pith. sign in

arxiv: 2603.00415 · v2 · submitted 2026-02-28 · 📡 eess.SY · cs.SY

Grid Integration of AI Data Centers: A Critical Review of Energy Storage Solutions

Sina Mohammadi , Wayne Wang , Marcus Chen I Wada , Rouzbeh Haghighi , Ali Hassan , Hualong Liu , Archit Bhatnagar , Ang Chen
show 1 more author
Wencong Su
This is my paper

Pith reviewed 2026-05-15 18:54 UTC · model grok-4.3

classification 📡 eess.SY cs.SY
keywords AI data centersenergy storage systemsgrid integrationhierarchical controlload smoothingbattery energy storagepower system stabilityuninterruptible power supply
0
0 comments X

The pith

AI data centers require coordinated energy storage across chip, rack, facility, and grid levels to manage their unique rapid power fluctuations.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper reviews energy storage technologies to integrate growing AI data centers with the electric grid. It identifies that AI workloads cause sub-second power variations unlike steady traditional loads, rendering standard dispatch methods inadequate. The authors organize the review around a four-layer hierarchy of storage solutions from the chip level up to grid-scale systems, emphasizing the need for coordination between layers. Without this approach, grid reliability and stability face significant risks from the dynamic demands of AI computing infrastructure. The review also points out remaining gaps in modeling, forecasting, and sizing tools.

Core claim

The central discovery is that the highly dynamic, sub-second variable power profiles of AI data centers make conventional energy storage dispatch strategies insufficient, necessitating a hierarchical, coordinated deployment of energy storage systems across chip-level buffering, rack and server-level ESSs, facility-level UPS systems, and grid-scale BESSs, along with supplementary non-battery technologies, to achieve effective load smoothing and grid support.

What carries the argument

A four-layer hierarchical taxonomy for ESS deployment that evaluates each layer by its response timescale, power and energy ratings, operational role, integration challenges, and coordination requirements with other layers.

If this is right

  • Effective load smoothing and grid support require coordination across all layers of the hierarchy.
  • Significant research gaps persist in simulation tools, degradation modeling, load forecasting, and optimal multi-layer sizing.
  • Non-battery technologies such as fuel cells and thermal energy storage can supplement battery-based solutions.
  • AI DC load profiles differ fundamentally from traditional loads due to their sub-second variability.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Implementing this hierarchy could enable better integration of AI data centers with renewable energy sources by smoothing intermittent demands.
  • Future work might focus on developing unified control algorithms that span multiple layers to optimize overall system performance.
  • Real-world testing of coordinated ESS on actual AI workloads would help quantify the benefits over conventional approaches.

Load-bearing premise

That the reviewed literature covers the sub-second variability of real AI workloads sufficiently and that conventional ESS strategies are shown to be inadequate without multi-layer coordination.

What would settle it

An empirical study showing that a single-layer energy storage system can effectively smooth sub-second AI data center load variations without requiring hierarchical coordination, or comprehensive modeling that demonstrates current tools already address the variability adequately.

Figures

Figures reproduced from arXiv: 2603.00415 by Ali Hassan, Ang Chen, Archit Bhatnagar, Hualong Liu, Marcus Chen I Wada, Rouzbeh Haghighi, Sina Mohammadi, Wayne Wang, Wencong Su.

Figure 1
Figure 1. Figure 1: Overall architecture of an AI data center with hierarchical energy storage systems. The figure illustrates the power delivery path from the utility grid [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: UPS and GiUPS operation modes GiUPS systems enhance TUPS designs by incorporating bidirectional power converters, advanced digital controllers, and communication interfaces that allow coordination with the utility grid or market operator. These systems can modulate active power in response to grid frequency deviations or dispatch signals, enabling participation in services such as primary frequency regulat… view at source ↗
Figure 4
Figure 4. Figure 4: UPS-Integrated BESS operation modes As shown in [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 3
Figure 3. Figure 3: GFM BESS as a power smoothing unit [57] [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 5
Figure 5. Figure 5: FC integration with AI DC. hydrogen, FCs produce no carbon dioxide or conventional air pollutants during operation [68]. Among the several established FC chemistries, Solid Oxide Fuel Cells (SOFCs), in context of AI DCs, are often high￾lighted for their high steady-state efficiency and suitability for long-duration on-site power generation [69]. Polymer Elec￾trolyte Membrane Fuel Cells (PEMFCs) exhibit fas… view at source ↗
Figure 6
Figure 6. Figure 6: Rack-level BBUs and server-level capacitors [PITH_FULL_IMAGE:figures/full_fig_p012_6.png] view at source ↗
Figure 8
Figure 8. Figure 8: FFR by utilizing new BESS and SLBESS in GiUPS architecture [PITH_FULL_IMAGE:figures/full_fig_p014_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: ) to achieve high reliability, redundancy, and operational flexibility. Within this framework, the placement and coordina￾ [PITH_FULL_IMAGE:figures/full_fig_p018_9.png] view at source ↗
read the original abstract

Artificial intelligence (AI) is driving unprecedented growth in data center (DC) scale and power demand. AI workloads impose highly dynamic, difficult-to-forecast power profiles on the utility grid, creating reliability and stability challenges that conventional DC architectures are not designed to address. This paper provides a critical review of energy storage systems (ESSs) as the key enabling technology for reliable grid integration of AI DCs. We organize the review around a four-layer hierarchical taxonomy, namely chip-level buffering, rack/server-level ESSs, facility-level uninterruptible power supply (UPS) systems, and grid-scale battery energy storage systems (BESSs), supplemented by non-battery technologies including fuel cells (FCs) and thermal energy storage (TES). Each layer is analyzed with respect to response timescale, power and energy ratings, operational role, integration challenges, and coordination requirements. Key findings include: (i) AI DC load profiles differ fundamentally from traditional loads in their sub-second variability, making conventional ESS dispatch strategies insufficient; (ii) hierarchical, coordinated ESS deployment across all layers is necessary for effective load smoothing and grid support; and (iii) significant gaps remain in simulation tools, degradation modeling, load forecasting, and optimal multi-layer sizing. This review identifies open research challenges and future directions at the intersection of AI computing infrastructure and power system integration.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper is a critical review of energy storage systems (ESS) for grid integration of AI data centers. It proposes a four-layer hierarchical taxonomy (chip-level buffering, rack/server-level ESS, facility-level UPS, grid-scale BESS) supplemented by fuel cells and thermal storage. Each layer is assessed on response timescale, ratings, role, challenges, and coordination needs. Central claims are that AI workloads exhibit sub-second variability fundamentally unlike traditional loads (rendering conventional single-layer ESS dispatch insufficient), that hierarchical coordinated deployment across layers is therefore necessary for load smoothing and grid support, and that major gaps persist in simulation tools, degradation modeling, forecasting, and multi-layer sizing.

Significance. If the literature synthesis is accurate and balanced, the work is significant for framing a timely interdisciplinary problem at the intersection of AI infrastructure growth and power-system stability. The taxonomy provides a useful organizing lens for comparing ESS technologies by scale and speed, and the gap identification could usefully direct future modeling and control research. The absence of new derivations or simulations is appropriate for a review, but the strength hinges on the breadth and critical depth of the cited prior work.

major comments (3)
  1. [Abstract and §1] Abstract and §1 (Introduction): The claim that 'AI DC load profiles differ fundamentally from traditional loads in their sub-second variability, making conventional ESS dispatch strategies insufficient' is load-bearing for the necessity of the hierarchical approach, yet the manuscript provides no explicit quantitative comparison (e.g., measured or simulated ramp rates, frequency content, or forecast-error statistics) drawn from the reviewed literature to demonstrate where single-layer strategies fail.
  2. [Taxonomy and layer-analysis sections] Taxonomy and layer-analysis sections (presumably §3–§6): The assertion that 'hierarchical, coordinated ESS deployment across all layers is necessary' rests on synthesis, but the paper does not tabulate or meta-analyze response-time coverage or coordination requirements across the cited studies; without such a summary it is unclear whether the literature actually shows that uncoordinated single-layer solutions are demonstrably inadequate for real AI workloads.
  3. [Gap-identification section] Gap-identification section (presumably near end): The listed gaps in simulation tools, degradation modeling, load forecasting, and optimal multi-layer sizing are presented as findings, but the manuscript does not reference specific prior attempts that fell short or define concrete evaluation criteria (e.g., required model fidelity for sub-second dynamics), leaving the gaps high-level rather than actionable.
minor comments (2)
  1. [Throughout] Ensure every acronym (ESS, BESS, UPS, FC, TES) is defined at first use in the main text and used consistently thereafter.
  2. [Taxonomy figure/table] If a figure or table presents the four-layer taxonomy, add explicit arrows or annotations showing the coordination signals or data flows between layers to make the 'hierarchical' claim visually concrete.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the positive assessment of the paper's significance and for the constructive major comments. These have highlighted opportunities to make the synthesis more explicit and the gap analysis more actionable. We address each point below and will incorporate revisions in the next version of the manuscript.

read point-by-point responses

  1. Referee: [Abstract and §1] The claim that 'AI DC load profiles differ fundamentally from traditional loads in their sub-second variability, making conventional ESS dispatch strategies insufficient' is load-bearing for the necessity of the hierarchical approach, yet the manuscript provides no explicit quantitative comparison (e.g., measured or simulated ramp rates, frequency content, or forecast-error statistics) drawn from the reviewed literature to demonstrate where single-layer strategies fail.

    Authors: We agree that an explicit quantitative comparison would strengthen the central claim. Although the review synthesizes multiple studies reporting sub-second GPU and workload fluctuations, we did not consolidate the specific metrics. We will add a new table (or subsection) in §1 that extracts and contrasts ramp rates, power spectral density features, and forecast-error statistics from the key cited works on AI DC loads versus traditional IT loads. This will directly support the argument that single-layer dispatch is insufficient. revision: yes

  2. Referee: [Taxonomy and layer-analysis sections] The assertion that 'hierarchical, coordinated ESS deployment across all layers is necessary' rests on synthesis, but the paper does not tabulate or meta-analyze response-time coverage or coordination requirements across the cited studies; without such a summary it is unclear whether the literature actually shows that uncoordinated single-layer solutions are demonstrably inadequate for real AI workloads.

    Authors: We concur that a consolidated meta-summary would make the coverage argument more transparent. We will insert a summary table in the taxonomy section (§3) that compiles, across all reviewed studies, the reported response timescales, power/energy ratings, and any coordination mechanisms discussed. The table will explicitly flag coverage gaps for sub-second dynamics when only single-layer solutions are considered, thereby demonstrating the necessity of the hierarchical approach based on the existing literature. revision: yes

  3. Referee: [Gap-identification section] The listed gaps in simulation tools, degradation modeling, load forecasting, and optimal multi-layer sizing are presented as findings, but the manuscript does not reference specific prior attempts that fell short or define concrete evaluation criteria (e.g., required model fidelity for sub-second dynamics), leaving the gaps high-level rather than actionable.

    Authors: We acknowledge that the gap section would benefit from greater specificity. We will revise it to cite concrete examples from the reviewed literature (e.g., prior single-layer BESS or UPS studies that reported inadequate performance under rapid AI load transients) and to define concrete evaluation criteria, such as required simulation timestep resolution (<10 ms for sub-second dynamics), target accuracy metrics for degradation models, and quantitative benchmarks for multi-layer sizing optimization. This will render the identified gaps more actionable for future research. revision: yes

Circularity Check

0 steps flagged

No significant circularity identified

full rationale

This is a review paper synthesizing external literature on ESS for AI data centers. The central claim of needing hierarchical coordinated ESS deployment is presented as a synthesis of reviewed studies on load variability and grid challenges, with no new equations, fitted parameters, predictions, or derivations. The four-layer taxonomy is a descriptive organizational framework drawn from existing work, not a self-referential renaming or ansatz. No self-citation load-bearing steps, uniqueness theorems, or self-definitional reductions are present; all findings rest on cited external sources rather than reducing to the paper's own inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

This is a literature review; no free parameters, new axioms, or invented entities are introduced by the authors.

pith-pipeline@v0.9.0 · 5568 in / 1020 out tokens · 18501 ms · 2026-05-15T18:54:05.621110+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

148 extracted references · 148 canonical work pages · 1 internal anchor

  1. [1]

    Data center growth and grid readiness (tr131),

    B. Chalamalaet al., “Data center growth and grid readiness (tr131),” IEEE Power and Energy Society, Technical Report TR131, 2025

    work page 2025

  2. [2]

    Characteristics and risks of emerging large loads,

    North American Electric Reliability Corporation (NERC), “Characteristics and risks of emerging large loads,” NERC, White Paper, July 2025, large Loads Task Force (LLTF). [Online]. Available: https://www.nerc.com/globalassets/who-we-are/standing-committees/ rstc/whitepaper-characteristics-and-risks-of-emerging-large-loads.pdf

    work page 2025

  3. [3]

    What is an ai data center?

    A. Jonker and A. Gomstyn, “What is an ai data center?” 2025. [Online]. Available: https://www.ibm.com/think/topics/ai-data-center

    work page 2025

  4. [4]

    Data center development in an ai-driven market,

    Stream Data Centers, “Data center development in an ai-driven market,” Stream Data Centers, Tech. Rep., Feb. 2024. [Online]. Available: https://www.streamdatacenters.com/wp-content/uploads/2024/02/SDC- BTPS-Whitepaper-240222.pdf

    work page 2024

  5. [5]

    Dynamic load model for data centers with pattern-consistent calibration,

    S. Lu, C. Xiao, and Y . Weng, “Dynamic load model for data centers with pattern-consistent calibration,”arXiv preprint arXiv:2602.07859, 2026

    work page arXiv 2026

  6. [6]

    Measurement of Generative AI Workload Power Profiles for Whole-Facility Data Center Infrastructure Planning

    R. Vercellino, J. Willard, G. Campos, W. d. S. Pereira, O. Hull, M. Selensky, and J. Mueller, “Measurement of generative ai workload power profiles for whole-facility data center infrastructure planning,” arXiv preprint arXiv:2604.07345, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  7. [7]

    A soft- switching multiresonant switched-capacitor converter for data center applications,

    X. Li, Y . Guan, T. Yao, Y . Wang, W. Wang, and D. Xu, “A soft- switching multiresonant switched-capacitor converter for data center applications,”IEEE Transactions on Power Electronics, vol. 41, no. 7, pp. 11 079–11 097, 2026

    work page 2026

  8. [8]

    Sequential operating simulation of solid state transformer-driven next-generation 800 vdc data center,

    J. Xu, X. Jiang, Y . Bao, Y . Zheng, X. Chen, Q. Xu, S. Liao, D. Ke, and X. Gao, “Sequential operating simulation of solid state transformer-driven next-generation 800 vdc data center,”arXiv preprint arXiv:2601.16502, 2026

    work page arXiv 2026

  9. [9]

    arXiv preprint arXiv:2509.07218 , year=

    X. Chen, X. Wang, A. Colacelli, M. Lee, and L. Xie, “Electricity demand and grid impacts of ai data centers: Challenges and prospects,” arXiv preprint arXiv:2509.07218, 2025

    work page arXiv 2025

  10. [10]

    Big tech’s data center boom poses new risk to us grid operators,

    T. McLaughlin, “Big tech’s data center boom poses new risk to us grid operators,” 2025. [Online]. Available: https://www.reuters. com/technology/big-techs-data-center-boom-poses-new-risk-us-grid- operators-2025-03-19/

    work page 2025

  11. [11]

    Technical challenges of ai data center integration into power grids—a survey,

    E. Ginzburg-Ganz, P. Lifshits, R. Machlev, J. Belikov, Z. Krieger, and Y . Levron, “Technical challenges of ai data center integration into power grids—a survey,”Energies, vol. 19, no. 1, p. 137, 2025

    work page 2025

  12. [12]

    Energy storage systems for ai data centers: A review of technologies, characteristics, and applicability,

    S. Rahman and T. A. Khan, “Energy storage systems for ai data centers: A review of technologies, characteristics, and applicability,”Energies, vol. 19, no. 3, p. 634, 2026

    work page 2026

  13. [13]

    A research-industry perspective of battery systems technology for next-generation data centers,

    A. Safari, F. Blaabjerg, and A. Oshnoei, “A research-industry perspective of battery systems technology for next-generation data centers,”Journal of Energy Storage, vol. 152, p. 120386, 2026. [Online]. Available: https://www.sciencedirect.com/science/article/pii/ S2352152X26000502

    work page 2026

  14. [14]

    Alibaba hpn: A data center network for large language model training,

    K. Qian, Y . Xi, J. Cao, J. Gaoet al., “Alibaba hpn: A data center network for large language model training,” inProceedings of the ACM SIGCOMM 2024 Conference. New York, NY , USA: Association for Computing Machinery, 2024, pp. 691–706. [Online]. Available: https://dl.acm.org/doi/10.1145/3651890.3672265

    work page doi:10.1145/3651890.3672265 2024

  15. [15]

    N. V . Savant. (2025) Gpu-to-gpu communication: Unlocking parallelism beyond the core. Medium. [Online]. Available: https: //medium.com/@nikheelvs/gpu- to- gpu- communication- unlocking- parallelism-beyond-the-core-a80de2974078

    work page 2025

  16. [16]

    Charac- teristics and risks of emerging large loads,

    North American Electric Reliability Corporation (NERC), “Charac- teristics and risks of emerging large loads,” North American Electric Reliability Corporation, Tech. Rep., Jul. 2025, large Loads Task Force White Paper

    work page 2025

  17. [17]

    (2025) An overview of popular nvidia gpus

    Atlantic.Net. (2025) An overview of popular nvidia gpus. Atlantic.Net. Accessed: 2026-02-27. [Online]. Available: https://www.atlantic.net/ gpu-server-hosting/an-overview-of-popular-nvidia-gpus/

    work page 2025

  18. [18]

    N. S. Rouslan Dimitrov, Harry Petty and M. Blake. (2025) How new gb300 nvl72 features provide steady power for ai. NVIDIA Developer Blog. [Online]. Available: https://developer.nvidia.com/blog/how-new- gb300-nvl72-features-provide-steady-power-for-ai/

    work page 2025

  19. [19]

    (2024) Accelerating ai inference with google cloud tpus and gpus

    Google Cloud. (2024) Accelerating ai inference with google cloud tpus and gpus. Google Cloud Blog. Accessed: 2026-02-27. [Online]. Available: https://cloud.google.com/blog/products/compute/ accelerating-ai-inference-with-google-cloud-tpus-and-gpus

    work page 2024

  20. [20]

    (2024) What changes in storage will ai drive? Micron Blog

    Micron Technology. (2024) What changes in storage will ai drive? Micron Blog. [Online]. Available: https://www.micron.com/about/ blog/storage/ai/what-changes-in-storage-will-ai-drive

    work page 2024

  21. [21]

    (2019) Gpudirect storage: A direct path between storage and gpu memory

    NVIDIA Corporation. (2019) Gpudirect storage: A direct path between storage and gpu memory. NVIDIA Developer Blog. [Online]. Available: https://developer.nvidia.com/blog/gpudirect-storage/

    work page 2019

  22. [22]

    Rao, Bruno Ribeiro, and Mohit Tawar- malani

    A. Gangidi, R. Miao, S. Zheng, S. J. Bondu, G. Goes, H. Morsy, R. Puri, M. Riftadi, A. J. Shetty, J. Yang, S. Zhang, M. J. Fernandez, S. Gandham, and H. Zeng, “Rdma over ethernet for distributed training at meta scale,” inProceedings of the ACM SIGCOMM 2024 Conference, ser. ACM SIGCOMM ’24. New York, NY , USA: Association for Computing Machinery, 2024, p....

    work page doi:10.1145/3651890.3672233 2024

  23. [23]

    (2023) Networking for data centers and the era of ai

    NVIDIA Corporation. (2023) Networking for data centers and the era of ai. NVIDIA Developer Blog. [Online]. Available: https://developer. nvidia.com/blog/networking-for-data-centers-and-the-era-of-ai/

    work page 2023

  24. [24]

    Evolving a data center into a microgrid: Industry perspectives and lessons learned,

    S. Sheehan and A. Rakow, “Evolving a data center into a microgrid: Industry perspectives and lessons learned,”IEEE Electrification Mag- azine, vol. 11, no. 3, pp. 16–25, 2023

    work page 2023

  25. [25]

    Department of Energy

    U.S. Department of Energy. (2025) Advantages and challenges of nuclear-powered data centers. Office of Nuclear Energy. Accessed: 2026-02-27. [Online]. Available: https://www.energy.gov/ne/articles/ advantages-and-challenges-nuclear-powered-data-centers

    work page 2025

  26. [26]

    Hydrogen fuel and fuel cells: Potential candi- dates for sustainable, dispatchable, and environmentally friendly power generation and transport technologies,

    H. Nehrir and C. Wang, “Hydrogen fuel and fuel cells: Potential candi- dates for sustainable, dispatchable, and environmentally friendly power generation and transport technologies,”IEEE Energy Sustainability Magazine, vol. 1, no. 3, pp. 52–65, 2025

    work page 2025

  27. [27]

    (2024) Understanding direct-to-chip cooling in hpc infrastructure: A deep dive into liquid cooling

    Vertiv. (2024) Understanding direct-to-chip cooling in hpc infrastructure: A deep dive into liquid cooling. Vertiv. [Online]. Available: https://www.vertiv.com/en- us/about/news- and-insights/articles/educational-articles/understanding-direct-to-chip- cooling-in-hpc-infrastructure-a-deep-dive-into-liquid-cooling/

    work page 2024

  28. [28]

    Data centres as a source of flexibility for power systems,

    M. T. Takci, M. Qadrdan, J. Summers, and J. Gustafsson, “Data centres as a source of flexibility for power systems,”Energy Reports, vol. 13, pp. 3661–3671, 2025. 20

    work page 2025

  29. [29]

    Uninterruptible power supplies (ups) for data center,

    A. Karpati, G. Zsigmond, M. V ¨or¨os, and M. Lendvay, “Uninterruptible power supplies (ups) for data center,” in2012 IEEE 10th Jubilee In- ternational Symposium on Intelligent Systems and Informatics. IEEE, 2012, pp. 351–355

    work page 2012

  30. [30]

    Minimizing data center uninterruptable power supply overload by server power capping,

    A.-H. Fawaz, J. Lorincz, and A. F. Mohammed, “Minimizing data center uninterruptable power supply overload by server power capping,” IEEE Communications Letters, vol. 23, no. 8, pp. 1342–1346, 2019

    work page 2019

  31. [31]

    Z. Wang, Z. Yin, J. Yang, and J. Wang, “Coordinated optimization of distributed energy system and storage-enhanced uninterruptible power supply in data center: A three-level optimization framework with model predictive control,”Energy Conversion and Management, vol. 342, p. 120137, 2025

    work page 2025

  32. [32]

    Grid-interactive data centers enabling energy transition: Data center’s hidden potential to provide essential grid services of a future power system,

    J. Paananen, “Grid-interactive data centers enabling energy transition: Data center’s hidden potential to provide essential grid services of a future power system,”IEEE Electrification Magazine, vol. 11, no. 3, pp. 26–34, 2023

    work page 2023

  33. [33]

    Ai data centres as grid-interactive assets,

    P. Colangelo, A. K. Coskun, J. Megrue, C. Roberts, S. Sengupta, V . Sivaram, E. Tiao, A. Vijaykar, C. Williams, D. C. Wilsonet al., “Ai data centres as grid-interactive assets,”Nature Energy, pp. 1–8, 2025

    work page 2025

  34. [34]

    How to maximize revenues from your data center energy storage system with grid interactive ups,

    A. Di Filippi and L. Valentini, “How to maximize revenues from your data center energy storage system with grid interactive ups,” Vertiv, Tech. Rep., 2025, vertiv White Paper

    work page 2025

  35. [35]

    Grid-interactive data centers: enabling decarbonization and system stability,

    J. Paananen and E. Nasr, “Grid-interactive data centers: enabling decarbonization and system stability,”Dublin, Ireland, 2021

    work page 2021

  36. [36]

    Hyperscale meets grid stability: On.energy launches medium-voltage ups built for ai data centers,

    ON.energy, “Hyperscale meets grid stability: On.energy launches medium-voltage ups built for ai data centers,” 2025. [Online]. Available: https://www.nacleanenergy.com/alternative- energies/hyperscale-meets-grid-stability-on-energy-launches-medium- voltage-ups-built-for-ai-data-centers

    work page 2025

  37. [37]

    How to maximize revenues from your data center energy storage system with grid interactive ups,

    A. Di Filippi and L. Valentini, “How to maximize revenues from your data center energy storage system with grid interactive ups,”

    work page

  38. [38]

    Available: https://www.vertiv.com/4918e5/globalassets/ documents/white-papers/white-paper-maximize-revenues-data-center- energy-storage-grid-ups 329440 2.pdf

    [Online]. Available: https://www.vertiv.com/4918e5/globalassets/ documents/white-papers/white-paper-maximize-revenues-data-center- energy-storage-grid-ups 329440 2.pdf

    work page

  39. [39]

    Eaton and microsoft’s energyaware ups technology pilot project

    Eaton, “Eaton and microsoft’s energyaware ups technology pilot project.” [Online]. Available: https://www.eaton.com/us/en- us/products/backup- power- ups- surge- it- power- distribution/backup- power-ups/dual-purpose-ups-technology.html

    work page

  40. [40]

    Enhancing data center low-voltage ride-through,

    Y . Xie, W. Cui, and A. Wierman, “Enhancing data center low-voltage ride-through,”arXiv preprint arXiv:2510.03867, 2025

    work page arXiv 2025

  41. [41]

    Strengthening data center operations using grid-forming battery energy storage as a line-interactive uninterruptible power sup- ply,

    A. Azizi, S. Morovati, A. Zamani, J. Piruzza, D. Guo, Z. Liu, and P. Taimela, “Strengthening data center operations using grid-forming battery energy storage as a line-interactive uninterruptible power sup- ply,”International Journal of Electrical Power & Energy Systems, vol. 175, p. 111638, 2026

    work page 2026

  42. [42]

    V oltage ride-through: A key ingredient in data center resilience,

    C. Tozzi, “V oltage ride-through: A key ingredient in data center resilience,” 2026. [Online]. Available: https: //www.datacenterknowledge.com/uptime/voltage-ride-through-a-key- ingredient-in-data-center-resilience

    work page 2026

  43. [43]

    Texas loads ride toward grid stability: V oltage ride through of large power electronic loads,

    J. Conto, Y . Cheng, J. Rose, and J. Schmall, “Texas loads ride toward grid stability: V oltage ride through of large power electronic loads,” IEEE Power and Energy Magazine, vol. 23, no. 5, pp. 56–67, 2025

    work page 2025

  44. [44]

    Best practices for large load interconnections: A north american perspective on data centers,

    R. Zahedi, A. Zamani, and R. Anilkumar, “Best practices for large load interconnections: A north american perspective on data centers,”arXiv preprint arXiv:2601.12686, 2026

    work page arXiv 2026

  45. [45]

    Fault ride-through of renewable energy conversion systems during voltage recovery,

    R. Li, H. Geng, and G. Yang, “Fault ride-through of renewable energy conversion systems during voltage recovery,”Journal of Modern Power Systems and Clean Energy, vol. 4, no. 1, pp. 28–39, 2016

    work page 2016

  46. [46]

    V oltage- profile-based approach for developing collection system aggregated models for wind generation resources for grid voltage ride-through studies,

    Y . Cheng, M. Sahni, J. Conto, S.-H. Huang, and J. Schmall, “V oltage- profile-based approach for developing collection system aggregated models for wind generation resources for grid voltage ride-through studies,”IET Renewable Power Generation, vol. 5, no. 5, pp. 332– 346, 2011

    work page 2011

  47. [47]

    An introduction to the life cycle assessment (lca) of bioelectrochemical systems (bes) for sustainable energy and product generation: relevance and key aspects,

    D. Pant, A. Singh, G. Van Bogaert, Y . A. Gallego, L. Diels, and K. Vanbroekhoven, “An introduction to the life cycle assessment (lca) of bioelectrochemical systems (bes) for sustainable energy and product generation: relevance and key aspects,”Renewable and Sustainable Energy Reviews, vol. 15, no. 2, pp. 1305–1313, 2011

    work page 2011

  48. [48]

    Optimized battery-management system to improve storage lifetime in renewable energy systems,

    R. Kaiser, “Optimized battery-management system to improve storage lifetime in renewable energy systems,”Journal of Power Sources, vol. 168, no. 1, pp. 58–65, 2007

    work page 2007

  49. [49]

    Optimal whole-life-cycle planning of battery energy storage for multi-functional services in power systems,

    Y . Zhang, Y . Xu, H. Yang, Z. Y . Dong, and R. Zhang, “Optimal whole-life-cycle planning of battery energy storage for multi-functional services in power systems,”IEEE Transactions on Sustainable Energy, vol. 11, no. 4, pp. 2077–2086, 2019

    work page 2077

  50. [50]

    Life-cycle performance analysis of a building integrated energy system considering equipment perfor- mance degradation,

    J. Wang, S. Ye, B. Wu, and B. Liu, “Life-cycle performance analysis of a building integrated energy system considering equipment perfor- mance degradation,”Energy Conversion and Management, vol. 347, p. 120593, 2026

    work page 2026

  51. [51]

    Life cycle performance of modular build- ings: A critical review,

    M. Kamali and K. Hewage, “Life cycle performance of modular build- ings: A critical review,”Renewable and sustainable energy reviews, vol. 62, pp. 1171–1183, 2016

    work page 2016

  52. [52]

    Lithium iron phosphate based battery–assessment of the aging parameters and development of cycle life model,

    N. Omar, M. A. Monem, Y . Firouz, J. Salminen, J. Smekens, O. Hegazy, H. Gaulous, G. Mulder, P. Van den Bossche, T. Coosemans et al., “Lithium iron phosphate based battery–assessment of the aging parameters and development of cycle life model,”Applied Energy, vol. 113, pp. 1575–1585, 2014

    work page 2014

  53. [53]

    Life cycle environmental assessment of lithium-ion and nickel metal hydride bat- teries for plug-in hybrid and battery electric vehicles,

    G. Majeau-Bettez, T. R. Hawkins, and A. H. Strømman, “Life cycle environmental assessment of lithium-ion and nickel metal hydride bat- teries for plug-in hybrid and battery electric vehicles,”Environmental science & technology, vol. 45, no. 10, pp. 4548–4554, 2011

    work page 2011

  54. [54]

    Automotive li- ion batteries: current status and future perspectives,

    Y . Ding, Z. P. Cano, A. Yu, J. Lu, and Z. Chen, “Automotive li- ion batteries: current status and future perspectives,”Electrochemical Energy Reviews, vol. 2, no. 1, pp. 1–28, 2019

    work page 2019

  55. [55]

    Power stabilization for ai training datacenters,

    E. Choukse, B. Warrier, S. Heath, L. Belmont, A. Zhao, H. A. Khan, B. Harry, M. Kappel, R. J. Hewett, K. Dattaet al., “Power stabilization for ai training datacenters,”arXiv preprint arXiv:2508.14318, 2025

    work page arXiv 2025

  56. [56]

    Grid forming functional specifications for bps-connected battery energy storage systems,

    North American Electric Reliability Corporation (NERC), “Grid forming functional specifications for bps-connected battery energy storage systems,” North American Electric Reliability Corporation, White Paper, Sep. 2023. [Online]. Available: https://www.nerc.com/globalassets/our-work/reports/white- papers/white paper gfm functional specification.pdf

    work page 2023

  57. [57]

    Understanding bess: Battery energy storage systems for data centers,

    P. Donovan, “Understanding bess: Battery energy storage systems for data centers,” Schneider Electric Energy Management Research Center, Tech. Rep. White Paper 185, 2025

    work page 2025

  58. [58]

    Understanding ai load profiles and their impact on power systems,

    Quanta Tech LLC, “Understanding ai load profiles and their impact on power systems,” Online Seminar (Webinar), Sep. 2025

    work page 2025

  59. [59]

    Ai-driven data center energy profile, power quality, sustainable sitting, and energy management: A comprehensive survey,

    R. R. Ahrabi, A. Mousavi, E. Mohammadi, R. Wu, and A. K. Chen, “Ai-driven data center energy profile, power quality, sustainable sitting, and energy management: A comprehensive survey,” in2025 IEEE Conference on Technologies for Sustainability (SusTech). IEEE, 2025, pp. 1–8

    work page 2025

  60. [60]

    Data centers – a good grid citizen,

    K. Watson, “Data centers – a good grid citizen,” Presentation, Eaton, Jul. 2025, slide on grid support and batteries; Mission Critical Solutions presentation

    work page 2025

  61. [61]

    Coordinated fast frequency response from electric vehicles, data centers, and battery energy storage systems,

    X. Tao and R. Gadh, “Coordinated fast frequency response from electric vehicles, data centers, and battery energy storage systems,” arXiv preprint arXiv:2512.14136, 2025

    work page arXiv 2025

  62. [62]

    Impact of reverse power flow on distributed transformers in a solar-photovoltaic-integrated low-voltage network,

    I. B. Majeed and N. I. Nwulu, “Impact of reverse power flow on distributed transformers in a solar-photovoltaic-integrated low-voltage network,”Energies, vol. 15, no. 23, p. 9238, 2022

    work page 2022

  63. [63]

    Salt cavern redox flow battery: The next- generation long-duration, large-scale energy storage system,

    L. Pan, M. Song, N. Muzaffar, L. Chen, C. Ji, S. Yao, J. Xu, W. Wu, Y . Li, J. Chenet al., “Salt cavern redox flow battery: The next- generation long-duration, large-scale energy storage system,”Current Opinion in Electrochemistry, vol. 49, p. 101604, 2025

    work page 2025

  64. [64]

    Multi-objective optimization of demand response in a datacenter with lithium-ion battery storage,

    A. Mamun, I. Narayanan, D. Wang, A. Sivasubramaniam, and H. Fathy, “Multi-objective optimization of demand response in a datacenter with lithium-ion battery storage,”Journal of Energy Storage, vol. 7, pp. 258–269, 2016

    work page 2016

  65. [65]

    A grid-forming energy storage system capacity planning method considering device lifetime,

    G. Ye, J. Fang, N. Wang, Y . Gaogao, and K. Sun, “A grid-forming energy storage system capacity planning method considering device lifetime,”Energies, 2026

    work page 2026

  66. [66]

    Reliability and economic impacts of utilizing battery energy storage in data centers for energy flexibility services in smart grids,

    Y . Yu, K. Shan, H. Tang, and S. Wang, “Reliability and economic impacts of utilizing battery energy storage in data centers for energy flexibility services in smart grids,”Energy Conversion and Manage- ment, vol. 339, p. 119951, 2025

    work page 2025

  67. [67]

    Optimal sizing of data center battery energy storage system for provision of frequency containment reserve,

    L. Cupelli, N. Barve, and A. Monti, “Optimal sizing of data center battery energy storage system for provision of frequency containment reserve,” inIECON 2017 - 43rd Annual Conference of the IEEE Industrial Electronics Society, 2017, pp. 7185–7190

    work page 2017

  68. [68]

    Recent development of hydrogen and fuel cell technologies: A review,

    L. Fan, Z. Tu, and S. H. Chan, “Recent development of hydrogen and fuel cell technologies: A review,”Energy Reports, vol. 7, pp. 8421– 8446, 2021

    work page 2021

  69. [69]

    Fuel cell technology review: Types, economy, applications, and vehicle-to-grid scheme,

    D. Manzo, R. Thai, H. T. Le, and G. K. Venayagamoorthy, “Fuel cell technology review: Types, economy, applications, and vehicle-to-grid scheme,”Sustainable Energy Technologies and Assessments, vol. 75, p. 104229, 2025

    work page 2025

  70. [70]

    Understanding solid oxide fuel cell hybridization: a critical review,

    I. Nikiforakis, S. Mamalis, and D. Assanis, “Understanding solid oxide fuel cell hybridization: a critical review,”Applied Energy, vol. 377, p. 124277, 2025. 21

    work page 2025

  71. [71]

    A grid-forming control method for pemfc power conversion systems with power ramp rate limitation to prevent fuel starvation,

    I.-B. Kong, W.-S. Kim, and S. Chae, “A grid-forming control method for pemfc power conversion systems with power ramp rate limitation to prevent fuel starvation,”IEEE Open Journal of Power Electronics, 2025

    work page 2025

  72. [72]

    Enabling net zero data centers: A techno-economic analysis of bloom energy’s sofc systems,

    F. Perez, “Enabling net zero data centers: A techno-economic analysis of bloom energy’s sofc systems,” Ph.D. dissertation, Politecnico di Torino, 2025

    work page 2025

  73. [73]

    Power ramp rate capabilities of a 5 kw proton exchange membrane fuel cell system with discrete ejector control,

    K. Nikiforow, J. Pennanen, J. Ihonen, S. Uski, and P. Koski, “Power ramp rate capabilities of a 5 kw proton exchange membrane fuel cell system with discrete ejector control,”Journal of Power Sources, vol. 381, pp. 30–37, 2018

    work page 2018

  74. [74]

    AI load dynamics–a power electronics perspective,

    Y . Li and Y . Li, “Ai load dynamics–a power electronics perspective,” arXiv preprint arXiv:2502.01647, 2025

    work page arXiv 2025

  75. [75]

    Techno-economic assessment framework for 2.5 mw-scale grid-connected proton exchange membrane fuel cell power systems: A case study in china,

    P. Zheng, X. Xie, C. Zhang, S. Cai, J. Pan, H. Zhang, M. Yan, and Q. Mu, “Techno-economic assessment framework for 2.5 mw-scale grid-connected proton exchange membrane fuel cell power systems: A case study in china,”International Journal of Hydrogen Energy, vol. 167, p. 150942, 2025

    work page 2025

  76. [76]

    An overview of thermal energy storage systems,

    G. Alva, Y . Lin, and G. Fang, “An overview of thermal energy storage systems,”Energy, vol. 144, pp. 341–378, 2018

    work page 2018

  77. [77]

    Thermal energy storage for grid applications: Current status and emerging trends,

    D. Enescu, G. Chicco, R. Porumb, and G. Seritan, “Thermal energy storage for grid applications: Current status and emerging trends,” Energies, vol. 13, no. 2, p. 340, 2020

    work page 2020

  78. [78]

    Ai-driven optimization of fan-wall cooling system in a medium-density data center,

    M. Omrani and M. Ghassemi, “Ai-driven optimization of fan-wall cooling system in a medium-density data center,”International Journal of Heat and Mass Transfer, vol. 247, p. 127159, 2025

    work page 2025

  79. [79]

    Thermal energy storage in district heating and cooling systems: A review,

    E. Guelpa and V . Verda, “Thermal energy storage in district heating and cooling systems: A review,”Applied Energy, vol. 252, p. 113474, 2019

    work page 2019

  80. [80]

    A comprehensive review of thermal energy storage,

    I. Sarbu and C. Sebarchievici, “A comprehensive review of thermal energy storage,”Sustainability, vol. 10, no. 1, p. 191, 2018

    work page 2018

Showing first 80 references.