Gatic CEO says long-haul driverless trucks remain decades away
- Gatic targets regional middle-mile logistics while long-haul autonomous trucking faces decades of delays.
- Fixed-fee contracts shift fleet risk to corporate clients like PepsiCo and Kroger.
- Deterministic safety layers override non-deterministic AI models on autonomous trucks.
Long-haul driverless trucks are a distant dream.
While rivals spent billions chasing coast-to-coast automated freight, Gatic CEO Gautam Narang targeted regional middle-mile routes. The traditional long-haul play required massive transfer hubs outside major cities to swap human drivers for autonomous tractors. High terminal construction costs and expensive first- and last-mile transfers choked those economics. Gatic bypassed that overhead by running direct dock-to-dock deliveries between distribution centers and retail stores.
The shift relies on strict contract mechanics. As detailed on FYI - For Your Innovation, Gatic charges corporate shippers like PepsiCo and Kroger a fixed fee of roughly $200,000 per truck annually. Under five-year non-cancellable contracts covering 12 hours of daily operation, Gatic gets paid regardless of whether the client runs the vehicle. That shifts fleet utilization risk to corporate customers while insulating Gatic’s balance sheet.
Hardware partnerships provide the physical backbone for this commercial push. Isuzu Motors invested directly in Gatic and is constructing a South Carolina manufacturing plant capable of producing up to 50,000 L4-ready trucks per year by 2030, with vehicle integration slated for 2028.
Software adoption, not hardware perception, remains the real industry bottleneck. On This Week in AI, host Jason Calacanis argued that public safety debates will evaporate within 24 months as safety data proves machines outperform human drivers. Narang countered that physical AI follows a far slower curve than software because society holds autonomous hardware to far stricter standards than human error.
To satisfy those standards, Gatic eschews pure end-to-end neural networks. Its software relies on a hybrid architecture where neural networks handle scene perception while a deterministic heuristic layer overrides non-deterministic outputs before execution. Safety-critical decisions execute locally on truck edge hardware without cloud latency, reflecting how edge compute offers operational predictability that central cloud servers cannot match.
The future of autonomous freight belongs to predictable, short-haul corporate corridors.