MARVEL-AI: Accelerating Advanced Materials Innovation From Discovery To Deployment

SINGAPORE, 27 October 2026 - Advanced materials underpin technologies across semiconductors, advanced manufacturing, energy and chemicals. Newer and higher-performing materials can unlock better products, more reliable manufacturing and new industrial applications. Each material candidate must move through experimentation, characterisation, validation, process development and industrial translation before it can progress towards real-world use. This can involve multiple development cycles, substantial resources and separate hand-offs between research and downstream development.

MARVEL-AI is a new initiative designed to change the speed and scale of materials innovation, with a differentiated proposition that connects AI-enabled materials discovery, experimentation and validation with the pathway towards real-world industrial applications. Led by the Agency for Science, Technology and Research (A*STAR), MARVEL-AI is supported by the Ministry of Energy, Trade and Industry (METI), EnterpriseSG, Singapore Economic Development Board (EDB) and National Research Foundation (NRF), under Research, Innovation and Enterprise 2030 (RIE2030).

MARVEL-AI is focused on four materials classes: metal alloys, polymers and organics, ceramics, and composites, with applications in areas such as semiconductors, advanced manufacturing, energy and chemicals. It aims to create high value startups, anchor large multinationals with materials development mandate and uplift local enterprises with frontier technologies.

AI and Automation: Closing the Materials Innovation Loop

MARVEL-AI links prediction, experimentation and validation in an iterative closed loop. AI models identify promising candidates and guide experiments; while automated experimentation synthesise and test them. The resulting experimental and characterisation data can then inform what researchers and AI explore next.

Existing capabilities demonstrate what this approach can unlock:

• Polymers: Increasing experimental throughput. In one polymer workflow, automation transformed the process from a typical 8-hour operation window and a maximum of three samples prepared sequentially, to 24-hour continuous operation with up to 25 samples prepared in parallel. This enabled the automated workflow to produce 75 samples in 24 hours, around 25 times the daily output.
• Metal alloys: Accelerating the screening process for more possibilities. In-house synthesis and characterisation technologies enable researchers to screen and test up to 470 unique alloy samples a day, delivering more than 400 times the throughput of conventional methods and allowing a much broader range of compositions to be explored.
• Composites: Moving from materials discovery towards application. AI-enabled materials development helped take a new composite from initial discovery and validation through to a full-scale prototype of a hydrogen-carrying capsule that is 50% lighter and four times more durable, demonstrating the potential to translate promising materials into higher-performing real-world solutions.

Faster experimentation is only part of the MARVEL-AI proposition. Promising materials must still be characterised and validated under relevant conditions, manufactured at scale, and applied by industry. MARVEL-AI connects AI-enabled discovery and automated experimentation with advanced characterisation, manufacturing process development and industrial translation.

Combining Capabilities

Anchored by the A*STAR Institute of Materials Research and Engineering (A*STAR IMRE), the platform combines complementary capabilities across A*STAR, and Institutes of Higher Learning (IHLs).

A*STAR, EnterpriseSG and EDB will also connect companies and investors to MARVEL-AI, to facilitate collaborations across aerospace, advanced manufacturing, precision electronics and specialty polymers to address real-world materials challenges.

Professor Loh Xian Jun, Executive Director of MARVEL-AI, said, “AI is creating unprecedented opportunities for materials discovery, but discovery alone is not enough. The next challenge is proving they can work in the real world. MARVEL-AI builds on Singapore’s established strengths in materials science and connects them with AI, automated experimentation, manufacturing and industrial translation. This will help bridge the gap between discovery and deployment, enabling companies to evaluate whether promising materials can be validated, manufactured and applied in real-world settings.”

Strategic Industry Collaborations

MARVEL-AI works with industry, research and technology partners to apply different parts of the connected materials innovation journey to real-world challenges.

Industry Partner

ST Engineering (Advanced Manufacturing)

High-performance alloys for aerospace, defence and infrastructure applications

Heraeus Electronics (Advanced Manufacturing)

Powders and alloys with tailored properties for advanced manufacturing and electronics

Dura-Metal (Advanced Manufacturing)

Protective coatings and cladding alloys for harsh operating environments

Foxconn Industrial Internet (FII) (Sustainable Materials & Advanced Manufacturing)

AI-enabled design frameworks to develop sustainable materials alongside eco-friendly, printable alloys with reduced reliance on critical raw materials.

Suntar Research (Advanced Materials & Advanced Manufacturing)

Joint R&D of advanced membrane materials through AI-driven discovery facilities, integrated computational-experimental materials science, and pathways for commercial translation

Technology Partner

Chemlex (Chemicals)

High-throughput monomer synthesis and specialty polymer development

XtalPi (Intelligent & Automated Materials Discovery)

Customised laboratory automation solutions, joint research facilities, computational-experimental materials discovery, and talent exchange to accelerate commercial translation

CuspAI (Electronics, Energy, Environment)

Collaborating on closed-loop, AI-driven materials discovery using high-throughput experimentation, connecting CuspAI's frontier generative AI platform with MARVEL-AI’s autonomous lab capabilities to design next-generation materials.

Google Cloud (Frontier Scientific AI & Lab Robotics)

Partnering to establish a secure research environment on Google Cloud. This platform gives materials scientists early access to Google’s frontier AI, accelerating discovery while keeping sensitive institutional data strictly protected.

Shimadzu (Intelligent Characterization & Lab Automation)

Integrating advanced, automated analytical and intelligent characterization instruments into MARVEL-AI's multi-material self-driving laboratory setup, enabling real-time, closed-loop feedback loops for high-throughput polymers, coatings, and structural materials screening.

– END –