AI-driven RADIANT project aims to turn years of materials development into months

Project news

New materials are essential for clean energy, hydrogen infrastructure, quantum devices and sustainable manufacturing, but developing and industrialising them can still take five to seven years. RADIANT will create a Finnish AI-driven materials acceleration platform that uses self-driving laboratories, high-throughput synthesis and advanced computing to shorten this cycle to months. The project will bring together VTT and the University of Helsinki to build a national capability for faster, more sustainable and more competitive materials innovation.

VTT and the University of Helsinki are kicking off RADIANT (Radical AI-Driven Materials Acceleration for Next-Generation Technologies), a project planned for 1 September 2026-30 August 2029 with a potential extension to 2031. The first phase has total estimated costs of EUR 7.5 million, with EUR 6.0 million in funding requested from Business Finland.

The project aims to build a self-driving materials development platform in Finland to help industry discover, test and pilot advanced materials for the energy transition, hydrogen and green industrial processes, quantum technologies and substitution of critical raw materials.

"Materials are often the hidden bottleneck behind industrial renewal. With RADIANT, we want to move from slow, sequential materials development towards a self-learning system where AI, experiments and manufacturing knowledge reinforce each other. This can give Finnish industry a much faster route from idea to pilot-scale validation," says Research Professor Anssi Laukkanen, VTT, project PI.

A self-driving lab for materials

RADIANT combines AI models, automated experiments, high-throughput synthesis and characterization, multiscale simulations, and FAIR data workflows. In practice, the platform will propose new candidate materials, produce material libraries, measure their properties and learn from the results. 

In its first phase, it aims to manufacture and screen more than 100,000 material variants. A distinctive feature is that AI will be trained not only on predicted material properties, but also on synthesis and manufacturing data, so that promising candidates are designed with practical manufacturability in mind from the beginning.

Focus on Finnish industrial strengths

The first use cases focus on materials for small modular reactors and advanced nuclear systems, hydrogen-compatible alloys and coatings, high-quality thin films for quantum devices, and cobalt-free hard metals. These are areas where small improvements in materials can unlock major benefits, such as more durable energy systems, safer hydrogen infrastructure, better quantum devices and reduced dependence on critical raw materials. 

The longer-term goal is to create new intellectual property, industrial proof-of-concepts, deep-tech start-ups and export-driven business around AI-enabled materials innovation.

Next steps and expected impact

The first phase will establish the core platform, discovery pilots, automated synthesis-characterization pipelines and FAIR-aligned data management. The second phase would move towards industrial piloting and validation, targeting at least four validated business cases and stronger company participation. By 2030-2035, the ambition is to support industrial deployment, licensing, Materials-as-a-Service models and at least two deep-tech spin-offs, while building a national centre of expertise involving at least 25 researchers and supporting around 10 PhD theses.

"The strength of this approach is that we connect high-throughput synthesis with advanced characterization and real performance data. That means we can identify promising materials faster but also understand why they work and how they can be made reliably," says Professor Filip Tuomisto, University of Helsinki, project co-PI.

For further information:

VTT contact person
Anssi Laukkanen, Research Professor, VTT, project PI.

Partner contact persons: 
Filip Tuomisto, Professor, University of Helsinki, project co-PI
Kostas Sarakinos, Professor, University of Helsinki, project co-PI. 

 

Radiant in a nutshell:

Project name

RADIANT - Radical AI-Driven Materials Acceleration for Next-Generation Technologies

Partners

VTT Technical Research Centre of Finland and University of Helsinki

Planned first phase

1 September 2026-30 August 2029

Potential extension phase

1 September 2029-30 August 2031

First-phase total estimated costs

EUR 7.5 million: EUR 3.75 million VTT and EUR 3.75 million University of Helsinki

Core target

Shorten concept-to-pilot materials development from 5-7 years to approximately 6-12 months

Planned first-phase scale

More than 100,000 material variants manufactured and screened

Application domains

Energy transition and advanced nuclear; hydrogen and green processes; quantum materials; critical raw-material substitution/cobalt-free hard metals

 

RADIANT project summary
RADIANT project summary
Radiant Foundation model
Integrated computational materials engineering (ICME) workflow, coined AlloyOracle, merging design of material chemistry, synthesis and manufacturing processes, microstructure and application performance critical properties. The foundation model (“MatGPT”) integrates data across the ICME workflow, and agentic AI closes the materials discovery loop.
Share
Anssi Laukkanen
Anssi Laukkanen
Research Professor