SoftBank is betting $225 million that self-driving bulldozers can fix a labor shortage

SoftBank invested $225 million directly in Autonomous Solutions Inc. (ASI), a Utah-based company that builds self-driving technology for heavy equipment. Separately, the two companies formed a joint venture, also funded by SoftBank, focused specifically on bringing that technology to large-scale construction projects.

Why it matters: splitting the deal into a direct investment plus a dedicated joint venture signals SoftBank isn't just backing ASI as a company. It's betting on construction robotics as its own category worth building a separate business around.

ASI's Mobius platform coordinates mixed fleets of heavy equipment, haul trucks, dozers, loaders, and compactors, so they can work autonomously together on the same jobsite. The system is built to be OEM-agnostic, meaning it isn't locked to a single equipment brand.

• Haul trucks: move material across a site autonomously.
• Dozers: handle earthmoving tasks without a human operator.
• Loaders: work in coordination with trucks and dozers on the same schedule.
• Compactors: finish surface work as part of the same automated sequence.

Why it matters: most construction companies already own equipment from several different manufacturers. Software that works across brands means a company can automate the fleet it already has, instead of replacing everything to get autonomy.

The catch: a single self-driving dump truck is a neat trick, but a real jobsite needs dozens of machines doing different jobs in the right order, without colliding or blocking each other. That coordination problem is harder than automating any one vehicle on its own.

ASI CEO Mel Torrie put it plainly: "Skilled equipment operators are hard to find, and coordinating mixed fleets across a jobsite is a real bottleneck." He said the venture brings "autonomous, OEM-agnostic fleet coordination to scale."

In real life it's less like teaching one car to drive itself and more like choreographing an entire marching band, where every instrument comes from a different manufacturer and none of them were originally designed to play together.

The joint venture is targeting large infrastructure work specifically: roadway expansion, airport modernization, and rail infrastructure. These are projects that already require large fleets of heavy machinery working in tight coordination over long timelines.

Background: construction has struggled with a documented shortage of skilled equipment operators for years, a problem that predates this partnership and has only intensified as an aging workforce retires faster than new operators are trained. Large infrastructure projects, which need dozens of operators running different machine types simultaneously, feel that shortage most acutely.

Not quite. The technology automates specific heavy-equipment tasks, hauling, earthmoving, compacting, within construction projects that still rely on skilled trades, site managers, and human oversight for everything the machines aren't built to do.

Who's affected: this isn't about eliminating construction jobs broadly. It's aimed squarely at the equipment-operator bottleneck specifically, the roles companies say they already can't fill fast enough with human hires alone.

Background: SoftBank has a long history of large, concentrated bets on robotics and automation through its various investment vehicles, treating autonomous systems as core infrastructure for future labor markets rather than a niche technology. A $225 million direct investment, paired with a separately funded joint venture, fits that pattern of going deep rather than making a small, exploratory bet.

The financial details of the joint venture itself, including its total funding size and how ownership is split between SoftBank and ASI, were not disclosed alongside the direct investment figure.

Background: yes, repeatedly, and not always with a clean exit. SoftBank bought the walking-robot pioneer Boston Dynamics from Alphabet in 2017 for roughly $100 million, then added another $37 million in loans that later converted into equity as Boston Dynamics worked to turn decades of research into commercial products like its Spot robot.

SoftBank sold a controlling stake in Boston Dynamics to Hyundai in 2021, in a deal that valued the company at $1.1 billion, while keeping a smaller ownership stake for itself. As recently as July 2026, Hyundai moved to buy out SoftBank's remaining stake entirely, closing the loop on an eight-year robotics bet.

Why it matters: that history shows SoftBank treating robotics less like a single company to hold forever and more like a sector to seed, develop, and eventually pass off once the underlying technology matures into a real commercial product. The ASI deal, structured as a direct investment plus a separately funded joint venture, follows a similar shape: get in early, help build the commercial case, and figure out the long-term ownership picture later.

The catch: a lot of the hardest problems in autonomous vehicles come from sharing space with unpredictable humans, pedestrians stepping off curbs, other drivers making sudden turns, cyclists weaving through traffic. A construction site is a controlled environment by design, with clear boundaries, defined work zones, and far fewer people expected to be anywhere near heavy machinery in motion.

That doesn't make the engineering trivial. Mixed-fleet coordination across different manufacturers, in varying terrain, on active earthmoving sites, is still a genuinely hard problem. But it's a more contained one than teaching a vehicle to navigate an open city street, which is part of why autonomous mining trucks and now construction fleets have moved toward real commercial deployment faster than robotaxis reaching every city.

Mining companies have run autonomous haul trucks at scale for years in remote pits with tightly controlled access, and that track record is part of what makes investors comfortable betting real money on the same underlying approach spreading into construction.

• This is industrial automation, not a consumer product. You won't interact with Mobius directly, but it could show up in how fast the road, airport, or rail project near you actually gets built.
• It's a sign autonomous vehicle technology is spreading beyond warehouses, mines, and test tracks. Construction sites are becoming one of the more practical places to deploy self-driving equipment at scale.
• Watch for this to affect equipment-operator hiring, not general construction employment. The stated goal is filling a specific skilled-labor gap, not replacing construction workforces broadly.
• Large infrastructure projects (roads, airports, rail) are the near-term test case. If this works at that scale, expect similar mixed-fleet automation to spread to smaller commercial projects over time.