Relativity Networks Secures $22 Million to Revolutionize Data Center Connectivity with Advanced Fiber Technology

Developers of data centers are projected to invest up to $4 trillion by the decade’s end, facing significant limitations due to political and power-grid factors in site selection. While fiber speed is typically considered a constant, one company believes that advancements in fiber technology could transform the geographic landscape of data center construction.

On Tuesday, Relativity Networks revealed that it has secured $22 million in funding through SAFE notes from various investors, including Rhapsody Venture Partners, Bell Ventures Inc., and Faster Than Glass LLC. SAFE notes, which convert investment into equity after a company’s first priced round, are commonly used in early-stage funding. Additionally, the company received a $40 million follow-up order from a prominent, unnamed hyperscale client.

Relativity Networks specializes in hollow-core fiber, an innovative technology that facilitates data transmission at speeds 30% faster than standard fiber. Unlike traditional fiber, which conducts light through fiber-optic glass, hollow-core fiber uses a vacuum chamber at its core, bringing it closer to the maximum possible speed of light.

This speed difference equates to a matter of microseconds. According to CEO Jason Eisenholz, a signal travels approximately five microseconds per kilometer with conventional fiber, whereas hollow-core technology reduces that time to roughly three and a half microseconds.

In earlier AI computing setups centered on single racks of GPUs, fiber latency was less of a concern. However, with expanding scale and physical distances between GPUs increasing, it’s now typical for a data center campus to extend across hundreds of acres and multiple buildings. Eisenholz identifies a significant opportunity for multi-campus arrangements, where existing data centers are interconnected to function as a cohesive system.

“The largest systems distribute computing across multiple campuses to harness available power,” he explained. “They’re relocating to more efficient spaces but still need to function as a synchronized entity.”

This development offers a potential solution to the geographical challenges that have limited many ongoing data center projects. Reducing latency by 30% allows developers to expand operations over 30% greater distances without encountering latency issues. As computational demands continue to grow, Eisenholz believes this could signify a pivotal change in the industry.

“The first phase of AI focused on compute power,” he stated. “It was all about GPU performance. The next phase has been about optimizing networking within data centers to leverage that power. The emerging phase we anticipate is optimizing geographic location.”