Hollow-core fiber could reshape where data centers are built, Relativity Networks says
Relativity Networks raised $22 million and landed a $40 million order as hollow-core fiber promises about 30% lower latency, potentially letting data centers span substantially greater distances.
Relativity Networks announced a $22 million SAFE note financing and a separate $40 million follow-on order while pitching hollow-core fiber as a way to change the geography of large-scale data center deployments. The company and investors say the faster transmission could relieve some of the strict siting pressures that have steered the multitrillion-dollar buildout. If adopted at scale, the technology could let operators place compute farther apart without the latency penalties that now constrain campus design.
Funding round and hyperscaler order
Relativity Networks said the $22 million SAFE note round included participation from Rhapsody Venture Partners, Bell Ventures Inc., Faster Than Glass LLC and other investors. SAFE notes convert to equity at a later priced round, a common instrument for early-stage startups.
Separately, the company reported a $40 million follow-on purchase from a major hyperscaler that requested anonymity. Relativity described the order as validation that hyperscale operators are testing alternative fiber types to meet growing interconnect and latency demands.
How hollow-core fiber reduces signal delay
Hollow-core fiber carries light through an air-filled or vacuum center rather than through solid glass, which reduces material delay and brings optical transmission closer to the theoretical limit. That structural difference can cut propagation time by roughly 30% compared with conventional fiber.
Relativity’s chief executive provided concrete figures: where signal transit in standard fiber averages about five microseconds per kilometer, hollow-core fiber can reduce that to roughly three and a half microseconds per kilometer. Though microseconds sound small, they compound across many kilometers and hundreds of cross-links inside a sprawling data campus.
Why lower latency matters for multi-campus data centers
Historically, compute-heavy workloads could be consolidated within a single rack or building where fiber delays were negligible. As AI training and distributed workloads have grown, operators are increasingly linking multiple data center campuses to aggregate power and capacity.
Reducing per-kilometre latency by about 30% gives system architects more physical room to distribute compute while preserving synchronization. That shift could alter choices about where to locate new facilities, especially in regions where power availability and permitting have limited site options.
Data center buildout pressures and the $4 trillion figure
Developers are projected to spend as much as $4 trillion on data center infrastructure by the end of this decade, and many of those investments are already constrained by grid capacity, permitting and local politics. Those constraints have concentrated builds in specific corridors and regions, increasing competition for land and power.
Faster fiber does not eliminate the need for electricity or approvals, but it changes a key technical trade-off. By allowing operators to span larger distances without compromising latency, hollow-core fiber could make previously marginal sites — with adequate power but farther from core campuses — viable components of a unified compute footprint.
Barriers to adoption and industry considerations
Despite the latency gains, hollow-core fiber faces practical hurdles before becoming pervasive. Manufacturing capacity, higher production costs, connector and splice compatibility, and field deployment practices differ from those for standard single-mode fiber. Each of these factors will shape the pace at which operators integrate hollow-core into critical paths.
Network operators must also weigh the economics: for some workloads the incremental latency improvement won’t justify retrofit or new-route expenses. For latency-sensitive applications — high-frequency trading, distributed AI training, and certain real-time services — the business case is stronger, which may drive early adopter momentum among hyperscalers and cloud providers.
Operators will monitor the performance of early deployments, interoperability with existing fiber infrastructure, and whether suppliers can scale output and reduce unit costs. If those boxes are checked, hollow-core deployments could move from specialized links to broader backbone uses over the next several years.
Industry executives say the shift mirrors a progression in which compute density, internal networking, and now geography are all being optimized in turn. In that framing, hollow-core fiber is not a silver-bullet solution but a technical lever that, combined with evolving power strategies and site selection, can expand the feasible footprint for next-generation data centers.
Relativity Networks’ financing and the large hyperscaler order mark an early commercial test for hollow-core fiber, and observers will be watching whether the technology spreads beyond niche, latency-critical routes into mainstream data center interconnects. The coming quarters should show whether the latency improvements translate into measurable cost savings and new siting flexibility for large-scale compute operators.
