Which Emerging Data Center Technologies
Are Worth Evaluating Now?

Several emerging technologies are reshaping data center strategy. The most important things to evaluate now include:

Liquid Cooling: Direct-to-chip and immersion approaches enable higher rack densities while reducing fan energy, noise, and hotspots. They are especially valuable where racks are power constrained or thermal limits block expansion, making them a key enabler for AI and HPC workloads.

AI Accelerators: GPUs and purpose-built chips significantly increase throughput for both training and inference. To maximize ROI, plan for high-bandwidth interconnects, qualified firmware/software stacks, and shared accelerator pools that keep utilization high.

SmartNICs and Data Processing Units (DPUs): By offloading networking, storage, and security tasks from CPUs, these processors improve overall performance and free CPU cores for applications. They also support fine-grained isolation and micro-segmentation, strengthening data center security.

CXL and Composability: Compute Express Link (CXL) and composable infrastructure allow memory and device pooling, improving flexibility for memory-bound or bursty workloads and reducing stranded capacity.

NVMe-oF and High-speed Fabrics: High-speed storage fabrics can extend Non-Volatile Memory Express (NVMe) across the network to deliver fast, low-latency block storage. With quality-of-service features, organizations can ensure critical workloads always get the performance they need.

Silicon Photonics and Optical Interconnects: Still early—but promising for reducing latency and power per bit across dense clusters. These optical technologies could become critical as data centers scale AI training environments and exascale computing.

Organizations that pilot these technologies early can gain performance, efficiency, and scalability advantages that translate directly into competitive differentiation.