Arm launched the Total Design for Physical AI ecosystem and introduced the Robotics Capability Framework — a unified system for describing robot capabilities. More than 80 companies have already joined the initiative, including AWS, Hugging Face, NXP, QNX, Siemens, Unitree Robotics, Qwen, PSYONIC, and Liquid AI. The framework proposes dividing robotic systems into six levels — from RL0 to RL5 — where they progressively move from reactive behavior to contextual understanding, cognitive capabilities, and self-learning. This is not about standardizing hardware or software: the system links a robot's capabilities to real-world application scenarios and technical requirements, including latency, compute placement, memory capacity, power consumption, determinism, and safety. According to Arm's vision, this will enable more precise answers to the question of what a robot can actually do, rather than simply claiming it is "autonomous."
In parallel, it became known that Anthropic declined to acquire AI startup Decart, valued at $6 billion. This was reported by Bloomberg, citing sources familiar with the matter. Anthropic conducted comprehensive due diligence on Decart's financial condition and assets, but ultimately decided not to continue negotiations. Decart develops specialized chips that optimize computations during the training and deployment of AI models, reducing operational costs. The company also builds world models for simulating physical processes and applies them in autonomous driving and e-commerce projects; its Lucy AI model allows shoppers to virtually try on clothing and accessories.
The deal could have been the largest in Anthropic's history and would have helped scale its computing infrastructure ahead of a public offering, which analysts predict could be one of the largest in the technology sector. Anthropic is simultaneously developing tools for retailers, including AI agents to assist shoppers and automate inventory management and pricing.