Autonocast #357: Why AV's Need Real World Data w/Nexar's Zach Greenberger
4 snips
Feb 19, 2026 Zach Greenberger, CEO of Nexar — an edge-AI leader building massive real-world driving datasets — explains why real-world data is essential for autonomous vehicles. He describes Nexar’s scale, how rare edge cases are flagged and enriched, differences between fleet and crowdsourced data, and how simulation and multimodal world models complement real driving data.
AI Snips
Chapters
Transcript
Episode notes
Rideshare Drivers Produce High-Value Data
- Rideshare drivers form Nexar's primary user base because they need protection and produce rugged, varied data.
- Small fleets also adopt video telematics as a lower-cost safety/compliance option than full sensor suites.
Real Data Is The Long Pole For Edge Cases
- Real-world data becomes critical once AVs move from simulation to fixing edge cases beyond ~90% coverage.
- Nexar partners with AVs (Waymo, Lyft, Uber) to supply that real-world edge-case data at scale.
Big Fleets Face Hard Data-Scale Math
- Uber building an internal AB Labs acknowledges data's value but collecting enough rich data at scale is operationally slow and costly.
- Partnering with aggregated networks like Nexar may be faster than solo sensor installs on fleets.