Profile of Starsky Robotics, which hires truckers to train and eventually remotely control semi-autonomous trucks
Just before Stefan Seltz-Axmacher offers a job to an engineer at Starsky Robotics Inc., a driverless trucking startup in San Francisco, he gives them the talk. Tweets: @aexm , @planetmoney , @chafkin , @aaronrutkoff , @meganmurp , @josheidelson , and @wooglooskr Tweets: Ana Milicevic / @aexm : “We basically have people from two worlds, neither of which has ever talked to each other” http://www.bloomberg.com/... NPR's Planet Money / @planetmoney : The onslaught of automation is leading to some strange truckfellows. Fascinating story from @business http://www.bloomberg.com/... Max Chafkin / @chafkin : me and @josheidelson teamed up to look at AI and trucking, a job that we romanticize but pretty much sucks today https://www.bloomberg.com/... pic.twitter.com/O8eKz4Ymi7 Aaron Rutkoff / @aaronrutkoff : This @BW story is a sitcom premise: Startup hires truckers and coders to do long-haul routes in the same cab. https://www.bloomberg.com/... Megan Murphy / @meganmurp : The near-term future of driverless trucking is a fascinating mash-up of Silicon Valley and Iowa 80. Worth your time https://www.bloomberg.com/... Josh Eidelson / @josheidelson : .@Chafkin & I wrote a @BW feature about how automation could transform trucking (and how deregulation already did): https://www.bloomberg.com/... Paul Milovanov / @wooglooskr : Three and a half million Americans drive trucks for a living. The next decade will be anything but boring. https://www.bloomberg.com/... pic.twitter.com/mIh2mIHLD9
Context & Ripple Effects
This 2017 Bloomberg profile captures Starsky Robotics at the start of the autonomous-trucking wave, betting on a hybrid model: hire working truckers to train the system first, then keep them on as remote operators who take over when the semi-autonomous truck hands control back. That human-in-the-loop design put Starsky at odds with where the money later went — rivals like TuSimple, valued at $1.1B by 2019, and Sequoia-backed PlusAI pursued full autonomy with no driver role at all.
The arc since then has been rough: investors poured an estimated ~$100B into self-driving vehicle startups with little progress to show, and Aurora and Kodiak are still planning Texas expansion against a lack of federal regulation. Starsky's early wager — that a human stays in the loop — looks less quaint in that light.
First-order effects
- Truckers hired by Starsky get a new job description — training the autonomy stack and standing by to remotely control trucks — rather than being displaced outright, blending two workforces the company itself says have never talked to each other.
- Starsky must fund both a trucking-operations payroll and an engineering team simultaneously, a cost structure pure-autonomy rivals avoid.
Second-order effects
- Full-autonomy competitors like TuSimple and PlusAI can pitch investors a no-driver end state, forcing Starsky to defend why teleoperators beat removing humans entirely — a debate the funding market ultimately answered by backing the fully autonomous camp.
- The remote-operator model creates a global labor arbitrage question: Kiwibots already runs 'parallel autonomy' delivery bots piloted by workers in Colombia paid $2/hour, a template that pressures US teleoperator wages downward.
Third-order effects
- If the hybrid model survives, long-haul trucking splits into two structures — fully autonomous fleets versus supervised ones with distributed remote drivers — and regulators, currently absent per the industry's own complaints, will have to decide whether a remote operator counts as a licensed driver.
- The broader lesson from the ~$100B bet is that the hardest problem in freight automation may be economics, not perception: whichever model closes the cost gap with human-driven trucking first dictates how millions of driving jobs are restructured.
The trend: Autonomous trucking keeps oscillating between eliminating the driver and re-employing them as a remote supervisor, with Starsky's 2017 teleoperations model an early data point in that unresolved fight.