Alex Hawkinson keeps a collection of Roman artifacts at BrightAI’s robot demo center, and he has a saying to go with them: “The Romans are laughing at us.” Think of the way the Romans inspected their roads, Hawkinson told Fortune: “They sent teams of technicians out periodically on schedule” to check on their infrastructure. Two thousand years later, power companies, water utilities, and pipeline operators still do. In his telling, two out of three of those trips turn up nothing to fix, and the failures that matter happen in between visits anyway. Hawkinson is convinced this can change, and investors are paying for the idea. BrightAI said Wednesday it has raised $97 million in a round led by Smith Point Capital, the growth-stage firm started by former Salesforce co-CEO Keith Block. GE Vernova, already a BrightAI customer, also invested, along with Cobalt Capital, Marlinspike Partners, Zero Infinity Partners as well as existing backers Khosla Ventures, Inspired Capital, and Upfront Ventures. The round brings total funding to $184 million and values BrightAI at $750 million, according to a person familiar with the matter. Hawkinson insisted he “doesn’t care” about so-called unicorn status: The point is bring infrastructure out of the ancient Roman era. His pitch has three parts. Replace scheduled human inspections with sensors and AI that watch infrastructure all the time. Go after what BrightAI estimates is more than $7 trillion a year in waste worldwide. And along the way, create a kind of data that has never existed: a continuous record of how pipes, poles, and pumps actually wear out. If it works, the companies running the world’s power and water get to stop paying for trips that find nothing. The Roman method The Romans knew the problem well. Frontinus, who was named Rome’s water commissioner by the emperor Nerva in the late first century A.D., wrote a treatise on running the city’s aqueducts. He worried about leaks in underground channels that were hard to find and fix, about people tapping the lines without permission, and about trees whose roots could damage the masonry. BrightAI’s release says the systems that power modern life “still operate on a centuries-old model: inspect, detect, and react.” Hawkinson is a believer from personal experience. In 2011, the power went out at his family’s cabin in Leadville, Colo., a remote mountain town, and nobody knew until the pipes had frozen and burst and water had run through the walls. He built a connected sensor hub in his garage, launched it on Kickstarter as SmartThings, and sold the company to Samsung in 2014. He has cofounded eight companies and run four as CEO, all of which exited. He bootstrapped BrightAI, founded in 2019, for its first four years. He spent the first half of his career in software as a service, and his SmartThings put sensors in people’s homes, not substations or sewers. “I wasn’t from infrastructure,” he said. When he started going out in the field, at one point in a hard hat in a sewer, he was “shocked” by how the systems everyone depends on are actually monitored and break down. In industry after industry, he asked frontline technicians how often they get sent out and find nothing to fix. The answer kept coming back at 50% to 70% of the time. BrightAI’s release puts it more conservatively, at more than half of scheduled inspections. He assumed at first the operators were doing it wrong. He came to see it as an equilibrium. Send crews less often, and something breaks catastrophically, so you send them more often. Send them more often, and the labor, which is expensive and often dangerous, becomes unaffordable. Operators settle where “two out of every three times they send somebody out, even the most talented operators, it’s unnecessary,” he said, and failures still slip through between visits. The wooden power pole is his favorite example. By his count, the U.S. has about 150 million of them, “little matchsticks sticking up,” and each one sees a human inspector roughly every eight to 10 years. The 2018 Camp Fire—California’s deadliest wildfire—killed 85 people after a worn hook on an aging PG&E transmission tower gave way. Cal Fire determined the utility’s lines caused it. Hawkinson’s point is that the hook was “completely invisible in the world” until it failed. $7 trillion of waste The $7 trillion figure is BrightAI’s own. The company says it surveyed 12,158 operators across 16 industries and put the addressable annual impact at more than $7.5 trillion. It also found 98% of those operators lack adequate visibility into their assets. It tallies the wasted trips, the labor, and the cost of letting things break, which he said is far more expensive to fix than to maintain. The outside research the release cites—from the Energy Department, the EPA, Swiss Re and the National Safety Council—puts the cost of infrastructure failure alone at $2.5 trillion a year. The examples Hawkinson gives are mundane, which is part of his point. He cited one customer, a UK water utility that has about a thousand pumps and can’t tell which ones will be the next to fail. Its limited staff runs around “like chickens with their heads cut off” trying to catch them. When heavy rain hits, failed pumps send raw sewage into rivers. Hawkinson said he’s a Lord of the Rings fan, and this breaks his heart, to see raw sewage being pumped into a landscape that looks “just like the Shire.” (This Fortune interviewer agreed, saying Hawkinson’s physical AI descriptions sound a bit like something Gandalf would get up to; Hawkinson laughed, but didn’t quite agree.) Closer to home, a break in the Potomac Interceptor sewer line in January released an estimated 243 million gallons of untreated wastewater into the Potomac River, according to the Maryland Department of the Environment. There’s also copious waste in replacing parts that are fine. Hawkinson described a customer COO who suspected his company was swapping out spark plugs in its machines too early, at a cost of tens of millions of dollars a year. Once BrightAI’s models had the sensor data, Hawkinson said, they found about half of the replacements were premature. The call depended on some 30 variables no technician would track. A new kind of data BrightAI’s platform, Stateful, pairs edge sensors on poles, pipelines, compressors, and tanks with autonomous robots, drones and wearable AI copilots for field crews. The company says it runs at more than 50,000 locations and processes tens of trillions of sensor readings. Customers include GE Vernova, Reworld, Osmose Utilities Services, Azuria Water Solutions, Kodiak Gas Services, and Diamondback Energy. On average, Hawkinson said, customers see trouble coming 51 days before what would have been a catastrophic failure. GE Vernova, whose equipment helps generate about one-quarter of the world’s electricity, wanted in. Hawkinson said BrightAI hadn’t planned to take money from strategic investors this round, but GE Vernova insisted. The release says it will use BrightAI across its Power, Wind, and Electrification businesses. Most of the new money goes to taking the platform global. BrightAI says nearly 60% of its addressable market is outside North America, and it’s opening offices abroad after doubling its headcount in 2025. Hawkinson said bringing on a new customer used to take two to three years, and now takes less than six months. The coming year, he said, is about going from “proven concept” to deployments at scale with the biggest utilities. Until then, most of America’s 150 million wooden poles will keep getting checked the Roman way, about once a decade.
‘The Romans are laughing at us’: physical AI founder raises $97 million in quest to stop wasting time with archaic infrastructure maintenance
Full Article
Original Source
Read the full article at Fortune →KhanList aggregates and links to publicly available news content. We do not host full articles from third-party sources. Always verify important information with original sources.