S&P 500 slips, but losses kept in check as Nvidia climbs ahead of results
Investing.com -- Tesla’s full self-driving (FSD) capabilities remain a core driver of Piper Sandler’s price target of $400, and interest in the technology has intensified as the company prepares for a potential robotaxi launch in Austin.
"By a wide margin, Tesla (NASDAQ:TSLA)’s FSD software is the largest contributor to our price target," analyst Alexander Potter said in a note. "Given the materiality of this topic, as well as the rapidly-approaching launch date for robotaxis in Austin, we think investors are scrutinizing this issue more closely than ever."
The brokerage last week hosted a call with Elias Martinez, creator of the FSD Community Tracker, to assess the current state of the software. Martinez’s view is that while version 13 of the FSD system has made visible progress, it still cannot support fully autonomous driving. However, the lack of recent visible updates may reflect behind-the-scenes efforts to optimize the product specifically for Austin.
Below are some of Piper Sandler’s key takeaways from the call.
1) The switch from version 12 to version 13 brought a significant jump in performance. “Miles to disengagement jumped 2x between v12 and v13; improvement is clearly evident,” Potter highlighted, citing the tracker data. The progress is even clearer on a three-year timeline.
2) Viewed over a longer horizon, the software’s learning curve is even more pronounced. Version 13 has shown an 8x improvement compared to version 10, with notable gains in handling pedestrians and difficult weather conditions, including snow.
3) Despite the improvements, it remains challenging to compare Tesla’s efficacy with competitors like Waymo. Data from Tesla’s fleet is filtered and not directly comparable.
4) A key metric under discussion was “miles to collision.” Tracker data showed that humans crash every 50,000 miles in city driving.
5) Waymo reportedly averages 45,000+ miles between crashes, while Tesla might be between 4,000 and 10,000 miles—though that number could understate actual performance in Austin.
6) Martinez speculated that Tesla may be building custom maps for Austin. "If true, FSD could ace Austin’s occluded stop signs, no-turn-on-red intersections, etc," Potter said.
7) Additional improvements could come from sensors Tesla currently does not use. “Elon Musk is a camera purist, but mapping and audio sensors would be a big help,” Martinez said, especially in detecting emergency vehicles.
8) Tesla could also benefit from using remote human operators during difficult scenarios, a method already employed by some other autonomous vehicle developers.