Analysis
Relativity Networks, an Orlando-based fiber-optics startup, raised $22 million in SAFE-note funding from Rhapsody Venture Partners, Bell Ventures Inc. and Faster Than Glass LLC, among others, TechCrunch reported Wednesday. The round, structured as a simple agreement for future equity rather than a priced round, closed at more than double the company's original fundraising target.
The physical-layer bet
Founder and CEO Jason Eichenholz built Relativity around a specific physical-layer problem: conventional fiber-optic cable carries light through solid glass, which is slower than light traveling through air. Relativity's patented hollow-core fiber, branded ChronoCore, guides light primarily through an air channel instead, carrying data roughly 47% faster than glass fiber for a given distance. In practical terms, that means two data centers connected by ChronoCore can sit farther apart geographically while staying within the same latency budget a shorter run of conventional fiber would require -- a meaningful constraint-loosener for AI training clusters that increasingly span multiple physical sites and need low-latency interconnects between them to function as one logical system.
“With just $22 million raised and one disclosed hyperscaler order, Relativity remains an early-stage bet on a specific physical-layer thesis.”
Real commercial validation, not just a funding round
Beyond the equity raise, an unnamed hyperscaler placed a $40 million follow-on order after successfully testing ChronoCore to connect two of its data centers -- a customer commitment that's larger in dollar terms than the funding round itself, and a rare instance of a hardware startup landing paying commercial revenue this early rather than purely investor capital. Relativity has not disclosed which hyperscaler placed the order, which limits how much outside observers can independently verify the deal's scale or exclusivity, but a follow-on order specifically (rather than an initial pilot purchase) suggests the first deployment performed well enough to justify expansion.
Where this fits in AI infrastructure
As AI training runs scale past what a single data-center campus can power, cluster-to-cluster interconnect speed has become a genuine bottleneck alongside the more heavily covered constraints of chip supply and electricity. Relativity's pitch competes less directly against chip or GPU vendors and more against incumbent fiber providers like Corning, whose glass-fiber products remain the default choice across the industry -- meaning Relativity's actual competitive test is convincing hyperscalers used to a mature, well-understood glass-fiber supply chain to adopt a newer, less-proven physical medium at scale.
With just $22 million raised and one disclosed hyperscaler order, Relativity remains an early-stage bet on a specific physical-layer thesis. The company hasn't disclosed manufacturing capacity or unit economics for ChronoCore at scale, both of which will determine whether hollow-core fiber becomes a meaningful alternative to glass across the AI data-center buildout or stays a niche solution for latency-critical interconnect segments specifically.
Why a SAFE note instead of a priced round
Structuring the raise as a simple agreement for future equity rather than a priced round is itself a signal worth noting: SAFE notes typically move faster and avoid the valuation-setting negotiation a priced round requires, which fits a company that had just landed a large hyperscaler order and wanted to close financing quickly rather than spend weeks on term-sheet negotiation. It also means Relativity's actual valuation remains undisclosed, unlike the funding rounds covered elsewhere in this issue -- a data point outside observers can't yet use to benchmark the company against comparable physical-layer infrastructure startups.
The broader capital intensity of the AI buildout is what makes a company like Relativity investable in the first place: Starcloud's $250 million raise this same week for orbital compute and Muon Space's parallel Series C both reflect the same underlying dynamic -- physical infrastructure constraints, whether interconnect speed, power, or orbital real estate, are becoming as investable a category as the AI models themselves. Relativity's specific bet is narrower than either of those, but it's targeting a bottleneck that scales directly with how many physically separate data centers hyperscalers need to link together as AI training clusters keep growing.
