Relativity Networks Raises $22 Million to Improve Fiber Connectivity for Data Centers

Data center developers are projected to invest up to $4 trillion by the decade’s end, encountering substantial challenges from political and power-grid influences impacting their construction sites. While many assume that fiber speed is consistent, one company asserts that advancements in fiber technology could reshape the geographic landscape of data center development.

On Tuesday, Relativity Networks announced it has secured $22 million through SAFE note funding, with backing from investors like Rhapsody Venture Partners, Bell Ventures Inc., and Faster Than Glass LLC, among others. A SAFE note allows for the investment to convert into a predetermined number of shares upon the company’s completion of its initial priced round, making it a popular financing method for pre-seed and seed funding. Furthermore, the company has received a $40 million follow-on order from a leading hyperscaler, although the client’s identity remains confidential in this report.

Relativity Networks is concentrating on hollow-core fiber, an underexploited technology that enables data transmission at speeds 30% faster than conventional fiber optics. In contrast to traditional fiber, which conveys light through glass, hollow-core fiber transmits light through a vacuum core within the cable, closer to the ultimate limit of light speed.

The speed difference is measured in microseconds. CEO Jason Eisenholz estimates that a signal takes roughly five microseconds to traverse one kilometer in standard fiber, a duration that can be cut down to approximately three and a half microseconds with hollow-core technology.

In its initial stages, when AI processing was confined to a single rack of GPUs, fiber latency was not a pressing concern. However, as scaling has progressed, the physical distance among GPUs has increased significantly. Currently, it is common for a data center campus to span hundreds of acres across multiple buildings. Eisenholz envisions substantial possibilities in multi-campus setups, where separate data centers are linked to operate seamlessly as a single entity.

“The largest systems are distributing compute across multiple campuses to utilize the available power,” he informed TechCrunch. “They are relocating to areas with the necessary infrastructure while still needing to function as a cohesive, synchronized system.”

This innovation could alleviate the strict spatial constraints that have impeded many data center initiatives. In terms of latency, a 30% reduction in time allows developers to extend operations over distances 30% further before encountering latency challenges. As computational requirements continue to escalate, Eisenholz believes this could be a pivotal moment for the industry.

“The first era of AI centered on compute,” he remarked. “It prioritized GPU, GPU, GPU. The second era improved networking within data centers to optimize that compute. The next era will focus on optimizing geographic placement.”