AI Optimizing France’s Nuclear Future: Metroscope Prevents Energy Loss at Scale
Nuclear power is the backbone of France’s energy system, supplying over 75% of the country’s electricity. In a world racing toward carbon neutrality, ensuring nuclear plants run at peak efficiency is more critical than ever - not just for economic reasons, but also for environmental sustainability.
Enter: Metroscope, a startup founded in 2018 by three former EDF R&D engineers: Aurélien Schwartz, Julien Lagarde, and Joel Bentolila. The company has built an AI-powered diagnostics platform that is already monitoring 95 plants across 12 countries, including nuclear, conventional, and cooling systems. Their digital twin technology detects inefficiencies in real-time, helping operators recover lost energy, prevent costly shutdowns, and improve plant performance.
With AI causing a global surge in energy demand, CEO Schwartz said the stakes couldn’t be higher: “Every power plant in the world loses, on average, 7 megawatts of energy. That’s enough to power an entire city. Our mission is to find those losses, fix them, and put that energy back into the grid.”
The Problem: Hidden Energy Losses in Power Plants
A modern nuclear power plant is a marvel of engineering, but its complexity makes inefficiencies inevitable. Over time, tiny operational issues—clogged condensers, faulty sensors, or microscopic leaks—chip away at performance. Without early detection, these inefficiencies can lead to huge energy losses and, in the worst cases, unexpected shutdowns.
One critical area where problems occur is the steam cycle, the system responsible for converting nuclear energy into electricity.
In a nuclear reactor, energy is released through fission, the process of splitting atomic nuclei. This energy heats water or gas, generating steam that drives turbines to produce electricity - much like in conventional fossil fuel plants. Once the steam has done its job, it must be cooled and condensed back into water so the cycle can continue.
That’s where the condenser comes in. In boiling water reactors, it recirculates water back to the heat source. In pressurized water reactors, it cools and filters the water before releasing it back into the environment. Because condensers often rely on natural water sources, they are vulnerable to external disruptions.
“Take mussels, for example,” says Schwartz. “They can enter the condenser, block cooling pipes, and reduce efficiency without anyone realizing. You can’t physically see inside these systems, and by the time an operator detects a problem, significant energy has already been wasted.”
The Solution: AI + Digital Twins for Predictive Diagnostics
Metroscope’s AI-driven digital twin technology creates a virtual replica of the power plant, mirroring every sensor reading, valve movement, and thermal fluctuation in real-time.
When something deviates from the expected performance, even slightly, Metroscope’s system flags the anomaly before it escalates. False alarms are a major problem in industrial AI, which is why Metroscope has built Infra Engine, an advanced algorithm that calculates the probability of an actual fault.
“Our AI isn’t just saying ‘Something is wrong,’” Schwartz explains. “It tells the operator exactly what’s wrong and what the likely cause might be. We provide transparency and trust, so decisions can be made with confidence.”
According to Schwartz, Metroscope's solution means a 5x reduction in plant failures and 3x faster detection of energy losses, all with a nearly 90% accuracy rate in diagnostics.
From Startup to Global Player
Since its founding in 2018, Metroscope has rapidly scaled from a niche AI startup to an international leader in power plant diagnostics. Originally spun out of EDF’s R&D division, the company secured €4 million in initial investment from EDF to develop its AI-powered diagnostic technology.
That cash allowed it to launch operations and support the interests of the nuclear division in France. Since then, the company has expanded internationally, acquiring customers across the globe.
Today, Metroscope has expanded beyond France into the United States, the UK, Belgium, Greece, South Africa, and Chile. Its headquarters in Paris is complemented by a growing team in Washington, D.C., supporting its expansion into the U.S. energy market.
The company has grown to 65 employees and continues to invest in R&D to enhance its digital twin technology. With its technology already detecting energy losses equivalent to the power consumption of a large city, Metroscope is positioning itself as a key player in the global energy transition.
Why It Matters: The Future of Energy Efficiency
As the world moves towards decarbonization, nuclear energy remains one of the most reliable low-CO2 power sources. However, building new plants takes years, while improving existing ones can deliver massive efficiency gains today.
The transition to an electrified economy - from EVs to AI-driven industries - will demand unprecedented amounts of power. Metroscope is betting that AI-powered diagnostics will be key to making that shift sustainable.
“The energy infrastructure of tomorrow will be a mix of renewables, small modular reactors (SMRs), batteries, and hydrogen,” Schwartz says. “For all of this to work together, AI will be essential. Managing a decentralized energy grid requires smart systems that can predict failures and optimize performance in real-time.”
Metroscope is proof that AI isn’t just for software - it’s now also a core technology for industrial infrastructure.
The Bigger Picture: AI in Industrial Decision-Making
Metroscope’s success highlights a broader shift: AI is not just about deep learning and generative models—it’s about real-time decision-making in complex, high-stakes environments.
In an industry where trust is everything, Metroscope has built an AI system that doesn’t just predict failures - it explains them. And in the world of nuclear energy, where a single misstep can mean billions in losses, Schwartz states that it's exactly what operators need.
“We’re at the beginning of a massive industrial transformation,” says Schwartz. “AI is no longer optional—it’s the backbone of a smarter, more resilient energy grid.”