Aug 31, 2026 · 6 min
Physical AI pushes robotics beyond the lab
Signal over Noise: How to position for the physical AI era
As robots learn to adapt in unfamiliar environments, investors must weigh the promise of autonomy against the hardware and supply chains required to deliver it.
- 1Physical AI gives machines the ability to perceive, understand, and act in the real world.
- 2In-context learning could help robots adapt to unfamiliar tasks without requiring entirely new training.
- 3The shift creates potential investment implications across robotics hardware and the broader supply chain.
Don't miss
The central turn is the idea that in-context learning could let robots adapt to unfamiliar real-world tasks more quickly.
The brief
Physical AI marks a shift from software that generates outputs to systems that perceive, understand, and manipulate the world around them.
The episode focuses on in-context learning, a concept associated with large language models, as a possible route to faster robotic adaptation.
That promise depends on more than algorithms: sensors, machines, components, and the wider robotics supply chain must support real-world deployment.
Ulrike Hoffman-Burchiadi, UBS’s Americas CEO and Global Head of Equities, frames the technology through its potential investment implications.
The episode closes with UBS’s reminder that its investment views are informational and not tailored to individual circumstances.
What was said on this episode
9 statements · 9 positive
Physical AI has officially entered its current era.
“We have officially entered the era of physical AI.”
Listen at 0:08
Leading robots are being built with vision-language-action models.
“Today's cutting-edge robots are being built with foundational vision language action models.”
Listen at 1:07
Advanced robots can learn tasks from brief demonstrations and immediate environmental data.
“these robots can now watch a human perform a task just once or twice via a camera. or take a few quick data points from their immediate surroundings and adapt instantly.”
Listen at 1:15
Robotics is approaching a shift from pre-programmed automation to physical autonomy.
“So we are now at the precipice of moving away from pre-programmed automation and stepping into genuine physical autonomy.”
Listen at 1:35
Markets for advanced robotics sensing technologies are expanding rapidly.
“the market for advanced LiDAR, time-of-flight cameras and tactile sensors is expanding rapidly.”
Listen at 3:01
Electronic-skin and high-frequency optical-sensor companies are well positioned.
“The companies developing electronic skin and high frequency optical sensors are well positioned for this opportunity.”
Listen at 3:23
Local robotic AI workloads will drive demand for specialized edge hardware.
“This drives demand for specialized edge AI hardware designed specifically for robotics.”
Listen at 3:43
Physical AI will unlock economic value across logistics, manufacturing, and healthcare.
“physical AI is here and is going to unlock economic value across logistics, manufacturing and healthcare.”
Listen at 4:02
Humanoid-robot gear, sensor, and chip makers will become more valuable over time.
“companies making the gears that move the humanoid robots, the sensors that let them see and the chips that let them think will become more valuable over time.”
Listen at 4:19
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.