Two landmark research programs funded by Singapore's National Research Foundation will harness artificial intelligence to transform how scientists discover new catalysts and verify software reliability, with combined investment of nearly S$20 million (approximately $15 million USD). The initiatives represent a significant shift in how computational chemistry and materials science are conducted globally, moving from decades-long discovery cycles to AI-accelerated workflows that could reshape industries from pharmaceuticals to clean energy.

How Are AI Systems Accelerating Catalyst Discovery?

The first program, titled "Overcoming the Terascale Design Challenge: Next-Generation Efficient AI for Discoveries in Surface Science and Catalysis," brings together researchers from Imperial College London and Singapore's Agency for Science, Technology and Research (A*STAR). The S$10 million, five-year initiative aims to develop a cloud-based artificial intelligence platform that will dramatically speed up the process of finding new catalysts for industrial applications.

Catalysts are chemical substances that enable reactions essential for producing everything from clean fuels to medicines, but discovering new ones has historically been slow and prohibitively expensive. Traditional trial-and-error approaches can take decades to yield viable candidates. The Imperial-A*STAR team plans to achieve a 500-fold acceleration in the computational screening of catalyst compositions, allowing researchers to rapidly identify promising candidates for sustainable feedstocks like carbon dioxide and bio-based chemicals.

The research team will develop what they call a "Surface Science Foundation Model," which functions as a comprehensive artificial intelligence-driven knowledge base of surface structures and interactions. This foundation model will enable rapid prediction of material properties and synthesis conditions, dramatically reducing both time and cost in the discovery phase before moving into wider development.

"Catalysts underpin everything from sustainable fuels to pharmaceuticals, yet their discovery remains slow and expensive. Our goal is to create open-source tools that support the development of adaptive, autonomous AI systems that can navigate this enormous design space and deliver breakthroughs at unprecedented speed," said Professor Aron Walsh, from the Department of Materials at Imperial.

Professor Aron Walsh, Department of Materials, Imperial College London

The program will deliver a cloud-based AI platform for catalyst design that will be accessible to researchers worldwide, supported by cutting-edge computational infrastructure and high-throughput catalyst screening platforms. Collaborative exchanges between London and Singapore will ensure knowledge transfer and strengthen the international research partnership.

What Role Will AI Play in Software Verification and Safety?

The second major grant, worth S$10 million over four years, addresses a critical challenge in computer science: ensuring the correctness and security of software systems in an era increasingly dominated by artificial intelligence-generated code. The program, called "AI for Program Reasoning," is co-led by Imperial's Professor Cristian Cadar and the National University of Singapore's Professor Abhik Roychoudhury, with collaboration from Singapore Management University, MIT, and ETH Zurich.

Software reliability has become increasingly urgent as artificial intelligence systems generate larger portions of code. Errors in software can have catastrophic consequences, from research retractions to security breaches affecting critical infrastructure. The research team will develop next-generation AI technologies for program reasoning, combining symbolic analysis with generative artificial intelligence to automate both formal and informal reasoning about software.

"Modern infrastructure and technology rely on software, and reliability failures can lead to serious consequences that impact the economy and society. As AI begins to write significant portions of code, we must ensure that this code is not only functional but also safe and secure. Our research will create AI systems that can reason about programs, verify their correctness, and even repair them when necessary," stated Professor Cristian Cadar.

Professor Cristian Cadar, Department of Computing, Imperial College London

The program aims to democratize program reasoning through AI-generated analysis tools that work across diverse programming languages and scientific domains, making verification accessible to researchers beyond computer science specialists.

Why Is Singapore Investing Heavily in AI for Science?

Both programs are funded through Singapore's National Research Foundation as part of the government's broader "AI for Science" initiative, which supports the development and adoption of artificial intelligence methods and tools that enhance research productivity and advance scientific discovery. The programs are hosted by Imperial Global Singapore, the university's research and innovation hub located at the National Research Foundation's Campus for Research Excellence and Technological Enterprise.

The strategic investment reflects Singapore's commitment to positioning itself as a global hub for AI-driven scientific research. By partnering with leading international institutions like Imperial College, the National University of Singapore, and world-class research organizations, Singapore is building infrastructure and expertise that will benefit researchers across multiple disciplines and geographies.

• Catalyst Discovery Program: S$10 million investment targeting 500-fold acceleration in computational screening of catalyst compositions for sustainable fuels and pharmaceuticals
• Software Verification Program: S$10 million investment developing AI systems to verify correctness and security of software code across programming languages
• International Collaboration: Partnerships between Imperial College London, A*STAR, National University of Singapore, MIT, and ETH Zurich bringing together world-leading expertise in materials science and computer science
• Open-Source Accessibility: Both programs aim to deliver tools and platforms accessible to researchers worldwide, democratizing access to advanced AI capabilities
• Infrastructure Support: Funding supports cutting-edge computational infrastructure, high-throughput screening platforms, and collaborative exchanges between research institutions

The timing of these investments aligns with broader global trends in AI-driven materials discovery. The U.S. Department of Energy's Genesis Mission, for example, is similarly leveraging artificial intelligence to accelerate scientific discovery across disciplines from high-energy physics to materials science to fusion-energy research. The Fermilab Data Platform has been selected as a key infrastructure partner for the Genesis Mission's American Science Cloud, providing petabytes of storage and data-access tools to ensure that artificial intelligence systems can access and process scientific data efficiently.

Meanwhile, the ion exchange materials market is experiencing significant investment momentum driven by artificial intelligence applications. Over $73 million in venture funding has been deployed across the sector in recent quarters as water scarcity challenges and semiconductor manufacturing demands drive adoption of AI-powered purification solutions. This broader investment activity demonstrates that artificial intelligence is becoming essential infrastructure for materials science research and development globally.

"AI is transforming the landscape of scientific research, opening up new opportunities in discovery and innovation. These two major grants illustrate how AI can accelerate breakthroughs from identifying novel chemical catalysts to improving the reliability of scientific software, with benefits that extend across scientific domains," explained Professor Azra Ghani, Academic Director of Imperial Global Singapore.

Professor Azra Ghani, Academic Director, Imperial Global Singapore

The Imperial College programs represent a significant milestone in the convergence of artificial intelligence and materials science. By combining advanced computational methods with domain expertise and international collaboration, these initiatives are establishing new models for how scientific discovery can be conducted faster, more efficiently, and more accessibly across the global research community.