Subvolts take Physical AI is transforming industries by optimizing operations and coaching workers, moving beyond screen-based applications. Dec 16, 2025
For you
Developers
Learn about the challenges and opportunities of building and deploying AI models on resource-constrained edge hardware.
Business
Discover how AI is driving efficiency and new capabilities in asset-heavy industries like logistics, construction, and field service.
Hype note
3/5Notable

This is a solid interview with a CEO scaling AI in the physical world, offering practical insights into a major AI application area.

Training Data Episode

Why the Next AI Revolution Will Happen Off-Screen: Samsara CEO Sanjit Biswas

38 min Interview Hosts: Sequoia Capital
0:0038:21

Audio streams from Sequoia Capital. Your place is remembered on this device.

From Subvolts

Summary and analysis by Subvolts editorial, AI-assisted.

Summary

Sequoia Capital's Training Data podcast features Samsara CEO Sanjit Biswas discussing AI's next revolution happening off-screen. Biswas explains how physical AI differs from cloud AI, focusing on edge devices, real-world data diversity, and new capabilities like video reasoning. He highlights how AI is improving frontline worker performance in asset-heavy industries.

Explained simply

This episode explores how AI is being used in the real world, like in trucks and construction sites, not just on computers. It explains how AI can help workers do their jobs better and make things more efficient, even with messy, real-life data.

Why listen

This episode offers insights into the practical application of AI beyond screens, a crucial area for industries like logistics and construction. It's for business leaders and technologists interested in how AI drives efficiency in physical operations. The discussion on distributed edge compute and foundation models for video reasoning connects to ongoing research and development in multimodal AI by frontier labs.

On the mic
  • HostSequoia Capital
  • GuestSanjit BiswasCEO, Samsara
About
Key facts
  • Guest: Sanjit Biswas, CEO of Samsara
  • Show: Training Data (Sequoia Capital)
  • Length: 38 minutes
  • Samsara's scale: Captures 90 billion miles of driving data annually
  • Samsara's market cap: $20B+ public company
Show notes from Sequoia Capital

Sanjit Biswas is one of the rare founders who has scaled AI in the physical world – first with Meraki, and now with Samsara, a $20B+ public company with sensors deployed across millions of vehicles and job sites. Capturing 90 billion miles of driving data each year, Samsara operates at a scale matched only by a small handful of companies. Sanjit discusses why physical AI is fundamentally different from cloud-based AI, from running inference on two- to ten-watt edge devices to managing the messy diversity of real-world data—weather, road conditions, and the long tail of human behavior.

He also shares how advances in foundation models unlock new capabilities like video reasoning, why distributed compute at the edge still beats centralized data centers for many autonomy workloads, and how AI is beginning to coach frontline workers—not just detect risk, but recognize good driving and improve fuel efficiency. Sanjit also explains why connectivity, sensors, and compute were the original “why now” for Samsara, and how those compounding curves will reshape logistics, field service, construction, and every asset-heavy industry.

Hosted by Sonya Huang and Pat Grady, Sequoia Capital


Jargon decoded
inference
Sanjit Biswas
The process of an AI model making a prediction or decision based on new data.
edge devices
Sanjit Biswas
Small, low-power computing devices located close to where data is generated, like in vehicles.
foundation models
Sanjit Biswas
Large, general-purpose AI models that can be adapted for various specific tasks.
video reasoning
Sanjit Biswas
The ability of AI to understand and interpret the content of video footage.
Published
Show
Training Data · Sequoia Capital
Length
38 min
Summary by
Subvolts, using an AI model (how we work). Spotted a mistake? Tell us.

Same guests, other shows

Questions people ask

What is physical AI?
Physical AI refers to AI applications that operate in the real world, often using sensors on devices like vehicles or job sites, rather than solely in cloud-based systems.
How is AI improving frontline workers?
AI can coach frontline workers by detecting not only risks but also recognizing good performance, such as improving driving habits for better fuel efficiency.

Page generated Oct 19, 2026. Summaries are Subvolts' own; the audio, artwork and notes belong to Sequoia Capital.