. In good conditions, on familiar roads, recent versions can feel genuinely impressive.
The catch is unpredictability. FSD can handle a left turn across traffic in a suburban intersection dozens of times without incident and then behave erratically on the same turn in slightly different lighting or weather. This isn't a knock unique to Tesla; it's a fundamental challenge in neural-net-based driving systems that still don't generalize perfectly across all conditions. But it is a significant gap between what buyers imagine they're getting and what they actually receive.
Some buyers have been vocal defenders of FSD, arguing it genuinely reduces fatigue on long drives and improves over time via over-the-air updates. That feedback is real. Other owners have reported paying $8,000 for a feature they rarely use because it behaves inconsistently enough that they don't trust it for their regular commute. Both experiences exist simultaneously within Tesla's customer base.
The Regulatory and Competitive Pressure Building Up
Regulators in the US, EU, and China are all tightening their definitions of what companies can say about driver assistance technology. The EU's new vehicle safety regulations include stricter rules about how ADAS features must be described in marketing materials. California's DMV has previously warned Tesla that FSD marketing may violate state advertising law.
Waymo's fully autonomous commercial robotaxi service in San Francisco and Phoenix operates under a completely different regulatory framework because it is genuinely driverless in geofenced areas. Waymo has its own safety incidents and limitations, but its product positioning is honest about what it does. General Motors' Cruise faced a brutal regulatory reckoning in 2023 after a serious pedestrian incident in San Francisco, resulting in suspended permits and a significant reputational hit. The message to the industry from regulators has been consistent: overstating capabilities has consequences.
Tesla's path to a genuine robotaxi product runs through its planned Cybercab, which the company unveiled in late 2024 and has promised to bring to market in 2025. Given Tesla's track record on product timelines, skepticism is reasonable. Getting from a supervised Level 2 system to an unsupervised Level 4 or 5 product is not an incremental step; it's a categorical jump that requires regulatory approval, extensive safety validation, and a fundamentally different liability framework.
The Trust Problem Is the Real Problem
Ten billion miles of data is genuinely valuable. If Tesla's AI training pipeline is working as intended, that accumulated experience should be making the system measurably better over time, and by most accounts it is getting better. That matters for the long-term competitive position of the company in what will eventually be a very large autonomous vehicle market.
But the jury verdict points to something the mileage number can't fix. Tesla has spent years building excitement around a product name that implies capabilities the product doesn't yet have. Buyers have paid thousands of dollars based partly on promises that haven't been kept. Some of those buyers have been involved in crashes where the system's behavior was a contributing factor.
The 10 billion mile announcement and the $10,000 verdict exist in the same week because they're part of the same story. Tesla is genuinely advancing autonomous driving technology at a scale few others can match. Tesla is also selling that technology to consumers under a brand name and with marketing language that consistently outpaces the product's actual capabilities. Both of those things are true, and the tension between them isn't going away just because the odometer keeps climbing.
For anyone considering buying FSD today, the honest advice is straightforward: test it on your actual regular routes before committing $8,000. Read the disclaimer language carefully. Understand that you remain legally and physically responsible for the vehicle at all times. The technology may genuinely help you, or it may spend most of its life switched off because it doesn't match your driving environment. The jury in California found out the hard way that the name on the box isn't always the product in the box.