Management at the Bay Area transportation startup Glydways wants you to be clear about what the company is not: It may plan to move people in futuristic autonomous pods, but it’s not hyperloop-grade vaporware. And its funding by big-name Silicon Valley investors does not make it a ride for the 1%.
“Public transit for everyone, everywhere,” says founder Mark Seeger.
But Glydways is starting smaller than that. Its first green-lit project (after a temporary test track now under construction next to an abandoned mall in Richmond, Calif.) and others under consideration by local governments will have Glydways’s four-seat electric vehicles plying short on-demand routes between existing business and transportation hubs.
That debut pilot effort—a half-mile route linking a convention center and arena to the last stop on a people mover outside Atlanta’s Hartsfield-Jackson International Airport—is on a small enough scale to evoke the Vegas Loop that Elon Musk’s Boring Company opened as a shortcut between three of the halls of the Las Vegas Convention Center.
“We want to see how well the system operates with various fluctuations of riders showing its ability to scale and that it is indeed a viable transit option,” says Krystal Harris, program director for ATL Airport Community Improvement Districts.
After two years of free-fare service, that agency and the Metropolitan Atlanta Rapid Transit Authority will assess how things worked and if the technology merits expansion.
Putting a cap on capex
The $18 million in construction and operational costs that Harris cited may seem steep for such a short distance, but not in the context of U.S. transit construction expenses that have made the country exceptional in the wrong way.
For example, the $3 billion Silver Line extension that Washington’s Metro system opened to Washington Dulles International Airport and beyond in late 2022 cost $263 million a mile, including a large rail yard built by the airport.
That, however, looks like an outright steal next to other U.S. rail projects, topped by the Long Island Rail Road’s East Side Access project in New York and its $3.5 billion a mile expense.
Glydways, meanwhile, touts a design for simple, narrow guideways that require neither rails nor electric power via overhead wires or third rails that it says will cost 90% less than traditional transit.
“We can do it for tens of millions,” Glydways CEO Gokul Hemmady says, adding that at-grade costs could run still lower at just $2 to $3 million per mile while elevated paths needed to avoid grade crossings could run $15 million a mile.
“The moment you’re in pedestrian-class infrastructure, your costs plummet,” he says. “The world knows how to build this.”
Construction costs of some recent U.S. cyclist and pedestrian infrastructure fall roughly into that minor-league ballpark.
A trail being built along the SMART commuter-rail line in Sonoma and Marin counties in California has run about $4 million a mile. Two bridges on the Washington & Old Dominion Trail constructed over wide roads in Arlington and Fairfax counties in Virginia had project costs around $30 million a mile.
But a veteran transit consultant who has led projects in North America, Europe, and Australia and New Zealand warned against expense extrapolation. Writes Jarrett Walker: “They will have to build out a demo project before we know.”
No human operators, some operating costs
The operational part of the Glydways pitch involves leveraging autonomous-vehicle advances to provide high-frequency, on-demand service around the clock at fares not that far above traditional transit—with the ability to transport 10,000 people an hour.
“We offer ride-hailing-like experience at a fraction of the cost,” says Hemmady. Pressed for an example, he cites the Oakland Airport Connector, an automated, elevated train that runs between that airport and Bay Area Rapid Transit’s Oakland Coliseum station for a one-way fare of $7.45.
But while those fares covered 96% of the connector’s operating costs pre-pandemic, Hemmady says Glydways’s lower costs—30% of other modes of transit, the company says—will let it clear a profit: “We are the only mass transit system that is revenue positive.”
A Glydways vehicle shown off at CES 2025 was all shiny modernity, with a streamlined exterior hiding camera, lidar and radar sensors, and large doors that slid open to reveal a clean plastic interior with tap-to-pay terminals by those doors. The closest visual parallel: the pod-like Zoox robotaxis now rolling around Vegas in test drives.
The lack of human operators or attendants has led some critics to raise safety concerns, but Glydways emphasizes that short waits at stations and the limited number of passengers per vehicle will keep it safe.
Older, almost as small-scale “personal rapid transit” systems built on older autonomous technology—such as one that runs between campuses of West Virginia University in Morgantown, W. Va.—have operated without incident for decades.
Larger automated-train systems rely on a combination of surveillance and patrolling. For example, Vancouver’s SkyTrain equips its driverless trains with emergency intercom systems and contact systems while having attendants and transit police at stations.
Next stops
After the Atlanta pilot, Glydways has advanced to final stages of consideration in a San Jose project to link that city’s Caltrain and Amtrak train station with San José Mineta International Airport—a 3.4 mile route that Google Maps estimates as a seven-minute drive but a 40-plus-minute transit adventure.
Glydways says it can build that mostly elevated route, with its vehicles taking eight minutes between the station and the airport, for under the city’s $500 million cost cap but isn’t specifying a cost estimate. The city council should be voting on its proposal, which allows for possible extensions to such nearby traffic generators as Apple’s headquarters, in the coming weeks.
This company has a comparable plan not far north of San Jose in Contra Costa County, where it’s pitched its technology as an automated transit network to provide transportation from train stations to nearby destinations.
And in the Los Angeles suburb of Ontario, Glydways has advanced a proposal to use its vehicles in a tunnel to connect Ontario International Airport with the closest Metrolink commuter-rail station. The Boring Company had earlier offered a version of the Vegas Loop concept but abandoned that bid in 2022.
Glydways’s proposition of robotic transportation has the advantage of not having to coexist with human-driven traffic like robotaxis like Waymo. And the company has the advantage of funding from such deep-pocketed investors as the VC firm Khosla Ventures and OpenAI CEO Sam Altman.
But in the realm of transit, self-driving technology isn’t something Glydways invented, and many transit agencies outside the U.S. already employ it on higher-capacity subway and light-rail lines.
And as autonomous mobility continues to advance on public roads, Walker suggests that established transit operators will be able to make better use of it than any startup that has to pour new concrete—even if the technology goes into something as unfancy as buses.
Says Walker: “If driverless technology becomes available, debugged, and socially accepted, there will be all kinds of applications, including much bigger-vehicle services that will be a better use of scarce space in dense cities.”
