Anthropic is betting on a trillion-dollar track! Introducing a “model hardware standard” to help AI move into the real world

The Zhitong Finance App learned that Anthropic announced a new software standard to enable artificial intelligence (AI) assistants such as Claude to work better with robots and scientific and manufacturing hardware. The AI startup is launching its “Model Hardware Standard” (Model Hardware Standard), or MHS, in the form of a research preview so that developers in science, robotics, and manufacturing can test how Claude interacts with the real world and establish necessary safety precautions before the standard is widely released.

Essentially speaking, MHS stores information for AI assistants to understand how the hardware works, without having to read paper instructions or rely on hidden knowledge held by only a few experts. For example, a factory robotic arm supplier can specify in MHS how AI should safely control a heavy robotic arm, including limiting its movement speed or angle of motion.

As AI technology continues to advance, companies and investors are increasingly concerned about how AI can be applied to the real world, such as manufacturing, robotics, and scientific experiments. A report released earlier this year by Barclays predicted that AI-driven robots and autonomous machines have the potential to grow into a trillion-dollar market by 2035. Meanwhile, Google (GOOGL.US), OpenAI, and Nvidia (NVDA.US) are also developing AI models and software for robots.

Anthropic said in a statement on Thursday that industry participants can join the waitlist to apply for a trial of MHS. The company plans to open source the framework after the preview period so that any hardware manufacturer can build their own MHS specification, but Anthropic declined to reveal a specific timeline.

Jonah Cool, one of the people responsible for collaboration and deployment at Anthropic Life Sciences, said, “In many cases, the reason scientific research cannot be carried out is because you cannot use the relevant equipment or the technical operation of the equipment is too difficult.” “With MHS, Claude can be an enabler for scientists, enabling them to use the right equipment in an expert manner.”

Anthropic has also previously launched and promoted a standard for software applications — the “Model Context Protocol” (Model Context Protocol), or MCP, which makes it easier for AI assistants to interact with third-party applications. For example, this allows Claude and its chatbot competitor ChatGPT to connect to Gmail, Google Calendar, and Slack.

“What MCP does for software, MHS will do the same for the hardware world,” said Anthropic technical team member Alek Kemeny. “Where it can create value is in scientific laboratories, which often have dozens or even hundreds of devices.”

Anthropic has teamed up with a number of industry participants to test this hardware framework, including Amazon Web Services (AWS), life science and diagnostics company Danaher (DHR.US), AI open source community Hugging Face, and Raspberry Pi. In a promotional video released by Anthropic, the company said that a scientist at the biotech company Genentech Inc. was able to send Claude a PDF document of an experiment he designed, and the AI assistant was then able to perform the experiment independently, and the hardware used was already equipped with MHS specifications.