MACE vs MatterSim vs Orb (2026): ML Interatomic Potentials
MACE vs MatterSim vs Orb machine-learning interatomic potentials for materials screening: accuracy, speed, training data and licensing. 2026 comparison for self-driving labs.
AI for Science Landscape 2026: Periodic Labs, Lila Sciences, and the Self-Driving-Lab Race
A 2026 map of the AI-for-science landscape: Periodic Labs, Lila Sciences, Berkeley A-Lab, Emerald Cloud Lab, and academic SDLs — capital, technical approach, moats, and honest caveats.
The Autonomous Materials-Discovery Pipeline: Closed-Loop Synthesis and Characterization (2026)
Inside an autonomous materials-discovery pipeline: ML screening, robotic synthesis, automated characterization, and the learning loop — plus an honest look at the A-Lab novelty debate.
The Experimental-Data Moat: Why AI Labs Are Building Robots to Make Their Own Data (2026)
The public internet is running dry as a training source. Why frontier AI-for-science labs are manufacturing proprietary physical-world experimental data — the compounding-data thesis, explained.
Self-Driving Lab Architecture: The Closed Loop That Runs Experiments (2026)
A reference architecture for self-driving labs: the design-make-measure-learn closed loop, orchestration, safety interlocks, and human-in-the-loop control for autonomous experimentation.