Hoxo enters the nuclear sector: Is Capgemini redefining physical AI for high-risk industries?

Capgemini, the global technology and business transformation group, and Orano, the nuclear materials recovery and transformation specialist, have deployed Hoxo, the first intelligent humanoid robot designed for nuclear operational environments. The robot is now undergoing a structured four-month evaluation phase at the Orano Melox Ecole des Métiers in the Gard region of France. This milestone represents a shift in the integration of artificial intelligence into physical systems tailored for highly regulated and hazardous industries.

Hoxo is not a conceptual showcase but a real-world industrial platform equipped with artificial intelligence, embedded sensors, autonomous navigation, and human-gesture replication capabilities. It has been developed to support routine and precision tasks that typically involve physical strain or require repetitive execution within nuclear facilities. Orano’s decision to integrate humanoid robotics into its core innovation pipeline reflects broader trends across the European nuclear sector, where the need for next-generation automation is accelerating alongside renewed investment in low-carbon energy infrastructure.

How does Hoxo improve upon traditional robotics used in nuclear operations?

Traditional robotics in the nuclear industry have focused primarily on fixed-location automation or remote-controlled manipulators. These systems are typically confined to controlled enclosures and are highly task-specific. Hoxo, by contrast, introduces mobility, perception, and adaptive learning into the equation. The humanoid design enables the robot to operate in spaces originally designed for human technicians without requiring architectural changes to the facility.

Its real-time perception capabilities are powered by advanced computer vision and multimodal sensors, allowing Hoxo to navigate the environment, recognize tools, and interact with people. Capgemini’s AI Robotics and Experiences Lab has led the development of this platform, ensuring that artificial intelligence is embedded not just at the software level but across physical embodiment, task planning, and environmental simulation layers.

Digital twin technology is a core feature of the system, enabling operators to simulate Hoxo’s tasks before actual deployment, reducing risk while improving operational planning. This functionality is critical in nuclear environments, where errors can have safety and regulatory consequences.

Why is the nuclear sector an ideal testing ground for physical artificial intelligence?

The nuclear industry requires precise, repeatable actions conducted under strict regulatory controls. Operations such as fuel processing, waste handling, decontamination, and equipment maintenance demand high levels of technical expertise, concentration, and adherence to safety protocols. Hoxo is being tested in this environment because of its ability to deliver consistent, programmable performance without succumbing to fatigue or error under routine strain.

Moreover, the nuclear sector is undergoing a demographic shift. Many facilities face aging workforces, and recruitment of young professionals into nuclear careers has become a challenge in several regions. Robotic augmentation provides a way to extend workforce capacity without compromising quality, safety, or procedural integrity.

Orano’s leadership views Hoxo as a complementary solution that enhances technician efficiency, particularly in environments that expose personnel to ergonomic fatigue or radiation thresholds. According to Arnaud Capdepon, Director of Orano Melox, the initiative is designed to evolve alongside industrial requirements, contributing directly to operational competitiveness and safety.

What role does Capgemini’s physical AI platform play in this collaboration?

Capgemini has increasingly moved from software-dominated artificial intelligence services into physical AI and robotics integration. The company’s AI Robotics and Experiences Lab has been at the forefront of combining machine intelligence with robotic embodiment, using digital twin environments and real-time analytics to enable responsive, adaptable machines.

In the case of Hoxo, Capgemini’s physical AI stack merges mobility algorithms, environmental simulation, and human-machine interface design. The goal is not only to automate a task but to create a machine that can learn from feedback, operate in unstructured spaces, and collaborate with humans in complex workflows.

Pascal Brier, Capgemini’s Chief Innovation Officer, noted that Hoxo’s deployment reflects the convergence of AI, robotics, and human ergonomics, extending the concept of industrial automation beyond traditional robotic arms and into intelligent, human-like assistants capable of real-world interaction and environmental awareness.

How could humanoid robots change the future of nuclear workforce operations?

The introduction of intelligent humanoid robots like Hoxo may have a cascading effect on how nuclear facilities are staffed, structured, and maintained. Rather than replacing nuclear technicians, the model leans toward cooperative robotics, where machines take on routine or high-risk tasks while personnel focus on oversight, exception handling, and advanced diagnostics.

Over time, training programs may shift toward skills in robotics management, AI interaction, predictive analytics, and digital twin simulation. This transformation aligns with the broader move toward Industry 5.0, where human-centric design, sustainability, and machine intelligence converge.

In terms of economic value, robotic assistance could reduce unplanned downtime, lower risk exposure costs, and optimize task scheduling in long-life nuclear infrastructure. As maintenance cycles and safety inspections become more demanding with age, Hoxo-style systems could enable operators to meet regulatory benchmarks more efficiently.

What future opportunities could emerge if Hoxo’s pilot phase proves successful?

If the ongoing trial validates Hoxo’s performance across safety, task execution, and human collaboration metrics, Orano and Capgemini could scale the deployment into operational reactor facilities, decommissioning zones, or waste handling units. Hoxo could also be modularized for use in other industries with similar high-precision, high-risk environments.

Aerospace assembly lines, pharmaceutical cleanrooms, semiconductor fabs, and offshore energy platforms represent logical extensions. In each case, the core value proposition of physical AI — reducing risk while enhancing productivity — remains relevant. Capgemini’s strategic positioning as a transformation partner across industries gives it a clear path for cross-sector adaptation.

This initiative also aligns with French and European industrial strategies around technological sovereignty, industrial resilience, and energy transition. By showcasing leadership in robotic intelligence within a safety-critical environment, Capgemini and Orano are positioning themselves at the frontier of industrial innovation.

What are the key takeaways from Capgemini and Orano’s deployment of the humanoid robot Hoxo?

• Capgemini and Orano have introduced Hoxo as the first intelligent humanoid robot designed specifically for nuclear environments, marking a notable advance in the application of physical artificial intelligence within high-security industrial settings.

• The robot is currently undergoing a four-month testing phase at the Orano Melox Ecole des Métiers facility in the Gard region of France, where its mobility, precision, stability, and real-time decision capabilities will be closely evaluated in realistic operational workflows.

• Hoxo stands apart from traditional industrial robots because it is built to interact directly with human teams, adapt to changing spatial conditions, and replicate human gestures while navigating confined or safety-restricted zones.

• The development is led by Capgemini’s AI Robotics and Experiences Lab, which integrates artificial intelligence, computer vision, digital twins, and robotic systems to support complex, high-risk industrial operations.

• Orano positions the robot as an augmentation tool that can support technicians in tasks requiring endurance or operating conditions that may lead to operator fatigue, while maintaining adherence to nuclear safety and regulatory standards.

• The initiative arrives at a strategic moment when France and several European countries are reaffirming nuclear energy’s role in long-term low-carbon energy security, increasing the importance of operational efficiency and workforce sustainability.

• Capgemini interprets the deployment as an example of how human-machine collaboration is evolving from automated machinery to intelligent, context-aware robotic partners in sensitive industrial sectors.

• If the trial proves successful, the same physical AI platform could be adapted for sectors such as aerospace, pharmaceuticals, energy infrastructure maintenance, and semiconductor fabrication, all of which require precision in controlled environments.

• The introduction of humanoid robotics in nuclear facilities is expected to shape future workforce training, with human roles shifting further toward supervision, programming, data interpretation, and process planning supported by robotic assistance.