Einride plans to test Nvidia’s Alpamayo reasoning model within its Einride Driver autonomous trucking system, with a live road evaluation scheduled for the second half of 2027. The Swedish freight technology company aims to become the first freight operator to run the software on Nvidia’s Hyperion platform.
The initiative builds on a partnership between the companies agreed in September 2026. It will focus on adapting and evaluating Alpamayo for autonomous freight operations without giving the model direct control over the truck.
Alpamayo to Operate in Shadow Mode
Nvidia’s Alpamayo is designed to interpret driving scenes, propose vehicle trajectories and explain the reasoning behind its proposed actions. Einride plans to post-train and distil the model using data collected from its own freight operations.
The adapted model will initially run in shadow mode, generating driving recommendations that can be compared with decisions from Einride’s existing planning system and human drivers. It will not control the truck during this evaluation.
Einride’s safety architecture will retain authority over vehicle operation while the company assesses the model’s behaviour in real-world freight scenarios. This approach allows the company to examine the system’s performance without relying on it to make driving decisions.
Freight Data to Support Model Adaptation
The use of operational freight data is intended to adapt Alpamayo to the routes and conditions encountered by Einride’s autonomous trucking system. The company expects this process to help refine the model for highway and point-to-point freight operations.
Einride aims to use the adapted model to reduce the development effort and cost associated with expanding autonomous operations across additional routes and vehicle types. The approach is intended to complement its existing autonomous driving system rather than replace it.
The planned evaluation will also give the company an opportunity to compare the model’s proposed trajectories and reasoning with the outputs of its established planner and the decisions made by human drivers.
Einride and Nvidia Expand Autonomous Freight Work
The project extends the companies’ collaboration around Nvidia’s Hyperion platform, which provides a reference architecture for autonomous driving technology.
Einride develops and operates autonomous freight systems alongside its wider freight technology business. Testing a reasoning model within its existing system will allow the company to assess how the technology could contribute to future development while maintaining its current safety controls.
Henrik Green, Chief Technology Officer at Einride, said:
“Our advantage has always been running autonomous freight in the real world, for real customers. Alpamayo brings valuable reasoning capabilities we can build on as we continue to design, build and operate the Einride Driver and our safety architecture. Being the first trucking company in this engagement gives us the opportunity to help shape the model for freight.”
Road Evaluation Planned for 2027
Einride expects to conduct a live road evaluation in the second half of 2027. Before then, the company plans to adapt the model using its freight data and assess its outputs in shadow mode.
The programme is intended to establish how a reasoning-based AI model can contribute to autonomous trucking development while remaining separate from direct vehicle control during testing. The extent to which it improves system performance, development costs or deployment timelines will depend on the results of the evaluation.