physical AI
Open Robotics Models Lower the Entry Cost for Industrial Automation Startups
Perceptron AI released Isaac 0.5 on August 28, an open-weight 36-billion-parameter model combining video understanding, embodied reasoning and robot control. For smaller robotics and industrial-software companies, the opportunity is not training another foundation model. It is adapting an increasingly capable shared model to one repetitive physical workflow where customers already understand the economic value of automation.
Ropedia raises 22 million dollars to scale physical AI training data infrastructure
Ropedia, a Singapore-based physical AI company founded in 2025 and led by co-founder and CEO Zhaoxi Chen, has raised 22 million dollars in pre-Series A funding, bringing total funding to 30 million dollars. The company builds data infrastructure that captures human movement through wearable devices to train robots, and says it now serves more than 20 robotics and foundation-model companies.
NCS expands Sunshine.AI: sovereign platform targets production scale across Asian enterprises
Singapore based NCS has expanded its Sunshine.AI portfolio with platforms for AI agents, no-code software creation, enterprise assistants, physical AI and production assurance. The announcement moves beyond a single model or chatbot by combining reusable infrastructure, sector partnerships and implementation guidance. SaaS founders and technology buyers should note the emphasis on controlled deployment, measurable operating results and data sovereignty.
Agility opens Fremont AI hub: nearly 200 hires support expanding Digit deployments
Agility Robotics has opened a 60,000 square foot facility in Fremont, California, to develop software and artificial intelligence capabilities for its Digit humanoid robot. The company plans to hire nearly 200 people and says it has more than 300 million dollars in multiyear orders, subject to contractual milestones. The expansion shows physical AI moving closer to operational deployment.