SoftBank has invested $200 million [1] in Gravis Robotics to develop autonomous systems for heavy construction equipment.
This investment marks a significant push into physical AI, aiming to automate high-risk and labor-intensive infrastructure projects. By retrofitting existing machinery, the technology could transform how roads, quarries, and mines are developed.
Gravis Robotics, a spin-out from ETH Zurich in Switzerland, secured the funding in a Series A round led by SoftBank [3]. The $200 million [2] injection is reported as the largest Series A ever recorded in the construction robotics sector [3].
CEO and co-founder Ryan Luke Johns said the company is focusing on giving excavators a brain, enabling them to work autonomously on construction sites.
The technology targets the automation of earth-moving equipment. This includes machinery used for large-scale infrastructure, such as the creation of new roadways, and the operation of mining sites.
Johns said the funding reflects SoftBank's strong confidence in the physical AI space.
SoftBank's strategy focuses on the integration of artificial intelligence into physical hardware. This move follows a broader trend of investing in robotics that can navigate and manipulate complex, real-world environments—unlike traditional AI that exists primarily in digital spaces.
The Swiss startup intends to use the capital to accelerate the development of its autonomous retrofitting kits. These kits allow standard construction equipment to operate without a human driver in the cab, potentially increasing safety and efficiency on industrial sites.
“We’re giving excavators a brain, enabling them to work autonomously on construction sites.”
The scale of this investment suggests a shift in AI deployment from generative software to heavy industry. By targeting the 'physical AI' sector, SoftBank is betting that the next major leap in productivity will come from automating the physical labor of infrastructure, which has historically been slower to digitize than office work or logistics.
