BrightAI bets sensors can retire a 2,000-year-old maintenance habit
The Khosla-backed physical AI startup wants utilities and pipeline operators to stop inspecting on a schedule and start monitoring all the time
Alex Hawkinson keeps Roman artifacts at BrightAI’s robot demo center. They aren’t there for decoration. They’re there to make a point.
The point is that the people who run power grids, water systems and pipelines still check their assets the way Roman road crews did: on a calendar. Hawkinson’s San Francisco company is betting that habit is finally ready to die.
“The Romans are laughing at us.”
That’s Hawkinson’s line, and he means it more as diagnosis than insult. He told Fortune that ancient Rome dispatched technicians on a fixed schedule to look over its infrastructure. Two millennia later, modern utilities operate on much the same principle.
The math, as Hawkinson tells it, is unflattering. Roughly two of every three scheduled inspection trips find nothing that needs repair. Meanwhile, the breakdowns that actually matter tend to happen in the gaps between visits.
What BrightAI actually builds
BrightAI describes itself as a physical AI company. That’s a growing label for firms that apply machine learning to things in the real world, like pipes, transformers and pumps, rather than to text or images on a screen.
Its core product is a sensor-driven platform, known as Stateful, that watches infrastructure continuously. The company uses peel-and-stick sensors, drones and multimodal AI to keep a running read on asset health.
The scale is already meaningful. BrightAI has 250,000 sensors deployed across its customer base, according to the research findings.
AI, tech, and the markets they move—in one daily briefing.
Daily. Free. Join 34,000+ readers across crypto, finance, and policy.
The peel-and-stick approach matters more than it might seem. Retrofitting old infrastructure is usually where these projects go to die. A sensor you can slap onto an existing asset sidesteps a lot of expensive rewiring and downtime.
The money behind the machines
BrightAI’s financial story is a bit unusual for a venture-backed AI company. It spent several years bootstrapping before taking outside capital in a big way.
During that stretch, the company reportedly built revenue to somewhere between $80 million and $100 million. It did so with just seven major enterprise customers.
The company later closed a $51 million Series A, around mid-2025. That round brought its total funding to roughly $78 million. Post-funding, BrightAI was valued at approximately $265 million to $300 million.
Backers include Khosla Ventures, Inspired Capital and Upfront Ventures.
A founder who has done this before
Hawkinson founded BrightAI in 2019. Before that, he co-founded SmartThings and served as its CEO.
SmartThings was a smart-home platform, built around connecting household devices and sensors to a central hub. BrightAI is in many ways the industrial sequel: same core idea of networked sensors feeding software, applied to assets where a failure can mean a burst water main instead of a lightbulb that won’t turn on.
What this means for utilities and the AI market
The prize BrightAI is chasing is enormous. The research findings place the global infrastructure management market at more than $4.5 trillion, describing that as the amount spent each year on managing infrastructure.
There are real risks to watch. Revenue concentrated among seven enterprise customers means losing even one would sting. Utilities are also famously cautious buyers, with long procurement cycles and strict regulatory oversight, which can slow expansion.
Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.

