August 27, 2026 at 12:49 PM ET
US Army selects 5 companies to deploy microreactors at military installations
The $2.2B in federal funding is intended to boost the sites’ resilience and the commercial viability of next-generation nuclear power.
Cloud to Edge: Benchmarking LLM Inference On Hardware-Accelerated Single-Board Computers
Power & Cooling Benchmarks & Research AI Compute Cloud & Colocation
Large language models (LLMs) are becoming increasingly capable at small parameter scales. At the same time, conventional cloud-centric deployment introduces challenges around data privacy, latency, and cost that are acute in operational technology and defence environments. Advances in model distillation, quantisation, and affordable edge accelerators now make local LLM inference on single-board computers feasible, but the high dimensionality of the configuration space makes identifying optimal deployments difficult without structured evaluation. Existing LLM-specific edge benchmarking efforts rely on CPU-only inference, poor coverage of genuine single-board computers, and generic evaluation tasks that lack multi-dimensional assessment of hardware effectiveness. This paper proposes a multi-dimensional benchmarking methodology that jointly evaluates inference performance and hardware efficiency across four IoT-suitable edge platform configurations testing single-board computers with the latest available hardware accelerators. Our results reveal the benefits of using hardware accelerators such as NPUs and GPUs, along with multi-dimensional evaluations quantifying the trade-offs between power efficiency, physical device size and token throughput; offering practical guidance for deploying generative AI in privacy-sensitive and connectivity-limited environments such as unmanned vehicles and portable, ruggedised operations.
More signals
August 27, 2026 at 9:02 AM ET
Data center load made up 9% of PJM wholesale costs so far in 2026: market monitor
Overall wholesale power costs in the PJM Interconnection jumped 46% to $56.7 billion so far this year, driven by rising energy and capacity costs, per the grid operator’s market monitor.
August 26, 2026 at 3:07 PM ET
Hull Street Acquires Two PJM-Based Power Stations from Rockland Capital
Maryland-based investment firm Hull Street Energy announced it has completed its acquisition of two Midwestern power plants that provide electricity in the PJM Interconnection.
August 26, 2026 at 12:28 PM ET
The POWER Interview: How Is Rising Scrutiny Changing the Way Data Centers Secure Power?
A recent flurry of state and federal action points to a marked pivot from treating data center growth as a grid-planning challenge to requiring large-load customers to show who pays for grid upgrades, what capacity supports the load, and how projects will operate during system stress. Morgan Lewis partner Arjun P. Ramadevanahalli spoke with POWER [...]
August 26, 2026 at 9:58 AM ET
Westinghouse eVinci Reaches Criticality Milestone at Nevada Test Site
Westinghouse’s eVinci microreactor has completed zero-power criticality testing at the National Criticality Experiments Research Center (NCERC), a National Nuclear Security Administration (NNSA) facility at the Nevada National Security Site (NNSS). The company said the milestone, achieved at 10:39 a.m. PT on Aug. 24 in partnership with Los Alamos National Laboratory (LANL) and Idaho National Laboratory [...]
August 23, 2026 at 6:14 PM ET
d-Matrix Raptor 3D-DRAM Accelerator for Generative Inference at Hot Chips 2026
At Hot Chips 2026, d-Matrix showed off its Raptor 3D-DRAM accelerator for AI breaking free of using HBM for memory by stacking DRAM and logic
