Anthropic makes first move into physical AI with new way for scientists, manufacturers to bring equipment to life

Anthropic, the artificial intelligence company, has unveiled its first venture into physical AI with the introduction of the Model Hardware Standard (MHS). This new framework enables scientists and manufacturers to seamlessly integrate their equipment with advanced large language models, like Anthropic's Claude, transforming factories and laboratories into autonomous systems.

MHS simplifies the process of connecting AI with physical objects, reducing the time from weeks or months to just hours or minutes. Typically, this integration would require specialized teams to perform custom builds, but MHS streamlines this task, making it accessible to anyone. The framework is model-agnostic, working with any LLM, not just Claude, and is built on the Model Context Protocol (MCP), a universal standard for connecting data sources that Anthropic introduced in 2024.

While not all existing equipment can connect to MHS without additional programming interfaces, the company is collaborating with device manufacturers to develop new products with necessary interfaces pre-loaded. Additionally, Anthropic is helping manufacturers add MHS connections to existing products, paving the way for a future where scientists can simply purchase devices that work out-of-the-box.

The research preview of MHS comes at a time when interest in combining AI and robotics is growing. Hugging Face, a company Anthropic partnered with, also unveiled its first physical AI product, a robotic duck, although it's not powered by MHS. Nvidia, a major player in the AI and hardware industry, has also expressed interest in physical AI, predicting that every industrial company will eventually become a robotics company.

Anthropic developed MHS in collaboration with the HHMI Janelia Research Campus, a biomedical research center in Virginia, and has already received early access from labs and manufacturers in various fields, including biotech, robotics, and quantum computing. Some of these partners include Genentech, Carnegie Mellon University, quantum computing company QuEra, Universal Robots, Amazon Web Services, Doosan Robotics, Danaher, and Hugging Face.