Relativity Networks Secures $22M for Ultra-Fast Data Center Fiber

The burgeoning data center sector is projected to invest up to $4 trillion by the decade's close, yet its expansion is significantly hampered by geopolitical and energy infrastructure limitations regarding suitable building sites. While the speed of fiber optic networks is often taken for granted, one innovative company is challenging this assumption, proposing that enhanced fiber performance could fundamentally alter the geographic calculus of data center development.

Relativity Networks recently announced a successful $22 million SAFE note funding round, attracting investments from Rhapsody Venture Partners, Bell Ventures Inc., and Faster Than Glass LLC, among others. A SAFE note, a common instrument for early-stage funding, converts into a specified number of shares upon the company's first priced equity round. This funding success was further bolstered by a substantial $40 million follow-on order secured from a prominent, yet unnamed, hyperscaler.

At the core of Relativity Networks' innovation is hollow-core fiber, a sophisticated but rarely deployed technology that accelerates data transmission by 30% compared to conventional fiber. Unlike traditional fiber optics, which guide light through a glass medium, hollow-core fiber channels light through a vacuum chamber at its center, pushing closer to the theoretical maximum speed of light.

This advancement translates into a critical reduction in latency, measured in microseconds. CEO Jason Eisenholz illustrates this by estimating that a signal traversing one kilometer in conventional fiber takes approximately five microseconds; with the adoption of hollow-core technology, this duration is dramatically cut to just three and a half microseconds.

In the early days, when AI computation was confined to a single rack of GPUs, fiber latency was largely negligible. However, as AI scale has surged, so too has the physical separation between GPUs. Today, it is not uncommon for a data center campus to spread across hundreds of acres and encompass dozens of buildings. Eisenholz identifies a significant opportunity in multi-campus deployments, enabling geographically dispersed data centers to function cohesively as a unified entity.

“The largest systems are distributing the compute across multiple campuses to reach the power that exists,” Eisenholz explained to TechCrunch. “They’re moving to where the warm shell is, but they still need to operate as one synchronized machine.”

This innovation offers a partial remedy to the stringent spatial demands that have constrained numerous ongoing data center expansions. In terms of latency, a 30% reduction effectively grants developers the flexibility to bridge 30% greater distances before latency becomes a prohibitive factor. As compute projects continue their trajectory of increasing scale, Eisenholz anticipates this could herald a profound transformation for the industry.

Reflecting on the industry's evolution, Eisenholz stated, “The first era of AI optimized for compute. It was GPU, GPU, GPU. The second era optimized the networking inside the data center to take advantage of that compute. The third era that we see coming is optimizing the geography.”