SoftBank invests $200M in Gravis construction robotics Series A

The construction industry has historically resisted automation, but converging pressures—infrastructure repair demands, AI data center expansion, U.S. reshoring initiatives, and defense needs—are creating an opening for autonomous solutions. Gravis positions itself to address a fundamental bottleneck: earthmoving work, which the company notes requires not simplification of the world but AI systems equipped with "physical intuition" to handle real jobsites. The core challenge that Gravis solves is the sim-to-real gap—the difficulty of translating simulated training into performance on unpredictable worksites where soil mechanics, machine vibration, and hydraulic resistance vary by the microsecond.

Gravis's competitive advantage lies in its equipment-agnostic approach. Rather than locking contractors into a single manufacturer's ecosystem, the Gravis Rack can be retrofitted onto excavators and heavy machinery from multiple vendors. This matters because the heavy equipment market is fragmented—roughly two-thirds of global demand sits outside the top three manufacturers—and contractors typically choose equipment based on regional service relationships and existing fleet investments. By supporting mixed fleets, Gravis avoids forcing a difficult transition and instead positions itself as a software layer that works with what contractors already own.

The $200 million Series A from SoftBank validates that physical AI—the application of machine learning to real-world physical tasks—has become a strategic priority for major investors. SoftBank's backing signals confidence in both the market opportunity (the company estimates the addressable market at a trillion dollars, bottlenecked by labor shortages) and Gravis's technical approach. Early validation comes from live deployments across four continents and the selection to lead a U.K. government-backed autonomous machinery project, which also demonstrates emerging policy support for construction automation.