SoftBank Group has committed $200 million to Gravis Robotics, a Zurich startup building physical artificial intelligence to run heavy machinery on construction sites. The capital is explicitly aimed at moving autonomous machines from controlled demonstrations into continuous operation in active work zones, according to reporting by CEO Insights Asia.
DIG Analysis — 2026-08-18: The size and profile of this round mark a clear stage shift. SoftBank is no longer testing the category with token bets. It is funding scaled engineering and pilot deployments that must solve reliability, safety and integration challenges within live construction projects.
What happened: the financing and the technical focus
Gravis Robotics raised the round to expand engineering teams, buy heavy test hardware and field multi-machine systems that can operate without a human in the cab. The company focuses on physical AI, meaning onboard perception, control and task planning that run on the vehicle rather than relying only on remote teleoperation. SoftBank led the round, a single anchor investor that brings appetite for multi-year capital intensity and operational support for rapid hardware iterations. CEO Insights Asia reported the investment details.
That technical choice matters. Running a hydraulic excavator or a paver requires real-time control across variable ground, changing loads and human crews nearby. Gravis is placing its bet on integrated stacks of sensors, model-based control and continuous learning to manage those edge cases. The new funds will go to prove those stacks under real working conditions where uptime and safety are non negotiable.
Why it matters to the construction sector now
Construction faces two acute problems that make this investment consequential. First, labor shortages are persistent in many developed markets, pushing contractors to accept higher labor costs and schedule risk. Second, jobsite safety remains a primary driver of costs and liability. Autonomy that reliably reduces human exposure to hazards and that can operate across long shifts could lower both direct labor spend and safety incidents, if systems meet regulatory and insurer requirements.

