After investors bet an estimated ~$100B on self-driving vehicles, startups see little progress, falling back on simulations and remote operators as losses mount
BloombergMax Chafkin
Context & Ripple Effects
This is the second act of a cycle the coverage has tracked for years: the trough of disillusionment was already visible in 2018, when Uber, Tesla, and Waymo all had disappointing years even as MVP-stage startups made progress. Investors kept writing checks anyway — CB Insights counted equity funding for autonomous vehicle companies topping $12B in 2021, up more than 50% year over year — pushing cumulative bets to an estimated ~$100B.
What changed by late 2022 is that the money stopped buying milestones. With full autonomy still out of reach, startups are taking shortcuts — leaning on simulations and human remote operators — while losses mount, validating the earlier argument that survival meant commercializing the least demanding applications of the technology first.
First-order effects
Startups that raised at 2021-era valuations are now burning cash on fleets that cannot drive themselves, substituting teleoperators and simulation for the autonomy they sold investors on.
Backers of the ~$100B aggregate bet face markdowns and down rounds as the gap between promised Level 4 capability and shipped product widens.
Capital consolidates toward the few players with balance sheets deep enough to keep testing, squeezing out the MVP-stage startups that looked promising back in 2018.
Third-order effects
If the pattern holds, the industry restructures around partial automation sold as a product today rather than full autonomy promised tomorrow — a repeatable lesson in what happens when a capital-intensive technology is funded ahead of its technical readiness.
The episode becomes a template for how later AI investment waves get priced: skepticism after hype peaks, then a flight to applications with near-term revenue, with regulators eventually stepping into a field where remote human operators quietly do part of the work marketed as autonomous.
The trend: Autonomous vehicle investing is cycling from broad full-self-driving bets toward narrowly scoped, revenue-first automation, with each funding peak followed by a commercialization retreat.
Anthony Levandowski has gone from driverless car evangelist to apostate: “You'd be hard-pressed to find another industry that's invested so many dollars in R&D and that has delivered so little.” @chafkin on the road to nowhere https://www.bloomberg.com/...
I've got a cover story @bw this week on the long unfulfilled promise self driving cars, and the industry pioneer who now says the whole thing has been wildly overhyped https://www.bloomberg.com/... https://twitter.com/...
The bull market created a world of impractical ideas that demoed well but weren't actually viable products that big tech branded as moonshots. It's a sad moment to see companies conceding that ads, e-commerce and cloud hosting are probably to only good products of big tech left. …
“Autonomous cars,” “self-driving cars,” “robot cars” — whatever you call them, they do not exist, will never exist. This has been one of the bigger scams I've seen the car industry pull. It was maddening to watch as otherwise smart people fell for it. https://www.bloomberg.com/..…
I'm about to get on a vehicle that isn't self-driving, but for me, and potentially as many as 2,000 other people, it is. One driver for all of us! It's called public transportation. Think what $100 billion could do for THAT! https://www.bloomberg.com/...
I first met Anthony Levandowski so many years ago, when he was building a self-driving motorcycle as a student. @chafkin caught up with him recently to talk about the state of self-driving, being pardoned by Trump and his longest, strangest trip in the self-driving field https://…
people forget now but really in 2017ish robotaxis (and not just Elon/Tesla - Waymo) were just around the corner and private vehicles by this year certainly https://twitter.com/...
So much tech-solutionism depends on underestimating and undervaluing humans' abilities to solve difficult problems—which include things we think are easy, like driving a car. It should be no surprise when the “solution” hits a wall far short of the expectations raised. https://tw…
The strongest use case for self-driving car technology has been to replace public transit with a more expensive, less-safe alternative operated by private companies. So basically the benefit is: no pesky human that has to be paid to drive. https://twitter.com/...
“The industry says its Derek Zoolander problem applies only to lefts that require navigating oncoming traffic. (Great.) It's devoted enormous resources to figuring out left turns” https://www.bloomberg.com/...
NEW @BW COVER! Driverless Cars Are Going Nowhere: After all the years and billions spent, we're still not even close to an autonomous future. And detractors are getting louder as the losses get bigger, by @chafkin https://www.bloomberg.com/... https://twitter.com/...
6-7 years ago I had a talk with my boss at the time who said he didn't expect our kids to ever have to drive cars and I asked him if he'd ever actually worked at a software company. We were both working for a software company. https://twitter.com/...
love the lede here, where Waymo kept sending its “self-driving” cars to do three-point turns in the same woman's driveway but still insists it was working as intended. It's actually the future, it's good actually. https://www.bloomberg.com/...
Media has long been way, way too sympathetic to the claims of the driverless car industry. (I'd argue a similar thing is happening now with coverage of AI) But this whole piece by @chafkin really good. https://www.bloomberg.com/...
NEW COVER: Tech and auto giants have spent years hyping self-driving cars as a just around the corner innovation that would make our streets safer. Instead, they're increasingly looking like a $100 billion flop https://www.bloomberg.com/... https://twitter.com/...