Great meeting the Manus team and seeing their work on embodied intelligence up close.
Manus, featured in the NVIDIA Keynote 2025, is building technology that translates human motion into machine skill. Their tagline says it all: “Where Human Motion Becomes Machine Skill.”
From Models to Machines
What stood out was how fast the conversation is moving from AI models to real-world robotics — from prediction to action, from software to machines.
The event “From Models to Machines: AI, World Models, and the Next Generation of Robotics” brought together key voices in the space, including Maarten Witteveen (CTO and Co-Founder, President North America at Manus) and Sebastien Willems (Co-founder and CEO of Antfarm), along with a hands-on robotics hackathon led by Dmitri Jarnikov (Senior Director Data Science at Prosus Group).
Seeing human motion translated into machine skill in person makes the opportunity feel very tangible. This is not a slide deck — it is hardware doing real things.
Why This Matters for AI Infrastructure
Embodied AI creates massive infrastructure demands. Robots operating in the physical world need:
- Real-time inference — latency budgets measured in milliseconds, not seconds
- Edge computing — models running on device, not in the cloud
- Simulation at scale — training in virtual environments before deploying to physical ones
- GPU orchestration — massive compute for training world models
This is where platform engineering and AI infrastructure intersect with robotics. The same Kubernetes clusters and GPU platforms powering LLM inference today will power robotic perception and control tomorrow.
The Next Wave
A strong reminder that the next wave of AI value will come from execution in the physical world, not only from better models. As the Prosus State of AI report noted, world models — systems that understand the physical world for robotics, autonomous systems, and manufacturing — represent a frontier where no one has yet established dominance.
Europe, and Amsterdam in particular, is positioning itself at the center of this conversation.
For more on AI platforms and edge computing, connect with me on LinkedIn or follow @TheLucaBerton.
