What is hollow-core fiber and how does it work?

Hollow-core fiber uses a vacuum chamber at its center instead of glass, transmitting data closer to the speed of light in a vacuum for a 30% speed increase over conventional fiber.

Why is a hyperscaler investing in Relativity Networks?

An unnamed leading hyperscaler placed a $40 million follow-on order, indicating a belief that hollow-core fiber is ready to address AI infrastructure latency bottlenecks.

How does hollow-core fiber reduce latency for AI data centers?

It cuts the travel time for a signal over one kilometer from roughly five microseconds in conventional fiber to about three and a half microseconds, enabling larger, geographically dispersed compute clusters.

What problem does hollow-core fiber solve for data center developers?

By reducing latency by 30%, it allows developers to build data centers over roughly 30% larger distances, mitigating constraints from power grid capacity, land availability, and politics.

How does Relativity Networks define the third era of AI infrastructure?

It's an era of 'geography optimization,' where faster fiber makes physical distance less critical, allowing multiple sites to function as a single campus and decoupling compute from strict location constraints.

Hyperscaler Bets Millions on Hollow-Core Fiber for AI

How Hollow-Core Fiber Rewires Data Center Physics

The core proposition is a 30% speed increase in data transmission. Where standard fiber optic cables transmit light through glass, hollow-core fiber uses a vacuum chamber at its center. This simple physical shift brings the signal closer to the theoretical speed of light in a vacuum.

The raw numbers are about microseconds, but at AI scale, they translate to miles. CEO Jason Eisenholz estimates a signal takes roughly five microseconds to travel one kilometer in conventional fiber. Hollow-core fiber cuts that to three and a half microseconds. As AI compute clusters expand from single racks to campuses spanning hundreds of acres, that latency difference becomes a critical constraint on design.

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

XOOMAR Analysis: The $40 million order is the story here, not just the VC raise. It shows a hyperscaler is willing to pay a premium now to solve a tangible bottleneck. This moves hollow-core fiber from lab curiosity to a deployment-ready tool for the largest AI builders.

Why 30% Less Latency Equals 30% More Location Options

The funding arrives as data center developers face a brutal site-selection equation. They are expected to spend up to $4 trillion by the end of the decade, yet are constrained by politics, power grid capacity, and land availability. Faster fiber rewrites that equation by making distance less punishing.

In latency terms, a 30% reduction allows developers to span roughly 30% larger distances before communication delays between computing nodes become a problem. This isn't about making a single data center faster; it's about enabling a new architectural model where multiple, geographically separated data centers can function as a single, logical "campus." It partially decouples compute from the strictest power and land constraints, a shift as we explored in our coverage of the FBI's $88 Million AI Deal: Hardware Buys Replace Cloud Dependence.

Scalability: From single rack to multi-campus grid. Constraint Alleviated: Physical proximity for low latency. New Flexibility: Build where power/land is available, not just where it's closest.

The Three Eras of AI Infrastructure, According to Relativity

Eisenholz frames this as a fundamental industry evolution. “The first era of AI optimized for compute,” he said. “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.”

This "geography optimization" era is about treating physical distance as a software-definable parameter. If you can make the fibers between sites faster, you effectively make the sites closer. This could enable more efficient use of existing "warm shell" buildings or allow new builds in previously untenable locations, mitigating some of the extreme physical demands of AI expansion highlighted in events like the Amazon Bulldozes Rare Books for AI-Fueled Data War.

What Happens After a $22 Million SAFE Note?

The funding mechanism itself is a standard SAFE note, which will convert into equity at Relativity's first priced funding round. It provides runway to scale production and fulfill its landmark hyperscaler order. The real test begins now: can Relativity manufacture hollow-core fiber reliably and at a cost that justifies its performance premium for broad deployment?

The forward look is about scaling a physics breakthrough into an industrial supply chain. The hyperscaler order is a powerful proof of concept, but the next milestone is demonstrating that this technology can move from custom installations to a standard, deployable product. If Relativity succeeds, it won't just sell faster cable; it will sell data center developers something more valuable: more places to build.

Impact Analysis

A $40 million hyperscaler order validates hollow-core fiber as a commercially viable solution, moving it from lab testing to real-world deployment.
The 30% speed increase directly addresses the latency bottleneck in distributed AI compute clusters, enabling larger, more efficient systems.
This funding and order signal a shift in data center infrastructure investment, prioritizing physics-level improvements to support future AI scale.