Gravis Raises $200M to Turn Mixed Heavy-Equipment Fleets Into Autonomous Systems
Gravis Robotics raised a $200 million Series A from SoftBank at a reported $1 billion post-money valuation, backing a retrofit-first autonomy model for excavators and mixed construction fleets.
What changed
Gravis Robotics announced a $200 million Series A investment from SoftBank Investment Advisers at a reported $1 billion post-money valuation.
The ETH Zurich spinout builds autonomy for construction machinery. Its core product is the Gravis Rack, a retrofit control and sensing kit designed to work across excavator brands rather than requiring customers to replace their fleets with a vertically integrated machine.
The company says its systems have been deployed with infrastructure customers across four continents and support equipment from manufacturers including Caterpillar, Case, Develon, John Deere, JCB, Hitachi, Sumitomo, Yanmar, and Volvo. These deployment and performance claims are company-reported.
Why it matters
Construction is a useful test for Physical AI because the environment changes as the robot acts.
An excavator does not navigate through a fixed scene. It breaks and moves material, encounters hidden rocks, changes the geometry of the worksite, and must respond to hydraulic resistance and machine-specific dynamics. That makes the problem materially different from structured warehouse transport or fixed-station manipulation.
Gravis is making two structural bets:
- retrofit beats replacement in a fragmented, capital-intensive equipment market; and
- machine telemetry plus simulation can teach a shared autonomy layer to generalize across equipment brands and sizes.
The first bet addresses customer economics. Contractors already own machines and depend on local dealer and service relationships. An autonomy kit can preserve that installed base while adding guidance, hazard detection, remote supervision, or full autonomy.
The second bet addresses technical scale. If one software stack can adapt across mixed fleets, the company can potentially learn from a broader range of machines and jobsites than a single-OEM system.
Deployment read-through
The financing is notable because it targets a workflow with direct labor, safety, and schedule economics.
Gravis offers a spectrum from AI-assisted manual control to full autonomy. That is strategically sensible: construction customers may adopt guidance and hazard detection before trusting unattended earthmoving. It also gives the company an incremental deployment path rather than requiring a site to jump immediately to lights-out operation.
The commercial proof burden remains high. Gravis cites up to a 30% productivity improvement versus peak manual operation, but this is a vendor claim and the public evidence does not yet disclose the task mix, site conditions, baseline definition, fleet size, or duration behind the figure.
The most important question is whether the retrofit remains reliable across real differences in hydraulics, wear, payload, soil, weather, maintenance, and operator practice. Supporting many brands in principle is not the same as delivering the same uptime and precision across all of them.
What to watch next
- paid fleet deployments and repeat orders by contractor
- hours of autonomous or AI-assisted operation in production
- productivity measured by task and site rather than a single headline percentage
- intervention rate, safety incidents, and near-miss reporting
- installation time and calibration burden per machine model
- performance across soil types, weather, machine age, and equipment brands
- whether operators supervise multiple machines without creating new bottlenecks
- service, warranty, and field-support economics
Interpretation
Gravis represents a form of Physical AI that is less visible than humanoid demos but potentially closer to measurable customer value.
The company is not asking construction firms to buy an entirely new robot category. It is trying to convert existing capital equipment into a software-defined fleet.
That makes the retrofit and integration layer strategically important. In fragmented industrial markets, the winner may not be the company with the most elegant standalone machine. It may be the company that can work with the equipment, workflows, service networks, and safety practices customers already have.
The $200 million round gives Gravis capital to scale that thesis. It does not prove the thesis. The next evidence must come from sustained jobsite performance, installation economics, repeat deployments, and independent safety and productivity data.
Not investment advice. Research notes only.