The National University of Singapore has secured four of the eight inaugural projects under Singapore's S$120 million AI-for-Science Initiative, launched on 16 June 2026 at the AI4X Accelerate Conference. The national programme pairs top AI researchers with domain experts to speed up discovery in materials, computing, genomics, and agriculture while building a generation of scientists fluent in both AI and traditional disciplines.
Professor Tan Chorh Chuan, Permanent Secretary (National Research and Development), announced the initiative alongside the project leads. The strategic research projects aim to develop "bilingual" scientists who can apply AI methods across life sciences, materials science, and quantum science. This focus on AI for Science & Research reflects a structural shift in how institutions fund and organise discovery work.
From materials discovery to program reasoning
The Materials Data Foundry, co-led by Professor Sir Konstantin Novoselov at NUS's Institute for Functional Intelligent Materials and Professor Alán Aspuru-Guzik at the University of Toronto's Acceleration Consortium, will deploy an autonomous lab powered by AI and robotics. The facility aims to build the world's largest dataset linking synthesis protocols to real-world material performance. Three testbeds are planned: beyond-silicon and quantum-topological materials, oxygen-evolution electrocatalysts, and corrosion-resistant high-entropy alloy coatings. Industrial partners Nvidia and VeChain supply digital infrastructure. The resulting dataset will train AI models to accelerate materials discovery for electronics, clean energy, and sustainable infrastructure.
The AI for Program Reasoning project tackles software correctness at a time when AI-generated code is growing fast. Co-led by Professor Abhik Roychoudhury from NUS School of Computing and Professor Cristian Cadar from Imperial College London, with collaborators from Singapore Management University, MIT, and ETH Zürich, the team is building AI tools that combine formal mathematical proofs with informal reasoning to analyse undocumented code. The tools will be tested on critical systems such as network protocols and components of the Linux kernel. The long-term aim is to produce specialised AI agents that can audit code produced by other AIs.
Multi-omics models and agricultural digital twins
The project Accelerating Genomic Research with AI, led by Professor Cheng Ching-Yu from NUS Yong Loo Lin School of Medicine and A*STAR Research Entities, is developing MultiOmicsFM - a unified AI foundation model that interprets DNA, RNA, and gene activity in concert rather than separately. By training on Singapore's multi-ethnic genomic datasets, the model targets faster disease risk prediction and more precise mRNA therapy optimisation, strengthening the country's position in AI-driven precision medicine.
Professor Roman Carrasco from NUS Biological Sciences and collaborators at the Illinois Advanced Research Center at Singapore are constructing agricultural digital twins. The platform uses knowledge-guided AI, combining sensor data with established crop science principles, to create virtual replicas of farmland. These simulations will deliver forecasting and decision-support tools to help farmers and policymakers optimise planting strategies, resource use, and supply chains. The work addresses Southeast Asia's food security and positions Singapore as a regional hub for climate-resilient agricultural innovation.
For researchers aiming to build the cross-disciplinary skills these projects demand, an AI Learning Path for Research Scientists can provide structured guidance on integrating AI into experimental and theoretical work.
Why this matters for science and research professionals
The AI-for-Science Initiative creates immediate demand for researchers who can bridge AI and domain expertise. Open datasets and tools produced by these projects - particularly the Materials Data Foundry's synthesis dataset and the MultiOmicsFM model - will become shareable resources that other labs can build on. Professionals in materials science, genomics, and computational chemistry can watch for early access to these assets and evaluate how similar AI-first approaches could shorten timelines in their own research.
Your membership also unlocks:
