Research connecting computing, power and thermal management

The rapid growth of large language model (LLM) inference is creating significant data-center loads that face increasing energy-management challenges under tightening grid conditions and demand response (DR) requirements. Conventional data-center energy management mainly relies on temporal and spatial workload shifting and campus-level energy asset scheduling, but it usually treats LLM inference demand as an aggregate load. As a result, these approaches fail to exploit the internal characteristics of LLM serving and therefore overlook the flexibility offered by LLM-specific techniques such as model quantization. To unlock this flexibility, this paper proposes a quantization-enabled energy management framework for grid-responsive LLM inference data centers. First, a quantization-to-power model is established to map each model--quantization configuration to a compact set of dispatchable parameters. Second, a two-stage quantization-enabled DR model is developed to account for model instance switching, request routing, and precision selection. Third, a multi-campus co-optimization method is introduced for DR participation by integrating grid-side electricity and carbon signals with the quantization-enabled DR model. Case studies show that the proposed framework reduces total data-center operating cost by 34.3\% without curtailing served token volume, validating model quantization as an effective flexibility lever for grid-responsive LLM data-center energy management.

Liquid cooled supercomputers dissipate tens of megawatts of waste heat through cooling plants organized as parallel subloops that serve coolant distribution units. The number of subloops and the assignment of units to them are design decisions fixed at construction, yet they have not been systematically optimized for facilities at this scale. As electricity grids decarbonize, embodied carbon becomes a larger share of facility life cycle emissions and the cost of an unnecessary subloop becomes harder to justify. We present a framework that integrates operational energy from a validated control optimizer based on sequential least squares programming, embodied carbon from a bill of materials, and expected unplanned downtime from a per subloop reliability model. The framework is applied to the Frontier supercomputer, evaluating all 611 ways of partitioning its 25 coolant distribution units into two through six subloops. The life cycle cost and carbon optimum is found at two subloops holding 14 and 11 units, achieving 3,320.7 tonnes of carbon dioxide equivalent and $3.99 million over a seven year horizon, a saving of 50.2 tonnes and $100,000 compared to built four subloop configuration. The optimum remains on the Pareto front in all 15 scenarios of a one at a time sensitivity sweep. A semi-analytical decision rule generalizes the result, predicting four subloops for Aurora, two for El Capitan, and one for LUMI. When reliability is treated as a hard constraint set by operations policy, the four subloop Frontier deployment is consistent with the constrained optimum.

With the increasing demand for edge computing and AI-driven workloads, integrating small and medium-sized edge data centers into distribution networks has become increasingly important. This paper investigates the hosting capacity of distribution networks for data center integration and identifies the key physical mechanisms that limit the maximum allowable data center load. The baseline analysis shows that data center hosting capacity varies significantly across candidate buses due to network topology and electrical distance. Three dominant limiting mechanisms are identified: current-constrained locations, voltage-constrained locations, and mixed-constrained locations where both current loading and voltage deviation jointly affect hosting capacity. To increase the hosting capacity, this study evaluates multiple flexible resources, including battery energy storage systems (BESS), dispatchable distributed generators (DDG), and static synchronous compensators (STATCOM). Numerical results demonstrate that these resources provide complementary benefits through active power support, sustained local generation, and reactive power compensation, effectively expanding data center hosting capacity in distribution systems.

The electric power supply for AI data centers is now the most significant bottleneck in the race toward Artificial General Intelligence, surpassing even the constraint of AI accelerator availability. To our knowledge, this paper is the first to describe the end-to-end power management process for a hyper-scale AI datacenter; from early power planning to accommodate next-generation accelerators 6--12 months before their general availability, to tuning power settings after large scale deployment, and finally to dynamic, runtime power management for evolving workloads. We present detailed power measurements for a 150 MW datacenter hosting a cluster of 83K GB200 GPUs. We share insights from building this state-of-the-art AI cluster. We hope this work encourages practitioners across the industry to share their own experiences as well.

The rapid growth of AI workloads is substantially increasing data center electricity demand and carbon emissions, motivating the development of carbon-aware scheduling methods. However, effective scheduling is challenging because renewable generation and AI workloads are subject to forecast errors, while training and inference workloads exhibit heterogeneity in computational characteristics. This paper proposes a contextual robust optimization framework for AI data center operation. The proposed model explicitly captures the heterogeneous computational characteristics of AI training and inference workloads. To deal with renewable generation and workload forecast errors, we develop loss-based uncertainty learning models that directly map contextual features to covariate-dependent uncertainty sets. The resulting contextual joint chance-constrained scheduling problem is reformulated into a tractable robust optimization problem, and a calibration algorithm is developed to provide finite-sample probabilistic feasibility guarantees for multiple joint chance constraints. Numerical experiments based on real-world AI workload traces and renewable generation data show that the proposed method reduces operating costs by an average of 5.57% compared to benchmark methods while maintaining reliable feasibility and strong computational scalability.

High-Performance Computing (HPC) has recently entered the Exascale era, and considerable efforts are being made to fully harness this potential power for large-scale applications, such as cutting-edge generative AI (training and exploitation). The corresponding energy consumption is very high, and forecasts are alarming, making this metric a critical systemic bottleneck. Addressing this issue presents a genuine challenge for the entire cloud-edge-HPC continuum at all scales, from low-power IoT microcontrollers to multi-megawatt data centers. Beyond financial costs, green computing is driven by considerations related to climate change and environmental concerns such as carbon footprint ($CO_2e$), as well as constraints on energy production and supply, leading to a real need to regulate {\em information and communication technology} (ICT) activities. This article presents a comprehensive overview of energy-efficient computing, taking into account the most recent and significant contributions. Based on this exploration of the state of the art, we design and describe a holistic taxonomy of the aforementioned publications, structured around various perspectives, including {\em hardware and software aspects, measurement instrumentation, software optimizations, dynamic task scheduling, voltage scaling, workload consolidation, federated learning}, and {\em cooling}. Particular emphasis is placed on large-scale AI, which receives significant attention due to its considerable resource requirements. We conclude with an analysis of a forward-looking roadmap that considers the main perspectives of sustainable computing.

The proliferation of large-scale data centers introduces spatially correlated demand profiles that challenge the long-standing assumption of statistical independence of loads in power system analysis. This paper examines the emergence of such load correlations and evaluates their impact on data-center-dominated grids. Analytical derivations reveal that correlated load fluctuations amplify aggregate stochastic disturbances, reduce voltage stability margins through weakened reactive power stiffness, and degrade frequency stability margin by erosion of natural load diversity effects. Real-time digital simulation studies confirm that moderate spatial correlation in distributed data centers produces simultaneous frequency deviations and voltage fluctuations across multiple buses. The findings offer transmission system operators a physics-based perspective to interpret emerging oscillatory phenomena and establish stability planning criteria grounded in measurable load-correlation structures rather than traditional diversity assumptions.

Hyperscale AI data centers induce spatially and temporally correlated load fluctuations that violate classical independence assumptions and are not captured by time-averaged spectral methods. These correlations are episodic and non-stationary, requiring analysis that resolves transient structure. This paper applies Dynamic Mode Decomposition (DMD) to the temporal evolution of pairwise inter-bus correlation coefficients to form a low-dimensional state representation that enables modal analysis without a stationarity assumption. DMD eigenvalues encode the correlation regime: their location in the complex plane distinguishes sustained coherence, decaying transients, and intensifying events, while oscillation frequency maps to underlying physical coupling mechanisms. Using an IEEE 39-bus Real-Time Digital Simulator (RTDS) testbed with three converter-interfaced AI data center loads driven by synthetic workload profiles, global DMD provides a time-averaged modal baseline in a slow thermal band ($f \approx 0.005$\,Hz, $|μ| = 0.91$) captures 93.6\% of total correlation energy. A sliding-window DMD formulation identifies transient intensification events: 51 of 775 windows (6.6\%) satisfy the $|μ_k^{(n)}| > 1$ criterion, which aligns with stochastic workload coincidences. Cross-validation with RTDS voltage coherence confirms elevated coupling during these intervals. The proposed modal growth indicator provides an early-warning signal of correlation intensification prior to peak pairwise coherence.