AI Is Taking Over Excavators: SoftBank Bets $200 Million on Autonomous Construction Equipment
The construction equipment industry is entering a new phase as artificial intelligence moves from software applications into the machines that physically build roads, foundations, quarries and infrastructure. Swiss construction robotics company Gravis Robotics has raised $200 million in Series A funding from SoftBank, marking one of the biggest investments yet in autonomous construction equipment. The company says the funding will accelerate the global deployment of its autonomous earthmoving technology.
Gravis Robotics, which originated as a spinout from ETH Zurich in 2022, is developing technology that can retrofit existing construction machines rather than requiring contractors to purchase completely new autonomous equipment. Its core system, called the Gravis RACK, combines sensors, computing hardware and software to give excavators and other earthmoving machines autonomous and machine-assistance capabilities. This approach is significant because construction companies already operate large fleets of expensive equipment, meaning autonomy could potentially be introduced by upgrading machines already in service.
The Gravis RACK uses technologies including LiDAR, cameras, GNSS and onboard computing to understand the machine's surroundings and the terrain around it. The system can create real-time three-dimensional information about the work area and support tasks such as excavation, grading and truck loading. Gravis says its terrain-aware excavation technology can increase throughput by up to 30% in appropriate applications, while also improving consistency and jobsite awareness.
The technology is not simply designed to remove the operator from the cab. Gravis has developed a model that combines human supervision with increasing levels of machine autonomy. Its systems can provide operators with additional information and automated assistance, while predefined repetitive tasks can be performed autonomously. This distinction is important because construction sites are highly variable environments where unexpected ground conditions, people, utilities and other equipment can require human judgement.
A demonstration by Hitachi Construction Machinery earlier in 2026 showed how this technology can work in practice. A 13-tonne Hitachi ZX135US-7 excavator equipped with the Gravis RACK was demonstrated moving from conventional operation to autonomous execution of repetitive trenching work. According to Hitachi, an operator could train the machine using site information and then allow it to execute a predetermined excavation path without continuously touching the machine's controls.
The technology also has an operator-assistance component. Gravis Copilot can provide real-time terrain information, 3D surveying and mapped utility visualization. The system is designed to give operators a better understanding of the work area from inside the cab, potentially reducing the need for manual measurements and improving precision during excavation and grading. Hitachi has also highlighted a safety function that can detect people entering designated work zones and alert the operator.
The business case for autonomous construction equipment is particularly strong around repetitive work. Excavation often involves repeating the same movement cycle thousands of times: digging, swinging, loading and returning the bucket. When the work is predictable and well defined, software can potentially automate portions of that cycle while allowing a human operator or supervisor to intervene when necessary. Gravis says its technology is already being applied to areas such as site preparation, stockpile management and loading operations.
Gravis has also been building partnerships across the construction equipment ecosystem. Its technology has been associated with equipment manufacturers, contractors and plant-hire businesses, including work with companies such as Hitachi Construction Machinery, HD Hyundai, Develon, Holcim and Taylor Woodrow. In 2025, Gravis announced deployments across several countries and described its technology as being used in real construction and quarry environments.
The latest SoftBank investment could therefore represent more than a funding event for one robotics company. It signals growing investor confidence in what is increasingly being called physical AI: artificial intelligence that interacts with the physical world through machines and robots. Unlike conventional generative AI, physical AI has to understand real-world environments, movement, physics and safety before it can perform useful tasks.
For equipment owners, the retrofit approach could become particularly important. Instead of replacing an entire fleet with autonomous machines, contractors may eventually be able to select suitable excavators and add autonomous hardware and software. This could create a new equipment technology category sitting between conventional machinery and fully autonomous robots.
The impact on construction operators is also likely to evolve rather than happen overnight. Autonomous systems could reduce the amount of manual control required for repetitive tasks, while creating greater demand for workers who can supervise machines, manage digital jobsite systems, troubleshoot sensors and maintain robotic equipment. In this model, the role of the operator could gradually shift from controlling one machine continuously to supervising multiple intelligent machines.
For construction equipment rental companies, dealers and fleet owners, the development could eventually create another opportunity: autonomous equipment as a service. A machine equipped with an autonomy package could potentially be offered with technology support, remote supervision and data services alongside the traditional rental model. Gravis has already highlighted partnerships in the plant-hire and equipment distribution ecosystem, showing that autonomy is beginning to enter the commercial equipment market rather than remaining purely a laboratory technology.
However, autonomous excavation still faces practical challenges. Construction sites are less predictable than controlled industrial environments, and machines must operate safely around workers, vehicles, underground utilities and changing terrain. Regulatory requirements, connectivity, operator training, cybersecurity, machine compatibility and the economics of retrofitting will all influence how quickly the technology scales.
The $200 million SoftBank investment nevertheless represents a major signal for the heavy construction equipment industry. The future may not be about replacing the excavator with a completely different machine. Instead, the excavator itself could become smarter through sensors, AI, connectivity and autonomous control.
For contractors and equipment owners, the question is increasingly shifting from “Can an excavator work autonomously?” to “Which construction tasks should be automated, and how can humans and intelligent machines work together?”
As AI moves deeper into heavy equipment, the construction site of the future could have fewer machines operated manually from start to finish and more machines working under intelligent supervision. The excavator may still look familiar, but its capabilities could be fundamentally different.