# Periodic Labs > Source: > Published: 2026-08-13 02:06:02+00:00 # Periodic Labs Periodic Labs builds AI scientists and autonomous laboratories that run real experiments, focused on materials discovery in the physical sciences. ## Overview ## Periodic Labs Periodic Labs is an [AI](/glossary/ai)-for-science company building autonomous laboratories where AI systems propose experiments, run them on real equipment, and learn from physical results. The bet behind Periodic Labs is that models trained on internet text have exhausted the readily available data, and that further progress in the physical sciences requires generating new experimental data rather than re-reading existing papers. Its initial focus is materials discovery, including the search for superconductors that work at higher temperatures, alongside custom agents that help semiconductor manufacturers interpret their own experimental data. Periodic Labs pairs frontier models with physics simulation and robotic synthesis in a closed loop. **Production credibility:** Periodic Labs launched in September 2025 with a $300M seed round, among the largest on record, backed by a16z, Felicis, DST Global, and NVIDIA's venture arm, with individual participation reported from Jeff Bezos, Eric Schmidt, and Jeff Dean. It was founded by Liam Fedus, who co-created ChatGPT at OpenAI, and Ekin Dogus Cubuk, who led materials AI at Google DeepMind and worked on GNoME. Reports in 2026 describe fundraising discussions at a valuation near $7.5B; that figure is unconfirmed by the company. It runs an academic grant program and has a scientific advisory board drawn from Stanford and Northwestern. ### Key Features - Autonomous laboratories that synthesize materials and characterize their properties - Closed-loop workflow where AI proposes a hypothesis, runs the experiment, and learns from the result - Frontier models paired with physics-based simulation rather than text pretraining alone - Materials discovery focused on superconductors and related physical-science targets - Custom agents for semiconductor manufacturers analyzing in-house experimental data - Founding team behind ChatGPT and DeepMind's GNoME materials work - Academic grant program supporting external research in its focus areas ### Ideal Use Case This is a research organization rather than software you sign up for, so the relevant audience is R&D leadership at materials, semiconductor, and energy companies weighing a research partnership, plus scientists tracking where AI-driven discovery is actually producing experimental results. For directory readers, Periodic Labs is worth watching as a marker of whether autonomous labs convert capital into reproducible physical findings. ### How Periodic Labs differentiates Most AI-for-science work still stops at prediction: a model proposes candidate materials and human labs test them later, if at all. Periodic Labs closes that loop by owning the experimental apparatus, so the system generates its own [training data](/glossary/training-data) from physical outcomes rather than from literature. Compared with drug-discovery peers such as Isomorphic Labs and Chai Discovery, its target is the physical sciences rather than biology, where measurements are cheaper to repeat and results are verifiable in a way biological assays often are not. ### FAQ **Q: What does Periodic Labs do?** A: Periodic Labs builds AI scientists paired with autonomous laboratories. Its systems propose experiments, run them on real equipment, and learn from the physical results, initially targeting materials discovery. **Q: Who founded Periodic Labs?** A: Liam Fedus, who co-created ChatGPT at OpenAI, and Ekin Dogus Cubuk, who led materials AI at Google DeepMind and worked on the GNoME materials project. **Q: Can I use Periodic Labs as a product?** A: Not as self-serve software. It operates as a research company and works through partnerships, including custom agents for semiconductor manufacturers, plus an academic grant program. **Q: How much has Periodic Labs raised?** A: It launched in September 2025 with a $300M seed round backed by a16z, Felicis, DST Global, and NVIDIA's venture arm. Reports in 2026 describe later talks at a much higher valuation, unconfirmed by the company. ### tl;dr Periodic Labs is an AI-for-science company building autonomous labs where AI proposes and runs physical experiments, focused on materials discovery. Founded by ChatGPT co-creator Liam Fedus and DeepMind's Ekin Dogus Cubuk, it launched with a $300M seed. ## Why Use Periodic Labs ## User Reviews ## Similar Tools [HealthcareIsomorphic LabsAlphabet's AI drug discovery company — DeepMind spin-out from AlphaFold. Computational drug design at the frontier of structural biology AI.Paid★ 4.93♥ 462](/tools/isomorphic-labs) [HealthcareChai DiscoveryAI-native drug discovery company building foundation models for biology. Forbes AI 50 2026 newcomer; backed by OpenAI Startup Fund and Thrive Capital.Freemium★ 4.75♥ 264](/tools/chai-discovery) [AI/ML ModelsSakana AISakana AI is the Tokyo AI research lab founded by ex-Google Brain David Ha and Transformer co-author Llion Jones. ~$200M+ raised; evolutionary model merginPaid★ 4.51♥ 110](/tools/sakana-ai) [AI InfrastructurePhysical IntelligenceFirst foundation model for general robot control — Pi-zero (π0) is a vision-language-action flow model from a team including Sergey Levine. Open source.Paid★ 4.75♥ 239](/tools/physical-intelligence) [AI InfrastructureSkild AIBuilding the general-purpose robotic brain — Skild's omni-bodied foundation model controls any robot, valued at $14B+ after acquiring Zebra's Robotics business.Paid★ 4.92♥ 484](/tools/skild-ai)