Matlantis Accelerates Materials Discovery with NVIDIA ALCHEMI Toolkit

New integration enables high-throughput materials simulation at industrial scale

Apr. 15, 2026 at 12:43am

A highly detailed, glowing 3D illustration of a complex molecular structure made up of interconnected neon cyan and magenta lines, representing the advanced computational capabilities of the Matlantis materials simulation platform.Matlantis' integration of NVIDIA's ALCHEMI Toolkit promises to unlock new frontiers in industrial-scale materials discovery through unparalleled computational power and efficiency.Cambridge Today

Matlantis, a joint venture between PFN and ENEOS, has announced the integration of the NVIDIA ALCHEMI Toolkit into its materials simulation platform. This integration is expected to deliver unprecedented computational throughput for industrial-scale materials discovery, building on previous collaborations that have achieved up to 10x speed improvements in key atomistic calculations.

Why it matters

The ability to rapidly simulate and test new materials at an atomic level is crucial for accelerating innovation in industries like catalysts, batteries, semiconductors, and chemicals. By integrating NVIDIA's high-performance computing toolkit, Matlantis is empowering its customers to dramatically speed up their R&D cycles while maintaining industry-leading accuracy.

The details

The new integration centers around Matlantis' LightPFP, a lightweight potential for large-scale simulations that now leverages the NVIDIA ALCHEMI Toolkit-Ops. This allows for more flexible resource allocation and stable operations, with Matlantis mitigating communication overhead challenges by replacing the neighborhood list construction process during inference. Looking ahead, Matlantis also plans to integrate its flagship Universal Machine-Learning Interatomic Potential (uMLIP), PFP, with the NVIDIA ALCHEMI Toolkit to further power both small-scale research and large-scale industrial materials design.

  • Matlantis announced the NVIDIA ALCHEMI Toolkit integration on April 14, 2026.
  • LightPFP, which integrates the NVIDIA toolkit, will be available in mid-April 2026.
  • Matlantis plans to integrate its PFP potential with the NVIDIA ALCHEMI Toolkit in the future.

The players

Matlantis

A joint venture between PFN and ENEOS that provides a universal atomistic simulator to support large-scale material discovery.

NVIDIA

A technology company that develops graphics processing units (GPUs) and related computing technologies.

NVIDIA ALCHEMI Toolkit

A toolkit developed by NVIDIA to enable high-throughput materials simulation at industrial scale.

Masateru Kawaguchi

The VP of Product Department at Matlantis.

PFN

Preferred Networks, a Japanese artificial intelligence research company that co-developed Matlantis.

ENEOS

A Japanese energy company that co-developed Matlantis with PFN.

Got photos? Submit your photos here. ›

What they’re saying

“Our continued collaboration with NVIDIA represents our commitment to providing the most advanced tools for material developers.”

— Masateru Kawaguchi, VP of Product Department, Matlantis

What’s next

Matlantis is planning the integration of its flagship Universal Machine-Learning Interatomic Potential (uMLIP), PFP, with the NVIDIA ALCHEMI Toolkit. This upcoming development aims to further power both small-scale research and large-scale industrial materials design, bridging the gap with unparalleled GPU efficiency.

The takeaway

The partnership between Matlantis and NVIDIA represents a significant advancement in the field of materials discovery, enabling researchers and industries to accelerate their R&D cycles through unprecedented computational power and efficiency. This integration highlights the growing importance of high-performance computing and AI-driven simulations in driving innovation across a wide range of industries.