NEC Corporation today announced the launch of its new NEC Composable Disaggregated Infrastructure Solution in Japan. This cutting-edge offering enables data centers, research institutions, and enterprise IT environments to flexibly deploy distributed computing resources—including servers, GPUs, and other high-performance components—across large network environments. By leveraging NEC’s proprietary ExpEther technology, the solution provides a reliable, low-latency method of transmitting IT equipment signals over Ethernet, helping organizations reduce both capital expenditures and operational costs while significantly improving energy efficiency.
As demand for advanced artificial intelligence continues to surge—particularly in generative AI, machine learning, and large-scale data analytics—organizations are struggling to scale their infrastructure effectively. Traditional data center architectures require servers to be built for peak performance, forcing operators to purchase and deploy large numbers of GPUs and CPUs even when they remain idle for most of the time. This over-provisioning results in wasted electricity, inefficient cooling usage, and the need for costly, frequent equipment refresh cycles. In parallel, the global rise in AI workloads has intensified pressure on organizations to build more sustainable and energy-efficient digital infrastructure. Recognizing these challenges, NEC has introduced a solution that offers both flexibility and efficiency through composable, disaggregated resource deployment.
At the core of this solution is the ability to physically separate and distribute resource components—such as CPUs, GPUs, and accelerators—from the server enclosure. Instead of being confined to a single rack or chassis, these components can be placed and managed across data-center-scale networks, maintaining high performance and low latency thanks to ExpEther’s advanced technology. This new architecture allows organizations to treat computing resources like a flexible pool that can be dynamically composed, expanded, or reconfigured depending on workload requirements.
NEC has conducted extensive verification testing to ensure the reliability and scalability of this approach. Joint studies with Osaka University, as well as internal trials at the NEC Inzai Data Center, demonstrated strong performance results in distributed GPU and server configurations. Findings from these proofs of concept are currently being published on NEC’s website, serving as a benchmark for organizations considering the transition from traditional, monolithic data center designs to more modern, composable architectures.
1. High Scalability and Flexible System Construction
The newly launched solution is built from several key components designed to deliver high scalability and flexible system deployment:
- ExpEther Boards with 100G ExpEther IP Core (1):
These boards enable 100Gbps Ethernet optical connectivity, forming the communication backbone of the distributed infrastructure. By utilizing the ExpEther protocol, signals can be transmitted reliably with extremely low latency, even when server components are placed far apart within a facility. - Expansion IO Box Supporting up to Eight GPUs:
This IO Box serves as a remotely deployable GPU enclosure, allowing the GPU resources typically found inside a server to be relocated to different floors, rooms, or even separate buildings. Organizations can optimize placement based on available power, cooling capacity, or physical layout. - NEC Composable Disaggregated Infrastructure Manager (2):
This management software orchestrates the allocation, composition, and monitoring of distributed resources. IT administrators can dynamically assign GPUs or other components to servers only when needed, enabling more responsive and efficient operations.
Together, these components free data centers from the physical limitations of conventional server installations. Operators can construct flexible, large-scale distributed systems where computing resources are physically decoupled but functionally integrated. This makes it possible to design data centers that adapt to varying cooling infrastructure, power density constraints, and evolving hardware roadmaps.
Additionally, this approach simplifies equipment upgrades. Instead of replacing entire servers to add new capabilities, organizations can upgrade individual components—such as swapping in newer GPUs—without discarding otherwise functional hardware. This modular approach reduces waste, lowers initial investment costs, and maintains future scalability as performance demands grow.
2. Efficient Utilization of Computing Resources and Cost Reduction
A major benefit of the NEC Composable Disaggregated Infrastructure Solution is its ability to dramatically improve the utilization of compute resources. By configuring CPUs, GPUs, and other accelerators only when they are required, operators avoid the idle periods common in traditionally architected systems. This ensures that resources are used more effectively, minimizing the over-provisioning that contributes to unnecessary power consumption and cost inflation.
During peak operating periods, resources can be aggregated and delivered to servers dynamically, ensuring that performance bottlenecks or shortages do not occur. Once workloads diminish, the system can automatically power down or deactivate unused GPUs and other components, reducing energy consumption. For industries where workloads fluctuate—such as AI research, data processing, and cloud-based development—this capability directly translates into substantial cost savings.
Beyond reducing electricity costs, the composable approach helps optimize the use of physical space. Data centers often struggle with balancing power density, cooling efficiency, and available rack space. With NEC’s solution, high-demand components can be strategically installed in locations with stronger cooling capacity or more available power, while servers operate in lower-density areas. This flexibility supports sustainable data center design and extends the usable life of existing facilities.
A New Stage in Optimized Computing Resource Management
The “NEC Composable Disaggregated Infrastructure Solution” marks a major advancement in the way computing environments are constructed and managed. By harnessing distributed IT architecture, NEC is providing organizations with a more agile and efficient method of operating data centers—one that aligns with the rapid growth of AI, big data analytics, and next-generation digital services.
NEC plans to initially release the solution for the Japanese market, with deployment beginning across industries that rely heavily on high-performance computing. As part of its commitment to long-term innovation, the company will continue to expand the technology and explore its applications in diverse sectors, including cloud services, academic research, and enterprise IT modernization.
Through this initiative, NEC aims to contribute to the development of a more sustainable and efficient digital infrastructure ecosystem—one capable of supporting the exponential growth of AI and data-driven businesses while addressing global concerns around energy usage and environmental impact.
Source Link:https://www.nec.com/
