Per-Vehicle Autonomy as Software Spend
Scott Sanders, Chief Growth Officer at Forterra, on autonomy for every vehicle
The real unlock in ground autonomy is not smarter models, it is getting the full kit cheap enough that buyers can treat autonomy like ordinary software spend instead of a capital project. Once the per vehicle stack looks more like an annual software budget than a custom robot program, autonomy stops being limited to billion dollar defense efforts and starts fitting yards, depots, and other high labor workflows where one more driver is already expensive.
-
Forterra is explicitly building for this crossover point. The company describes its product as a repeatable hardware and software kit that can move from a military truck to a yard tractor with the same core stack, which only works if sensors and edge compute are standardized and bought in volume.
-
This is the same playbook that made Anduril viable earlier. A startup can skip years of government funded development if it arrives with a working product that is meaningfully cheaper or better than the incumbent path. In defense and industrial autonomy, price matters because procurement friction is so high.
-
The markets that open first are the ones where labor is costly, environments are constrained, and customers can buy fleets in concentration. That is why Forterra keeps coming back to terminal tractors, container yards, and defense logistics, instead of broad on road autonomy where customer bases are fragmented and deployment costs stay high.
As LiDAR and edge compute keep getting cheaper, autonomy is likely to spread the same way enterprise software did, first into the workflows where the ROI is easiest to prove, then outward into adjacent fleets. The winners will be the companies that package autonomy as a low cost product, not a bespoke science project.