Case Study Deskcasestudydesk.com
RoboticsSourced

Case Study: Physical Intelligence runs real-time robot inference remotely on Modal

Physical Intelligence Case StudySourced & dated by Case Study Desk
Key facts · TL;DR
Company
Physical Intelligence
Industry
Robotics
Challenge
Physical Intelligence needed real-time remote inference for a 24/7 robot fleet without onboard GPUs.
Headline result
Physical Intelligence added only ~10-15ms latency for remote robotic control inference

Key results

10-15ms
Added network latency
remote inference
<30s
Checkpoint loading time
<1 min
Container startup to reachability

The challenge

Physical Intelligence builds general-purpose robotic systems that need continuous real-time inference across a fleet running 24/7. Local inference requires an expensive onboard GPU per robot, while conventional cloud inference adds latency and request-response bottlenecks incompatible with real-time control loops.

The solution

Physical Intelligence partnered with Modal to build a QUIC-based portal over UDP with automatic NAT traversal, replacing TCP request-response patterns. Persistent bidirectional channels between robot runtimes and Modal GPU containers enable continuous observation-to-action streaming without head-of-line blocking.

The results, in context

Physical Intelligence reported that the approach added only about 10-15ms of network latency for remote inference. Checkpoint loading completed in under 30 seconds and container startup reached service reachability in under a minute.

Products used

Modal Modal GPU computeModal Serverless inference