Are robotaxis poised to be the Next Big Thing™ in North America? It seems so, at least according to Goldman Sachs, which issued a report this week stating that robotaxis have officially entered the commercialization phase of the hype cycle. That assessment appears to be based on an analysis of the total ride-sharing market, which encompasses services that are currently almost 100% reliant on meat-based drivers, such as Lyft and Uber, and is valued at $58 billion. Autonomous ride-hailing services like Waymo, which has a fleet of 1,500 robotaxis operating in several cities across the US, are included in that market but account for less than 1% of the total right now. But, Goldman projects that the market will burgeon to over $336 billion in the next five years, driven in large part by “hyperscaling” of autonomous vehicles.
We suspect the upcoming launch of Tesla’s robotaxis in Austin, Texas, accounts for some of this enthusiasm for the near-term, but we have our doubts that a market based on such new and complex technologies can scale that quickly. A little back-of-the-envelope math suggests that the robotaxi fleet will need to grow to about 9,000 cars in the next five years, assuming the same proportion of autonomous cars in the total ride-sharing fleet as exists today. A look inside the Waymo robotaxi plant outside of Phoenix reveals that it can currently only convert “several” Jaguar electric SUVs per day, meaning they’ve got a lot of work to do to meet the needed numbers. Other manufacturers will no doubt pitch in, especially Tesla, and factory automation always seems to pull off miracles under difficult circumstances, but it still seems like a stretch to think there’ll be that many robotaxis on the road in only five years. Also, it currently costs more to hail a robotaxi than an Uber or Lyft, and we just don’t see why anyone would prefer to call a robotaxi, unless it’s for the novelty of the experience.
On the other hand, if the autonomous ride-sharing market does experience explosive growth, there could be knock-on benefits even for Luddite naysayers such as we. A report, again from Goldman Sachs — hey, they probably have a lot of skin in the game — predicts that auto insurance rates could fall by 50% as more autonomous cars hit the streets. This is based on markedly lower liability for self-driving cars, which have 92% fewer bodily injury claims and 88% lower property damage claims than human-driven cars. Granted, those numbers have to be based on a very limited population, and we guarantee that self-drivers will find new and interesting ways to screw up on the road. But if our insurance rates fall even a little because of self-driving cars, we’ll take it as a win.
Speaking of robotics, if you want to see just how far we’ve come in terms of robot dexterity, look no further than the package-sorting abilities of Figure’s Helix robot. The video in the article is an hour long, but you don’t need to watch more than a few minutes to be thoroughly impressed. The robot is standing at a sorting table with an infeed conveyor loaded with just about the worst parcels possible, a mix of soft, floppy, poly-bagged packages, flat envelopes, and traditional boxes. The robot was tasked with placing the parcels on an outfeed conveyor, barcode-side down, and with proper separation between packages. It also treats the soft poly-bag parcels to a bit of extra attention, pressing them down a bit to flatten them before flicking them onto the belt. Actually, it’s that flicking action that seems the most human, since it’s accompanied by a head-swivel to the infeed belt to select its next package. Assuming this is legit autonomous and not covertly teleoperated, which we have no reason to believe, the manual dexterity on display here is next-level; we’re especially charmed by the carefree little package flip about a minute in. The way it handles mistakenly grabbing two packages at once is pretty amazing, too.
And finally, our friend Leo Fernekes dropped a new video that’ll hit close to home for a lot of you out there. Leo is a bit of a techno-hoarder, you see, and with the need to make some room at home and maintain his domestic tranquility, he had to tackle the difficult process of getting rid of old projects, some of which date back 40 or more years. Aside from the fun look through his back-catalog of projects, the video is also an examination of the emotional attachments we hackers tend to develop to our projects. We touched on that a bit in our article on tech anthropomorphization, but we see how going through these projects is not only a snapshot of the state of the technology available at the time, but also a slice of life. Each of the projects is not just a collection of parts, they’re collections of memories of where Leo was in life at the time. Sometimes it’s hard to let go of things that are so strongly symbolic of a time that’s never coming back, and we applaud Leo for having the strength to pitch that stuff. Although seeing a clock filled with 80s TTL chips and a vintage 8085 microprocessor go into the bin was a little tough to watch.