Meta AI speeds up scientific research with new data set and model

Meta is pioneering scientific research through massive data release that could revolutionize chemistry, drug discovery, and artificial intelligence. The company’s Open Molecules 2025 project represents an unprecedented computational leap, generating 100 million quantum mechanics simulations requiring 6 billion compute hours — significantly larger than any previous academic dataset. This landmark scientific contribution could dramatically accelerate the process of developing new medications, materials, and more advanced AI systems.

The big picture: Meta has released an enormous chemistry dataset and a new AI model that dramatically accelerates scientific discovery while potentially advancing general AI capabilities.

• The Open Molecules 2025 dataset required 6 billion compute hours to produce 100 million quantum mechanical calculations of atomic and molecular interactions.
• Meta’s researchers deliberately omitted data that could help bad actors develop biological or radiological weapons.

Why this matters: The dataset transforms how scientific research can be conducted by making previously prohibitive quantum calculations accessible and dramatically faster.

• “It’s going to dramatically change how people do computational chemistry,” said Sam Blau, a research scientist at Lawrence Berkeley National Laboratory who collaborated on the project.
• The data represents “two orders of magnitude more compute than any kind of academic data set that’s ever been made,” significantly expanding scientific capabilities.

Technical breakthrough: The calculations model larger and more complex atomic systems than previous research efforts.

• While most density functional theory (DFT) calculations typically map interactions of 20-30 atoms, Meta’s new dataset includes systems of up to 350 atoms.
• These larger models provide a more complete picture of real-world atomic interactions, potentially leading to more accurate predictions.

The AI advantage: Meta used this expansive dataset to train UMA (Universal Frontier model for Atoms), an AI model that dramatically accelerates molecular research.

• UMA can achieve the same results as conditional density functional theory methods but 10,000 times faster.
• Larry Zitnick, a Meta FAIR researcher, explained the impact: “Instead of saying, ‘Let me try this new molecule, and I’ll check back in a couple days,’ it’s saying, ‘Let me try these 10,000 different molecules and just run them all simultaneously’ and then getting an answer in a minute.”

Broader applications: The dataset spans four key scientific domains with significant potential impact.

• The Open Molecules dataset covers small molecules, biomolecules, metal complexes, and electrolytes, supporting diverse research applications.
• Meta is offering the UMA model in three different sizes, allowing researchers to balance computational power against speed and cost considerations.

Strategic importance: Meta sees this work as crucial for developing more advanced artificial intelligence systems.

• The company’s Fundamental AI Research team believes that to reach “Advanced Machine Intelligence,” AI systems must develop “world models” that include understanding the physical world at the atomic level.
• By open-sourcing this dataset, Meta aims to accelerate innovation across the scientific community.