Amazon Web Services has unveiled the Randomized Network Graph (RNG) data-center networking architecture, making it the default for most AWS workloads.

This shift represents a fundamental change in how the company manages its global infrastructure to handle the increasing demands of cloud computing. By optimizing the physical and logical layout of its networks, AWS aims to lower operational costs and energy requirements.

The RNG architecture significantly reduces the physical footprint of the data center. The new system uses 69% fewer networking devices [1], [2]. This reduction in hardware allows the company to streamline its infrastructure while maintaining high levels of connectivity.

Beyond hardware savings, the new topology improves performance. AWS said that network throughput is 33% higher under the RNG system [1], [2]. This increase in efficiency allows for faster data movement across the cloud environment, a critical factor for high-performance computing and large-scale AI workloads.

Sustainability and cost management are also central to the rollout. The transition has resulted in power consumption that is 40% lower [1], [2]. Additionally, some reports indicate that these efficiencies have led to cost reductions of up to 45% [3].

The move to RNG allows AWS to scale its capacity without a linear increase in power or hardware investment. By solving the technical challenges associated with network graphs, Amazon is altering the traditional design of data-center connectivity to ensure long-term resilience.

“AWS said that network throughput is 33% higher under the RNG system”

The adoption of Randomized Network Graphs signals a move away from traditional, rigid data-center topologies toward more flexible, mathematically optimized structures. For the broader industry, this demonstrates that significant gains in throughput and energy efficiency can be achieved through architectural redesign rather than simply adding more hardware. As AI continues to drive unprecedented demand for data-center bandwidth, such efficiencies are essential for maintaining the economic viability of hyperscale cloud providers.