Swiss AI cuts cement emissions by 50% while maintaining strength

Researchers at Switzerland’s Paul Scherrer Institute (PSI) have developed an AI framework that can generate low-carbon cement formulations in seconds, potentially cutting CO₂ emissions by up to 50% while maintaining structural performance. This breakthrough addresses one of the world’s biggest climate challenges, as cement production accounts for about 8% of global carbon dioxide emissions and humanity consumes more cement than food—around 1.5 kilograms per person daily.

How it works: The PSI team built a specialized AI system from the ground up rather than adapting generic models for cement development.

• Their custom simulation software models how various cement ingredients react during the hardening process, then trains a neural network to predict mixture strength based solely on composition.
• This compresses what would typically take weeks or months of lab work into milliseconds of computation.
• Genetic algorithms—computational tools that mimic natural selection—are applied to find optimal mixes balancing strength and emissions, with top candidates flagged for real-world testing.

The results: Initial data shows the AI-generated formulations can match today’s performance standards while dramatically reducing environmental impact.

• Some of the most effective mixes emit up to 50% less CO₂ than conventional cement while maintaining equivalent strength.
• The system can adapt to regional raw materials and different production scenarios without requiring extensive manual testing of every possibility.

What they’re saying: The research team emphasizes the transformative potential of their approach for accelerating materials development.

• “This allows us to simulate and optimise cement formulations so that they emit significantly less CO₂ while maintaining the same high level of mechanical performance,” explains Romana Boiger, lead author of the study.
• “The range of possibilities for the material composition – which ultimately determines the final properties – is extraordinarily vast. Our method allows us to significantly accelerate the development cycle by selecting promising candidates for further experimental investigation,” notes Nikolaos Prasianakis, head of PSI’s Transport Mechanisms Research Group.
• John Provis, who leads PSI’s Cement Systems Research Group, puts the scale in perspective: “To put it bluntly, humanity today consumes more cement than food – around one and a half kilograms per person per day. If we could improve the emissions profile by just a few percent, this would correspond to a carbon dioxide reduction equivalent to thousands or even tens of thousands of cars.”

Competitive landscape: PSI joins a growing movement of institutions applying AI to cement and concrete optimization.

• University of Illinois researchers partnered with Meta and concrete supplier Ozinga to develop AI-optimized concrete that reduced carbon emissions by 40%.
• MIT has trained models to scan research papers and databases to identify novel, low-emission materials.
• PSI’s approach stands out by combining deep chemical simulations with machine learning rather than relying solely on historical data.

What’s next: The promising laboratory results still face real-world validation challenges before commercialization.

• These cement mixes must prove themselves under actual conditions including curing, durability, supply chain integration, and compliance with global standards like EN 197-1.
• The team is expanding the model to account for regional resource availability, cost considerations, and lifecycle performance—key factors for moving beyond research applications.

Why this matters: This development represents AI’s evolution from virtual outputs to reshaping physical materials that form the foundation of the built environment.

• With global cement demand continuing to rise, even modest efficiency improvements could yield massive emissions reductions at scale.
• The approach demonstrates how AI can tackle climate challenges through fundamental materials innovation rather than just process optimization or carbon capture systems.