Periodic Labs develops AI scientists integrated with autonomous laboratories to accelerate discovery in the physical sciences. These systems generate novel, high-quality experimental data to advance materials design and scientific understanding beyond internet-trained models. The company applies this capability to solve industry challenges, such as improving semiconductor performance and discovering new materials like high-temperature superconductors.
Funding
$300M raised to dateRaised to date based on public sources. This may differ from the amount the company actually raised and is based only on what is publicly available on the internet.
Founders
Product
Problem
Physical scientists and engineers often face long experimental cycles and limited high-quality data, which slows the discovery of new materials and hampers optimization of complex devices such as semiconductors. Traditional AI models rely on static internet text corpora, providing insufficient domain-specific insight for hypothesis generation and testing in the physical sciences.
Solution
Periodic Labs builds autonomous laboratory platforms that execute high‑throughput experiments and capture multi‑modal data streams (e.g., spectroscopy, microscopy, electrical measurements) at scale. The collected datasets train reinforcement‑learning agents—referred to as AI scientists—to propose, test, and refine material designs iteratively. By closing the loop between hypothesis generation, experimental execution, and data‑driven learning, the system reduces the time‑to‑insight for material discovery and device optimization. Cloud‑native analytics pipelines apply deep‑learning models to extract structure‑property relationships and predict performance under novel conditions. Engineers can query the AI scientist via an API or web dashboard to receive ranked candidate materials, experimental protocols, and confidence metrics, enabling faster iteration on challenges such as high‑temperature superconductors or chip heat‑dissipation solutions.
Target Audience
Primary customers are industrial R&D teams in semiconductor manufacturing, advanced materials companies, and academic laboratories focused on accelerated discovery of functional materials and device engineering.
Features
- Fully automated experimental rigs capable of running thousands of parallel trials, each generating gigabytes of raw sensor data
- Integrated multi‑modal data acquisition (optical, electrical, thermal) with real‑time metadata tagging for reproducibility
- Reinforcement‑learning framework that treats the physical lab as an RL environment, allowing agents to learn optimal synthesis and testing policies
- Hybrid simulation‑augmented training pipeline that combines physics‑based models with empirical data to accelerate convergence
- Scalable cloud infrastructure for data storage, model training, and inference, supporting distributed collaboration across sites
- RESTful API and web UI for on‑demand generation of material hypotheses, experimental designs, and performance forecasts
- Automated negative‑result logging and curation, enriching the training corpus with valuable failure data
- Secure, role‑based access controls and end‑to‑end encryption to protect proprietary experimental data