Key insights
- The article highlights a potential bottleneck in the robotics and physical AI supply chain: connectivity. It argues that as these technologies scale, reliance on terrestrial networks will be insufficient, creating an opportunity for space-based cellular networks like AST SpaceMobile. This could positively influence companies involved in satellite communications and related technologies, but the impact on broader US equities is limited.
Since early 2026, I’ve seen endless discussions around the robotics boom - specifically focusing on supply chain bottlenecks like compute chips, actuators, and battery density, LIDAR and etc. However, I almost never see the connectivity aspect brought up. As physical AI scales, reliance on local Wi-Fi or terrestrial 5G grids won't be enough. I see a massive mid-to-long-term opportunity for space-based Direct-to-Cell (D2C) networks like AST SpaceMobile. Here are a few areas where this becomes a critical bottleneck:
-
Humanoid Robots (Outdoors/Rural): While early humanoids will operate in controlled environments (houses, factories) on local broadband, the 10–15 year horizon will see them working outdoors. Agriculture, search & rescue, forestry, and construction robots cannot rely on terrestrial cell towers. They will need ubiquitous, un-droppable connectivity.
-
Autonomous Vehicles & Smart Cars: Beyond just navigation, cars are becoming rolling AI hubs with complex entertainment and AI voice agents. Dead zones on highways are going to be unacceptable for real-time edge-to-cloud AI synchronization.
-
Delivery Drones: This goes without saying. Beyond-Visual-Line-of-Sight (BVLOS) drones need seamless, cross-country connectivity that standard cellular towers simply can't guarantee due to altitude and terrain gaps. By bypassing terrestrial infrastructure limitations, space-based cellular broadband changes the game for autonomous machines.
What are your thoughts? Are there other physical AI use cases where satellite-to-cellular connectivity becomes the ultimate enabler, or am I overestimating the need for constant off-grid broadband?