What's Happening?
DriveNets has announced the first commercial deployment of an AI supercluster using long-distance scale-across AI networking. This development, part of Project Redwood by WhiteFiber, connects two data centers over 50 miles apart into a single logical
GPU supercluster. The network, validated at 111.2 Tbps of bandwidth with 0.9ms latency, represents a significant advancement in AI infrastructure. The scale-across architecture allows for larger clusters and greater resiliency by overcoming power and space constraints at single sites, enabling distributed GPUs to operate as a unified system.
Why It's Important?
This innovation addresses a critical challenge in AI infrastructure: the power and space limitations of single-site data centers. By enabling the distribution of AI workloads across multiple locations, DriveNets' solution enhances the scalability and efficiency of AI systems. This could lead to more robust AI applications and support the growing demand for AI processing power in various sectors, including technology, finance, and healthcare. The ability to manage large data flows and prevent congestion is crucial for maintaining performance in AI training and operations.
What's Next?
WhiteFiber plans to expand the network's capacity to 136 Tbps by Q3 2026, indicating ongoing investment in this technology. As the demand for AI capabilities continues to grow, further developments in scale-across networking could become a standard in AI infrastructure, potentially influencing how data centers are designed and operated globally.
