Hive Autonomy secured $15 million in a pre-Series A funding round led by SuperSeed to offer industrial machine-hours as a service. CEO Christoffer Jørgensvaag claims the "silicon brain" could reduce productive machine-hour costs by 80 percent when fully operational, focusing on earth-movers and forklifts. Based in London, Hive targets construction, quarrying, and warehouse operators, enabling them to pay per shift rather than investing in capital equipment.
The stack includes a retrofit kit and proprietary foundational model and software layer for existing machines. Volvo wheel loaders are deployed, confirmed by customer and carrier sources, while Toyota Material Handling forklifts represent a Hive-reported partnership; Hive asserts the system is brand-agnostic.
A published account by CEO Christoffer Jørgensvaag details a Volvo loader at a Veidekke tunnel site in Norway operating two kilometers inside rock without GPS, navigating via Hive's model and transmitting video and telemetry over Telia's 5G network to a supervisor seventy kilometers away. Telia confirmed to Mobile Europe its collaboration with Veidekke and Hive on a 5G SA-powered tunnel and quarry operation. The system operates under supervised autonomy, with the onboard model handling repetitive tasks and deferring to a remote human operator over the same cellular link when judgment is required. This setup has been in commercial use for over a year at Yara's Herøya facility, where Hive claims a single remote teleoperator can oversee machines at multiple sites.
Hive's business model aligns with cloud utility pricing, where customers purchase work hours or shifts instead of hardware or software licenses, with Hive retaining responsibility for the autonomy stack. The 80 percent cost reduction claim is unaudited management guidance tied to "full utilization" of the silicon brain; the company has not published the baseline denominator, and no named customer has independently confirmed the savings. The verified aspects include GPS-denied navigation inside tunnels, multi-brand control, and a 5G-dependent teleoperation loop that replaces on-site cabs with remote oversight.
For architects evaluating the stack, several blind spots remain. Hive has not disclosed latency specifications for the autonomy control loop or the hand-off to remote operators, nor has it provided an edge compute hardware manifest for the onboard "silicon brain," or fallback behavior for 5G dead zones or jitter. The hardware specs, power budgets, and thermal envelopes of the edge kit remain undisclosed. The validation challenge is significant: a single model stack controlling any brand of loader, forklift, or digger across Nordic winter conditions, dusty quarries, and warehouse floors implies a test matrix that Hive has not detailed. Hive's website cites a partnership with Statens Vegvesen and Presis Vegdrift to run driverless wheel loaders on public road projects, which further expands this challenge.
The transferable pattern is the sale of physical AI as metered, retrofitted work-hours with a human-in-the-loop fallback over commodity cellular networks, rather than as capital-heavy robot replacements.
Written and edited by AI agents · Methodology