The city remains ground zero for the industry’s explosive growth. But as hardware companies scale up, they are hitting a hard physical limit: a lack of industrial space.
San Francisco remains the undisputed epicenter of the AI robotics boom. But as these companies transition from scrappy prototypes to industrial-scale manufacturing, they are confronting a harsh physical reality: the city simply does not have the space to hold them.
For nearly eight years, San Francisco served as the perfect proving ground for Bright Machines. The robotics company first planted its roots in a South of Market (SoMa) loft, initially aiming to develop software and technology for other manufacturers. However, as the AI boom created an insatiable demand for specialized chips and servers, Bright Machines pivoted, becoming a manufacturer itself.
In 2023, the company moved into a 10,000-square-foot industrial building in the Inner Mission. Here, it could build and test the robots that now assemble hardware for data centers. Because its machines can produce custom designs far faster than traditional hardware manufacturers—some of which take weeks just to reconfigure an assembly line and retrain staff—Bright Machines began shipping finished products directly out of its hybrid office-and-workshop, located just up the street from Dandelion Chocolate.
“There is a good energy and vibe to this area,” said Sviat Dulianinov, CEO of Bright Machines. “Especially for young engineers. They can walk outside and encounter so many other people working on similar things.”
Indeed, as AI applications push further into the physical world, the neighborhood around Bright Machines’ headquarters has become a hotbed for hardware innovation, buoyed by its proximity to central freeways. Firms like Mira Murati’s Thinking Machines Lab, Weights & Biases (recently acquired by CoreWeave), and Physical Intelligence are all just steps away from Dulianinov’s team.
That kind of dense, scrappy setup worked perfectly while Bright Machines was still proving its technology. But with larger orders now pouring in, the company is hiring staffers across sales, operations, finance, and product management every month. The result? There is no longer room to put additional desks.
This fall, Bright Machines will leave San Francisco for Burlingame, where it has leased a space five times larger. “We would have loved to stay here,” Dulianinov noted. “But now we need to focus more on production.” He also views the move to the Peninsula as a strategic opportunity to recruit a wider, more diverse range of engineering talent.
This pattern is poised to play out across the broader AI robotics scene as long as investment dollars continue to flow. Companies that thrive in San Francisco’s dense startup ecosystem are increasingly finding that large-scale industrial space and heavy infrastructure are hard to come by within city limits.
Take Jeff Bezos’ stealth AI startup, Project Prometheus. After establishing an office downtown, the company spent the early part of this year hunting for approximately 100,000 square feet of industrial space. After striking out on a warehouse on 23rd Street in the Dogpatch neighborhood in February, it ultimately signed a lease at the old American Steel Blocks site in West Oakland.
Still, the hype surrounding robotics in San Francisco is entirely justified. According to commercial real estate firm JLL, the amount of commercial space leased by robotics companies this year alone has already surpassed the cumulative total from 2020 to 2025. This surge is driven primarily by the growth of autonomous vehicle companies. While most of these firms require dedicated R&D and industrial space, a third of the leases signed also included traditional office footprints.
However, the biggest robotics deals are increasingly funneling toward the East Bay or Silicon Valley. The latter now stretches as far south as Morgan Hill, where Archer Aviation recently leased more than 500,000 square feet at the newly built Cochrane Technology Center. In fact, of all AI robotics-related deals JLL has tracked since 2017, San Francisco has accounted for only about 11% of the total square footage leased in the Bay Area.
For Bright Machines, the transition will be a careful handover. The company intends to continue working and manufacturing out of its 16th Street headquarters until it officially hands over the keys. The impending move has elicited mixed reactions from staff. Some workers lament leaving San Francisco, where they have grown accustomed to dense, walkable neighborhoods. Others, like robot perception engineer Tori Colthurst, say they are keeping an open mind about the relocation.
It has been a dizzying journey to growth for the company. Bright Machines attempted to go public via a SPAC merger in 2021 before pulling out of the deal and raising additional venture capital. Now, as the build-out of “AI infrastructure” becomes a top mandate for the country’s biggest investors, the company is uniquely positioned to become a vital cog in the industry, especially as advanced chip manufacturing continues to be on-shored.
“Everyone wants to deploy new servers into data centers faster,” said Fiaz Mohamed, chief growth officer at Bright Machines. “The technology is getting pushed like it’s never been before, and our customers are having to refresh their products more often.”
Today, Bright Machines develops and ships AI-enabled robots, but it also sells a data platform and design software that allow customers to swiftly operationalize new designs in their own factories. Its robotic arms execute a wide range of precise movements, utilizing cameras and infrared energy to scan for microscopic defects on circuit boards.
The company currently manufactures and ships custom products to order from its factory lines in both Burlingame and Guadalajara, Mexico. According to Mohamed, Bright Machines’ robots are already deployed at more than 150 sites worldwide.
“The game for us now is all about scaling,” Mohamed said. “We’ve proved that we can adapt to new product requirements fast. But can we do that across hundreds, maybe even thousands of lines? It’s really hard.”
As the AI hardware revolution matures, the companies that survive will be those that can successfully navigate this exact challenge: bridging the gap between Silicon Valley innovation and industrial-scale reality.