What Samsara actually does

Samsara sells ruggedized IoT hardware — AI dash-cams, vehicle gateways, asset tags, environmental sensors — and then charges a software subscription to make sense of everything those devices see. The hardware is the hook; the software annuity is the business.

Founded in 2015 by Sanjit Biswas and John Bicket (the duo behind Cisco Meraki) and public since December 2021, Samsara targets the unglamorous, enormous world of physical operations: commercial trucking and last-mile fleets, field services, construction, logistics, utilities, manufacturing and local government. These are businesses that move atoms, not bits — and historically ran on clipboards, spreadsheets and disconnected legacy boxes. Samsara replaces that with a single platform where a fleet manager can see where every vehicle is, how it is being driven, whether the engine is about to fail, and whether a job got done — in one pane of glass, in real time.

From sensor to decision, in seconds

Every Samsara device is a data faucet. Edge hardware captures the signal, cellular pipes it to the cloud, AI models interpret it, and the output lands as a live map, a coaching nudge, a maintenance alert or an automated workflow. The loop is continuous — and the more it runs, the smarter the models get.

Four pillars, one platform

Customers usually "land" on one pillar to solve an urgent problem — compliance, an accident, a lost trailer — then "expand" across the others. Each module shares the same hardware, data layer and login, which is why attach rates and net retention are so high.

The AI dash-cam, explained

Dual-facing AI Multicam — computer vision runs on-device (edge AI) for instant alerts.

A Samsara dash-cam is really a computer with lenses. One camera faces the road, one faces the cab. On-device computer vision watches for distracted driving, phone use, missing seatbelts, tailgating and rolling stops — and reacts in the moment, not hours later in a report.

When an event fires, the driver gets an instant in-cab nudge, and the clip plus context flows to the manager's dashboard. Samsara Coach layers on an AI avatar that delivers personalized coaching based on each driver's own history, and lets managers rehearse safety conversations against an AI-simulated driver.

The moat is the data: trillions of road-miles of labeled video make the models hard to replicate. That same corpus is what makes Samsara's push into "Physical AI" — agents for safety, maintenance, compliance and dispatch — credible rather than aspirational.

Land cheap, expand forever

The economics are a classic razor-and-blades inversion: the hardware is sold near cost to get installed, and the recurring software subscription — ~98% of revenue — is where the value compounds. Because every new module rides the hardware and data already in place, expansion is cheaper than acquisition.

Anywhere atoms get moved or built

Samsara's stated addressable market is $96B+ and it has penetrated under 2% of it. The customer base spans the industries that run the physical world — and includes over 200,000 International-brand vehicles that now ship pre-installed with Samsara devices.

From dashboards to "Physical AI"

Samsara's next act is to become the orchestration layer between human-operated fleets, autonomous vehicles and robotics — the connective tissue that lets AI agents act on the physical world. Its agents for safety, compliance, maintenance and dispatch begin broad rollout in summer 2026.

The exhaust of operations
becomes the fuel for AI.

Every mile driven, every harsh-brake event, every engine fault and completed work order is a labeled training signal. That compounding dataset — not any single sensor — is the durable advantage.