Relativity Networks Raises $22 Million to Revolutionize AI Fiber Speed
Relativity Networks has secured $22 million in funding and landed a significant order from an unnamed cloud giant for its innovative hollow-core fiber technology. This fiber carries light through air, achieving nearly the speed of light, addressing a crucial constraint in the expansion of AI data centers.
The Investment and Order Details:
- Funding: Relativity Networks raised the funds via a Safe (Simple Agreement for Future Equity), converting into shares at a later priced round. New investors include Rhapsody Venture Partners, Bell Ventures, and Faster Than Glass.
- Order: The company received a $40 million order from a major cloud provider, following successful testing of its ChronoCore technology. This order signifies a real-world validation of their hollow-core fiber’s capabilities.
Unlocking the Speed of Light:
Relativity Networks offers hollow-core fiber, contrasting conventional glass fibers that transmit light through solid material. Their approach allows light to travel closer to its speed in a vacuum, achieving approximately 47% (or up to 50%, per TechCrunch) faster speed compared to traditional fiber.
Dense Cable Deployment:
In collaboration with Prysmian, a leading fiber manufacturer, Relativity has developed a dense cable with 24 fibers in a single 10-millimeter cable. This compact design, tested with Dura-Line, ensures the technology is ready for practical deployment.
Addressing AI Data Center Constraints:
The company attributes the shift to a new era of "AI Geography" where the distance between chips becomes a significant constraint. With AI systems now spanning multiple sites to access available power, the delay caused by light’s speed in fiber becomes a critical issue. Relativity refers to this cost as the "Propagation Tax."
“AI is no longer scaling inside a data center. It is scaling across geography.” – Jason Eichenholz, CEO of Relativity Networks
Relativity Networks aims to overcome these geographical limitations by enabling AI systems to behave cohesively despite scattered power sources over vast distances.