An AI data center needs more than powerful processors: those processors must keep communicating as workloads change and hardware fails. Spain's iPronics, which builds equipment for redirecting optical connections between them, announced a $125 million Series B on September 2 to expand commercial deployments.
Its main product, iPronics ONE, is a rack-mounted optical circuit switch. Instead of relying only on fixed connections, a data-center operator can use software to change which groups of processors are connected. The company supplies the switching hardware together with controls and monitoring interfaces, so engineers can configure paths and check optical signal levels as part of running the network.
One practical application is recovering from a failed computing chip. In an interview with Reuters, founder and chief technology officer Daniel Pérez-López described failures as frequent events in very large installations. The ability to route around them is therefore part of keeping a cluster useful, not simply an exercise in making individual links faster. Reuters reported that iPronics had been shipping switches for about a year and was working with large AI infrastructure providers whose names it did not disclose.
The company's approach uses silicon photonics, which puts optical functions on silicon chips. Its product documentation describes a solid-state design without moving mechanical switching parts and a dedicated manufacturing line at Fabrinet for assembly and testing. Reuters reported a goal of shrinking the equipment roughly twentyfold. That is a development ambition, not an independently demonstrated production result; the commercial question is whether smaller systems can also meet operators' reliability requirements at scale.
Maverick Silicon and Light Street Capital co-led the financing, with Nvidia among the participants. The company says its cumulative funding is now $177 million. Founded in 2019 as a spinout from Universitat Politècnica de València, iPronics has also opened a Santa Clara office to support its U.S. product and customer work.
Captables' read: this is a bet on making existing computing capacity easier to use. The money supports manufacturing, deployment and closer work with infrastructure operators. The next meaningful proof point is how the switching system performs in customers' operating networks—not simply how much funding the supplier can attract.