Comparing Tesla’s robotaxi rollout with Waymo’s on safety demonstrates how seriously the former is taking the matter.
Elon Musk, Tesla’s (TSLA +0.68%) CEO, says he expects Tesla’s robotaxis to operate without safety drivers by 2026. In stating on a recent earnings call that Tesla’s “single biggest expansion in production” in 2026 will be the Cybercab, he’s clearly expecting big things from robotaxis in 2026.
Tesla vs. Waymo on safety
There’s little doubt that Tesla is taking a cautious approach to its robotaxi rollout, and for good reason. As Musk also noted on the last earnings call, Tesla is “paranoid about safety,” as “All it takes is like one in 10,000 trips to go wrong, and you’ve got an issue.”
Building on the “one in 10,000” quote, interpolating National Household Travel Survey data suggests that the average daily non-highway trip in the U.S. is about 7 miles, so 10,000 non-highway trips is roughly equivalent to 70,000 miles. Alphabet–owned Waymo’s average ride length is approximately 6.7 miles , so let’s use that as a rough estimate of average trip length. Based on Musk’s “one in 10,000” trips comment, that makes 67,000 miles before a robotaxi company has an “issue.”
Today’s Change
(0.68%) $2.74
Current Price
$403.99
Key Data Points
Market Cap
$1344B
Day’s Range
$398.50 – $411.78
52wk Range
$214.25 – $488.54
Volume
1M
Avg Vol
89M
Gross Margin
17.01%
Dividend Yield
N/A
To date, Tesla has reported seven collisions with its robotaxi service to the end of September, and management claims its robotaxis had covered “more than” 250,000 miles at the end of the third quarter. This implies a collision rate every 35,700 miles, which is a significantly worse rate than the 67,000 miles rate interpolated above.
By way of comparison, Waymo data indicates 2.1 incidents per million miles driven, and using the average trip length of approximately 6.7 miles, yields about 71,400 miles before an incident in a Waymo vehicle. That’s better than Tesla’s implied robotaxi to date, and also better than my estimate of what Tesla might expect above.
A few considerations
It’s essential to keep a couple of key points in mind — one negative and two positives. First, the Tesla data doesn’t include incidents where the safety driver/passenger may have stepped in to override FSD to avoid a collision.
Image source: Tesla.
Second, the “law of small numbers” needs to be considered. Take one collision out of Tesla’s numbers, and the average is 41,666; add one more, and it’s 31,250.
Third, it’s still relatively early in Tesla’s rollout, and since Waymo’s safety has improved over time, as it builds data, it’s reasonable to expect Tesla’s robotaxi to do the same. Moreover, Tesla has vastly more data on its FSD than Waymo will ever come close to matching. While FSD data is not the same as robotaxi data, and unsupervised FSD is not the same as what Tesla uses in robotaxis, it still has a huge opportunity to leverage the mass of data it gathers from publicly available FSD to refine what is used in robotaxi FSD.
Tesla’s Full Self-Driving data
Tesla’s latest Full Self-Driving (FSD) vehicle safety report provides ample support for the idea that its FSD (which is still in ongoing development) is safe. The latest FSD safety data from Tesla demonstrates 986,000 miles driven before even a minor collision using FSD. Note that Tesla’s reports do “not attribute fault in our collision reporting.”
Not only does the data demonstrate the safety of the technology company’s FSD, but it also notes the superiority of using FSD and Tesla’s active safety features over driving without them.
|
North America Non-Highway |
Miles Before a Major Collision |
Miles Before a Minor Collision |
|---|---|---|
|
Tesla with supervised FSD |
2.9 million |
986,000 |
|
Tesla manually driven with active safety features |
1.7 million |
579,000 |
|
Tesla driven manually, without any safety features |
776,000 |
270,000 |
|
U.S. average |
505,000 |
178,000 |
Data source: Tesla FSD Vehicle Safety Report
What it means to investors
Clearly, Tesla is taking a cautious approach to the rollout of its robotaxis. While the robotaxi safety data is far from perfect, it’s of a magnitude that gives confidence that Tesla will achieve its safety aims. Moreover, it’s a small sample of data, and a fuller picture will emerge with more data.
Image source: Tesla.
A far larger sample of data, such as Tesla’s overall FSD vehicle safety data, provides considerable reassurance for investors. The latest FSD vehicle safety data is highly complementary, and it appears to be a matter of when, not if, Tesla receives approval to operate robotaxis without safety drivers. Whether it occurs before Tesla starts ramping up Cybercab volume remains to be seen, but it’s the next key milestone investors need to watch for.