While Tesla prepares to hold a launch event for its 2-seat, no-steering-wheel-or-pedals Cybercab on September 3rd, FSD crashes among Tesla owners in the US are skyrocketing. Recent data shows that FSD non-highway crashes are not only growing faster than total FSD city miles driven, but are also becoming increasingly serious. This is why the September 3rd Cybercab launch event (its second one since being introduced in October 2024) should be another “sell-the-news” event for Tesla’s stock: the Vested Interests are using the options market to pump Tesla shares back over $400 on this event. Unless Musk announces that NHTSA gave Tesla approval to produce up to 2,500 Cybercabs per year for 2 years (like it did with Zoox recently), Tesla’s stock could sell off after this launch event, as it did after the Cybercab’s original launch (Tesla shares dropped by 8.8% to $218 the day after the October 2024 Cybercab launch). The recent crash data from NHTSA is showing a marked deterioration in FSD safety, while Tesla’s own data supports this trend. Here are the key points: FSD serious injury crashes spike in Q2 2026: In the latest Level 2 ADAS crash report published by the National Highway Traffic Safety Administration (NHTSA), Tesla saw its non-highway (“city”) incidents double to 380 in Q2 compared with Q1, while estimated “serious injury or worse” crashes rose by 105% to 170. Cumulative FSD miles driven on non-highway roads only rose by 34% in Q2 to 4.3 billion miles, compared to Q1’s 3.2 billion miles, so crashes and serious injuries are outpacing FSD miles driven—a horrible trend for any AV operator. Even Tesla’s FSD Vehicle Safety Report is weak: Tesla’s latest FSD Safety Report from late July even showed weaker results than normal: city miles driven before a major collision came to 3.54 million miles, which was only 2.6% better than the previous quarter’s result of 3.45 million miles (an 18.3% improvement versus Q4 2025). FSD crashes on non-highway roads rose almost in line with city miles driven by Tesla’s own estimates: FSD Supervised saw its city miles driven rise by 34% to 2.89 billion miles in the latest July Safety Report, while major collisions grew by 31% to 819 incidents compared to the previous quarterly report. New estimates of serious crashes show FSD only 8% safer than humans: Using data that Tesla reports to NHTSA about its crashes, such as pre-crash vehicle movement and speed, what the Tesla crashed with, etc., serious or worse FSD city crashes come to a stunning 946 incidents since 2021, of which 909 crashes (96%) had the highest injury severity listed as “unknown” (see Figure 1). FSD crashes with human supervision where injuries were estimated to be “serious or worse” are now 21x greater than Waymo’s driverless robotaxis (see Figure 3 below). Digging up serious crashes from Tesla’s undisclosed injury levels: As detailed in my July report, Tesla’s Crash Reports Hide FSD Injuries (here), Tesla has filed reports where the injury severity level is “unknown” in 94% of its 3,693 crash reports. In a May 2025 crash on a street in Boca Raton, a Tesla on FSD had a pre-crash speed of 103 mph in a 40-mph zone (which sounds fatal), but Tesla not only marked the injury level as “unknown”, but also redacted information related to what the Tesla crashed into in that 40-mph zone. Figure 1 shows the trend of serious or worse injury FSD crashes, with the 25 cumulative fatal crashes on city streets disclosed by Tesla (albeit not the fact that the incidents occurred on FSD, which Tesla heavily redacts), and the 29 serious injury crashes disclosed by Tesla. Note that, in my screening of FSD crashes on non-highway roads (which includes determining whether roadway types labeled as “unknown” were non-highway roads), Tesla has reported only 29 serious-injury crashes, while I was able to identify 909 crashes that very likely involved serious injuries or fatalities (see Figure 2 below). Figure 1: Cumulative FSD Serious or Worse CrashesSource: NHTSA & MH estimates.