Artificial intelligence is increasingly moving beyond software development into scientific discovery, and CuspAI is positioning itself at the center of that transition. The company has unveiled the AI Materials Foundry, an enterprise platform that combines generative AI, scientific datasets, high-performance computing, laboratory infrastructure, and autonomous research workflows to accelerate the discovery of next-generation materials. More than 45 organizations have joined as founding members, including NVIDIA and Meta, signaling growing industry investment in AI-driven materials science.
The race to commercialize artificial intelligence is expanding beyond digital applications into one of the most challenging areas of scientific research: discovering entirely new materials.
CuspAI has announced the launch of its AI Materials Foundry, a collaborative platform designed to bring together AI models, laboratory networks, compute infrastructure, and scientific expertise into a unified ecosystem for industrial materials discovery. The initiative launches with more than 45 founding organizations and aims to reduce the time required to identify and validate materials for industries including semiconductors, clean energy, advanced manufacturing, and carbon capture.
Unlike traditional research environments where computational modeling, laboratory experimentation, and data analysis operate independently, the AI Materials Foundry integrates these capabilities through CuspAI’s proprietary autonomous AI platform, MIRA. The platform is designed to manage complete discovery cycles—from generating candidate materials and predicting their properties to planning synthesis routes, coordinating laboratory validation, and continuously refining models using experimental feedback.
The announcement reflects a broader trend across enterprise AI, where organizations are increasingly deploying agentic AI systems capable of executing complex, multi-stage scientific workflows with minimal human intervention.
AI infrastructure backed by NVIDIA and Meta
Among the Foundry’s founding members, NVIDIA will provide accelerated computing infrastructure needed to simulate molecular interactions across billions of potential compounds, while Meta’s Fundamental AI Research (FAIR) team contributes its Universal Model for Atoms (UMA), an open-source frontier model designed to improve atomistic chemistry simulations.
Together, these technologies address one of the largest computational barriers in materials science: accurately predicting atomic behavior at scales that would traditionally require years of laboratory experimentation.
The simulation layer is further supported by kUPS, an open-source molecular simulation toolkit jointly developed by CuspAI and NVIDIA’s ALCHEMI (AI Lab for Chemistry and Materials Innovation) initiative. By combining GPU-accelerated simulation with large-scale generative AI models, the platform seeks to create a continuous workflow from molecular design through physical property prediction.
Tackling the materials bottleneck
While industries such as semiconductor manufacturing and renewable energy continue advancing rapidly, many technological breakthroughs remain limited by the availability of suitable materials.
Whether developing batteries with higher energy density, catalysts for carbon capture, advanced polymers, or next-generation semiconductor compounds, engineers frequently encounter the same constraint: the required materials either do not exist or are prohibitively difficult to discover using conventional research methods.
CuspAI argues that AI can significantly compress these timelines by screening enormous chemical search spaces before laboratory validation begins.
The company points to a recent collaboration with Finnish chemicals manufacturer Kemira, where its platform evaluated approximately 300 trillion potential molecular structures and identified twenty novel candidates suitable for experimental testing. According to CuspAI, the project was completed in six months, compared with discovery timelines that traditionally extend over several years.
Agentic AI for scientific research
At the center of the Foundry is MIRA, an autonomous scientific agent built to coordinate multiple stages of materials research.
Researchers specify desired characteristics—such as thermal stability, catalytic activity, electrical conductivity, reaction performance, or manufacturing cost targets—and MIRA generates candidate molecular structures using generative AI models trained on extensive experimental datasets.
The platform then evaluates millions of possible candidates, predicts their physical properties, designs practical synthesis pathways, assigns experiments to partner laboratories based on available expertise and geographic proximity, and incorporates validated experimental results into future discovery cycles.
This feedback-driven architecture represents an emerging class of enterprise AI known as agentic AI, in which autonomous software systems perform coordinated decision-making across multiple tasks rather than executing isolated predictions.
Building a collaborative research ecosystem
Rather than functioning as a centralized laboratory, the AI Materials Foundry operates as a distributed research network connecting computational resources, scientific institutions, industrial partners, and experimental facilities.
One of its first major collaborations is a multi-year partnership with Singapore’s Agency for Science, Technology and Research (A*STAR). The organizations plan to combine AI-driven discovery with autonomous laboratory synthesis to accelerate research across semiconductor materials, carbon capture technologies, and advanced electronics.
To support enterprise adoption, CuspAI has designed the platform with private Foundry instances that allow partners to maintain confidentiality while integrating AI workflows into existing R&D operations.
Data as a competitive advantage
While generative AI models continue attracting attention, many AI researchers argue that long-term performance increasingly depends on proprietary training data.
CuspAI has secured exclusive AI training rights to several of the world’s largest experimental materials databases, including the Cambridge Structural Database (CCDC) and the Inorganic Crystal Structure Database (FIZ Karlsruhe). It has also licensed scientific literature from Wiley and other publishers to expand its materials knowledge base.
This combination of proprietary datasets, frontier AI models, experimental validation, and continuous learning creates what CuspAI believes is a compounding advantage, where each completed research program strengthens future discovery efforts.
The company’s leadership further reinforces its scientific ambitions. CTO and co-founder Professor Max Welling is widely recognized for co-inventing the Variational Autoencoder (VAE) architecture, while Chief Scientific Officer Professor Aron Walsh FRS is a leading computational materials scientist. Industry veteran John Giannandrea, formerly head of AI research at Google and later Apple’s Senior Vice President of Machine Learning and AI Strategy, is expected to help establish the Foundry’s U.S. operations.
As AI increasingly expands into scientific research, platforms capable of combining data, autonomous reasoning, simulation, and laboratory execution may become foundational infrastructure for the next generation of industrial innovation.
Market Landscape
AI for Science is emerging as one of the fastest-growing segments of enterprise artificial intelligence. Organizations across pharmaceuticals, chemicals, advanced manufacturing, energy, and semiconductor industries are investing in generative AI, foundation models, and autonomous research platforms to shorten discovery cycles and reduce R&D costs. According to McKinsey & Company, generative AI could generate between $2.6 trillion and $4.4 trillion in annual economic value, with scientific research and product innovation among its highest-impact applications. Gartner also expects autonomous AI agents to become increasingly integrated into enterprise workflows, extending AI beyond prediction into decision-making and execution. Against this backdrop, collaborative AI research ecosystems such as CuspAI’s Foundry represent a new model for industrial innovation, where cloud infrastructure, proprietary datasets, and distributed laboratory networks operate as a unified discovery platform.
Top Insights
- CuspAI has launched the AI Materials Foundry, combining generative AI, laboratory infrastructure, proprietary datasets, and scientific expertise to accelerate industrial materials discovery across multiple sectors.
- NVIDIA and Meta are founding partners, contributing GPU-accelerated computing and frontier atomistic AI models that significantly improve molecular simulation and scientific research workflows.
- MIRA, CuspAI’s autonomous AI platform, orchestrates end-to-end materials discovery by generating compounds, predicting properties, coordinating laboratory validation, and continuously improving future predictions.
- The Foundry addresses a growing materials bottleneck affecting semiconductors, clean energy, carbon capture, and advanced manufacturing by reducing discovery timelines from years to months.
- Proprietary experimental datasets and continuous feedback from laboratory validation position AI-driven materials discovery as a strategic advantage for enterprise R&D organizations.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
