Skild AI secures $1.4B to build universal robot ‘brain’

Skild AI is positioning itself as the physical-world analogue of OpenAI and Anthropic: a foundation model that can run on almost any robot, from humanoids to warehouse arms. A $1.4 billion Series C at a $14 billion valuation instantly puts it in the top tier of AI companies by capital firepower, and signals that investors now see “physical AI” as the next mega-wave after language and image models. If Skild’s omni-bodied “Skild Brain” works as advertised—general control across many embodiments with in‑context adaptation—it would be one of the first practical steps toward agentic systems that can reason and act in the real world, not just in text boxes.

Strategically, this moves the race to AGI into robotics and embodied intelligence. SoftBank, NVIDIA (via NVentures), Samsung, LG and Salesforce are not just chasing returns; they’re securing a front‑row seat in what could become a de facto operating system for robots. That threatens incumbent industrial automation vendors and also raises the bar for other embodied AI startups, who now have to compete against a massively funded platform bet. For the broader ecosystem, it tightens the feedback loop between simulation, video data and real‑world deployment—exactly the kind of scaling curve that has driven rapid capability jumps in language models.