Waypoints as a scaling tool: lessons from Aurora's Chris Urmson

Executive overview

Scaling a company is a long expedition — doing it without intermediate goals leads to drift, waste, and failed missions. Waypoints are short, flexible, achievable targets that keep you on course without locking you into a rigid path.

Reid Hoffman interviews Chris Urmson, co-founder and CEO of Aurora, who has used waypoint thinking from early autonomous robot experiments to building a self-driving trucking company. His journey — from Carnegie Mellon to Google X to founding Aurora — is a case study in how to set, adapt, and communicate waypoints across technology, business, and team integration.

The core insight: waypoints aren't about eliminating risk — they're about maximizing speed of learning while staying pointed at the destination.

From robot crawler to Google to Aurora

• Carnegie Mellon robot in the Atacama Desert drove 15–30 cm/sec; waypoints were measured in centimetres but built foundational capability
• DARPA Grand Challenge first attempt failed at 7.5 miles of 150; re-entered two years later and won — attracted Google's attention
• At Google, Larry and Sergey set three waypoints: 50k miles, 100k miles, then 1,000 "interesting" miles covering diverse road types
• The first two built data volume; the third tested breadth and prevented a narrowly-solved problem
• Pre-qualifying on Highway 101 (10 passes) before attempting Pacific Coast Highway — a shorter but harder waypoint to de-risk a scarier one
• Left Google after 7.5 years when he lost confidence in the business-side waypoints; applied the "fix it, get in line, or get out of the way" framework

Founding Aurora and choosing the right first market

• Founded Aurora in 2017 with Sterling Anderson (ex-Tesla Autopilot) and Drew Bagnell (ex-Uber autonomy)
• Chose to focus on one miracle at a time: build the self-driving system, not also build the car or the ride-hail platform
• In 2020, pivoted focus to autonomous trucks: trucking is a $700B market vs. $35B for ride-hailing; routes are more predictable; economics are stronger
• This was not a pivot — passenger vehicles remained a waypoint, just a more distant one
• Architected the system to handle both trucks and cars from the start, enabling the later move back to ride-hail

Building the technology: sensor fusion and simulation

• Aurora uses cameras, lidar, and radar in combination ("sensor fusion") rather than cameras only
• Each sensor type compensates for others' weaknesses: color detail, 3D distance measurement, all-weather penetration, velocity sensing
• Adding sensor types costs time and complexity but is the right long-term call
• Real-world driving alone cannot prove safety at scale (one fatality per 85M miles makes pure on-road testing impractical)
• Invested in advanced simulation to test edge cases that rarely occur in reality
• Adopted deep learning and moved from monolithic HD maps to sharded maps that update faster

Merging with Uber ATG: waypoint alignment in acquisitions

• Aurora absorbed Uber's autonomous driving unit (ATG) in 2020
• Integration approach: both teams wrote independent six-pagers on the same problem, shared learnings, and identified overlap before declaring any technology winner
• First priority stated clearly: ship a trucking product; second: ride-hail product
• Mistake made: created artificially parallel leadership structures out of respect for both sides — leaders clashed, and in some cases both were lost
• Lesson: prioritise simplicity and clarity in org structure over political symmetry

Waypoints as team motivation and stakeholder communication

• Long-horizon problems deprive teams of the "dopamine hit" of shipping
• Create discrete internal milestones to celebrate incremental progress — treat it like celebrating each mile of a marathon, not just the finish line
• Clearly signposting waypoints to customers matters too: TaskRabbit's Stacey Brown Philpott announced a major platform overhaul to the press and customers simultaneously — taskers revolted; advance notice would have prevented the backlash
• Waypoints also serve external audiences: clear, achievable steps gave investors like Reid Hoffman a "line of sight" on how the mission would be achieved, helping secure Series A funding

Aurora's roadmap for autonomous trucking and beyond

• Initial focus: long-haul highway driving between terminals
• Next: terminal-to-depot routes
• Ultimately: warehouse-to-store last-mile delivery
• Safety improvements: diverse sensors eliminate distraction/fatigue; trucks run at 65 mph instead of 75 mph for a 25% fuel efficiency gain; Houston-to-LA shipping time cut from two days to one
• Truck driver shortage (80k today, projected 160k by decade end) means displacement will be softer than feared; new jobs will be created
• Longer-term waypoint: smaller, on-demand transit vehicles replacing fixed-route buses, improving accessibility at the lower end of the economic spectrum