pith. machine review for the scientific record. sign in

arxiv: 2604.09705 · v1 · submitted 2026-04-07 · 💻 cs.NI

Recognition: no theorem link

Sustainability-Constrained Workload Orchestration for Sovereign AI Infrastructure: A Joint Compute-Network Optimization Framework

Sergio Cruzes
Authors on Pith no claims yet

Pith reviewed 2026-05-10 18:09 UTC · model grok-4.3

classification 💻 cs.NI
keywords sustainability-constrained orchestrationsovereign AI infrastructurejoint compute-network optimizationfeasible sovereign operating regionenvironmental impact reductionworkload placementoptical network routingcarbon intensity constraints
0
0 comments X

The pith

Joint optimization of compute placement and network routing under strict sustainability constraints reduces environmental impact for sovereign AI infrastructure.

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

This paper proposes a framework that treats carbon intensity, water usage, and power capacity as hard feasibility constraints rather than adjustable penalties when orchestrating AI workloads. It introduces the Feasible Sovereign Operating Region to mark the workloads that a given infrastructure can actually sustain under its physical and regulatory conditions. The method performs a single closed-loop optimization that decides both where to place compute tasks and how to route them across optical networks. Scenario-based comparisons show the joint approach produces lower environmental impact than optimizing compute and network separately. Infeasibility events are interpreted as precise signals indicating when infrastructure investment or workload reduction is required.

Core claim

The paper claims that modeling carbon intensity, water usage, and power capacity as strict feasibility constraints, then jointly optimizing compute placement and optical network routing in one closed-loop system, produces lower environmental impact than baseline separate optimizations and defines the Feasible Sovereign Operating Region of workloads that infrastructure can sustain, with infeasibility events serving as telemetry-grounded indicators for investment or scaling decisions.

What carries the argument

The Feasible Sovereign Operating Region (FSOR), defined as the set of workloads an infrastructure can sustain under its physical and regulatory endowment, which drives the joint closed-loop optimization of compute placement and optical network routing.

If this is right

  • Joint optimization yields lower environmental impact relative to baseline formulations that optimize compute and network independently.
  • Infeasibility events function as precise, telemetry-grounded signals that sovereign AI operation requires infrastructure investment or workload reduction.
  • Treating sustainability metrics as strict feasibility constraints rather than tunable penalties produces more realistic operational boundaries.
  • The single closed-loop system integrates compute placement and network routing decisions under one set of environmental limits.

Where Pith is reading between the lines

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

  • Real-time telemetry feeds could convert the FSOR into a live dashboard that adjusts workloads as energy conditions change throughout the day.
  • The same constraint-first approach could apply to other resource-capped settings such as edge computing clusters or shared scientific facilities.
  • Infeasibility signals could inform regulatory or investment decisions about where to site new AI data centers or upgrade power grids.

Load-bearing premise

The scenario-based analysis accurately models real infrastructure constraints and the FSOR construct can be practically implemented and measured in operational systems.

What would settle it

Running the joint optimization framework on live sovereign AI infrastructure, recording measured carbon, water, and power outcomes plus infeasibility frequency, and directly comparing them to results from a baseline separate-optimization orchestrator would confirm or refute the lower-impact claim.

Figures

Figures reproduced from arXiv: 2604.09705 by Sergio Cruzes.

Figure 1
Figure 1. Figure 1: The green-but-far effect as a constraint geometry. Panel A shows the two-dimensional space of candidate sites partitioned by the carbon/water sustainability threshold Γ¯ i (green dashed line) and the latency budget λk (red dashed line). The FSOR interior (green region) is the intersection of the sustainability-eligible and latency-admissible half-planes; the green-but-far zone (orange region) contains site… view at source ↗
Figure 2
Figure 2. Figure 2: Closed-loop control architecture for sustainability-constrained AI infrastructure orchestration. Streaming telemetry from compute, energy, cooling, and optical network domains is ingested and normalized to provide a unified observability plane. A state estimation and prediction module derives both the current and a short-horizon forecast representation of infrastructure state, which jointly parameterize th… view at source ↗
Figure 3
Figure 3. Figure 3: Feasible Sovereign Operating Region (FSOR) under three representative infrastructure profiles. In each panel, the shaded area denotes the feasible operating region: the set of workloads that simultaneously satisfy all active constraints, including energy availability, carbon intensity, water usage, and network latency. Region A (clean energy access) exhibits a larger sustainability envelope, but the latenc… view at source ↗
read the original abstract

AI infrastructure has transitioned from a software-centric paradigm to a system tightly bound by physical and environmental limits. Energy availability, cooling capacity, and network connectivity now impose hard operational boundaries that cannot be relaxed through software optimization alone. This paper proposes a sustainability-constrained orchestration framework that treats carbon intensity, water usage, and power capacity as strict feasibility constraints rather than tunable penalties, and that jointly optimizes compute placement and optical network routing in a single closed-loop system. We introduce the Feasible Sovereign Operating Region (FSOR) - a conceptual and operational construct that characterizes the set of workloads a given infrastructure can actually sustain under its physical and regulatory endowment. Scenario-based analysis demonstrates that joint optimization yields lower environmental impact relative to baseline formulations. Infeasibility events, rather than being optimizer failures, constitute precise, telemetry-grounded signals that sovereign AI operation requires infrastructure investment or workload reduction.

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 proposes a sustainability-constrained workload orchestration framework for sovereign AI infrastructure. It treats carbon intensity, water usage, and power capacity as hard feasibility constraints, jointly optimizes compute placement and optical network routing in a closed-loop system, and introduces the Feasible Sovereign Operating Region (FSOR) as a construct characterizing sustainable workloads under physical and regulatory limits. Scenario-based analysis is claimed to show that this joint optimization reduces environmental impact relative to baselines, with infeasibility events interpreted as telemetry-grounded signals for infrastructure investment or workload reduction rather than optimizer failures.

Significance. If the framework were equipped with explicit mathematical formulations, constraint sets, scenario definitions, and reproducible validation data, it could offer a meaningful contribution to sustainable AI systems by shifting from penalty-based to feasibility-constrained optimization and by reframing infeasibility as an operational diagnostic. The joint compute-network scope and sovereign-infrastructure focus address an emerging intersection of networking and environmental constraints that is currently underexplored in the literature.

major comments (3)
  1. The abstract asserts that 'scenario-based analysis demonstrates that joint optimization yields lower environmental impact relative to baseline formulations,' yet the manuscript supplies no model equations, objective function, constraint set, scenario definitions, input data, or quantitative results. Without these elements the central empirical claim cannot be evaluated or reproduced.
  2. The Feasible Sovereign Operating Region (FSOR) is presented as both a 'conceptual and operational construct' that is 'computable from telemetry,' but no definition, algorithm, or telemetry mapping is provided. This leaves the load-bearing construct for the paper's operational claims without a concrete realization.
  3. The claim that infeasibility events constitute 'precise, telemetry-grounded signals' for investment or workload reduction is asserted without any example telemetry traces, threshold definitions, or mapping from optimization output to actionable decisions, rendering the interpretive contribution unverifiable.
minor comments (2)
  1. The term 'sovereign AI infrastructure' is used repeatedly but never defined with respect to regulatory, geographic, or ownership boundaries; a brief clarifying paragraph would improve readability.
  2. The abstract refers to 'baseline formulations' without naming or citing the specific baselines (e.g., separate compute or network optimizers) that are being compared.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the insightful comments, which correctly identify that the submitted manuscript lacks the explicit technical details needed to substantiate its central claims. We will revise the paper to incorporate full mathematical formulations, definitions, algorithms, and illustrative examples as outlined below.

read point-by-point responses

  1. Referee: The abstract asserts that 'scenario-based analysis demonstrates that joint optimization yields lower environmental impact relative to baseline formulations,' yet the manuscript supplies no model equations, objective function, constraint set, scenario definitions, input data, or quantitative results. Without these elements the central empirical claim cannot be evaluated or reproduced.

    Authors: The referee is correct; the submitted version omitted the detailed optimization model, equations, and results. In the revision we will add a dedicated technical section presenting the joint compute-network optimization formulation (objective function minimizing environmental impact subject to hard sustainability constraints), the full constraint set for carbon intensity, water usage, and power capacity, explicit scenario definitions drawn from telemetry, input data sources, and quantitative comparison results against baselines. revision: yes

  2. Referee: The Feasible Sovereign Operating Region (FSOR) is presented as both a 'conceptual and operational construct' that is 'computable from telemetry,' but no definition, algorithm, or telemetry mapping is provided. This leaves the load-bearing construct for the paper's operational claims without a concrete realization.

    Authors: We acknowledge the absence of a formal definition and computation procedure in the current text. The revised manuscript will include a precise set-theoretic definition of FSOR, an algorithm for computing the region boundaries from telemetry (carbon intensity, water, and power traces), and the explicit mapping from infrastructure parameters to feasible workload sets. revision: yes

  3. Referee: The claim that infeasibility events constitute 'precise, telemetry-grounded signals' for investment or workload reduction is asserted without any example telemetry traces, threshold definitions, or mapping from optimization output to actionable decisions, rendering the interpretive contribution unverifiable.

    Authors: The referee correctly notes the lack of concrete examples. We will augment the paper with sample telemetry traces, explicit threshold definitions for detecting infeasibility, and a step-by-step mapping from optimization outputs (such as constraint violations or dual prices) to operational decisions regarding infrastructure investment or workload scaling. revision: yes

Circularity Check

0 steps flagged

No circularity detected; derivation remains self-contained

full rationale

The abstract and available description introduce the FSOR construct and a joint optimization framework at a conceptual level, then assert that scenario-based analysis shows lower environmental impact. No equations, objective functions, constraint formulations, or derivation steps are presented that reduce any claimed prediction or result back to its own inputs by definition, fitted parameters renamed as outputs, or load-bearing self-citation chains. The framework treats external physical limits as constraints and treats infeasibility as a signal, but these are presented as modeling choices rather than tautological redefinitions. Absent any quoted reduction of the form 'prediction X equals fitted input Y by construction,' the analysis chain does not exhibit circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 1 invented entities

The central claim rests on the validity of the FSOR as an operational construct and the premise that joint optimization under hard constraints produces measurable environmental benefits; no free parameters, standard axioms, or external evidence for the new entity are provided in the abstract.

invented entities (1)
  • Feasible Sovereign Operating Region (FSOR) no independent evidence
    purpose: Characterizes the set of workloads a given infrastructure can sustain under its physical and regulatory endowment.
    Introduced as a conceptual and operational construct in the abstract with no independent evidence or prior validation referenced.

pith-pipeline@v0.9.0 · 5443 in / 1346 out tokens · 72039 ms · 2026-05-10T18:09:47.625225+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

78 extracted references · 54 canonical work pages · 1 internal anchor

  1. [1]

    rep., IEA, Paris, licence: CC BY 4.0 (2025)

    International Energy Agency, Energy and AI , Tech. rep., IEA, Paris, licence: CC BY 4.0 (2025). URL https://www.iea.org/reports/energy-and-ai

    2025

  2. [2]

    Mytton, M

    D. Mytton, M. Ashtine, The ecology of artificial intelligence: energy, water, materials, and land limits of digital systems , Carbon Neutral Systems (Dec. 2025). doi:10.1007/ s44438-025-00018-8 . URL https://link.springer.com/article/10.1007/s44438-025-00018-8

    work page doi:10.1007/s44438-025-00018-8 2025

  3. [3]

    Mytton, Data centre water consumption , npj Clean Water 4 (1) (2021) 11

    D. Mytton, Data centre water consumption , npj Clean Water 4 (1) (2021) 11. doi: 10.1038/s41545-021-00101-w . URL https://doi.org/10.1038/s41545-021-00101-w

    work page doi:10.1038/s41545-021-00101-w 2021

  4. [4]

    Y. Ye, T. Huang, Z. Shi, Y. Luo, X. Zhang, Artificial intelligence adoption for advancing energy justice: a multidimensional perspective , Scientific Reports (mar 2026). doi: 10.1038/s41598-026-38421-x . URL https://doi.org/10.1038/s41598-026-38421-x

    work page doi:10.1038/s41598-026-38421-x 2026

  5. [5]

    B. W. Nkwawir, M. Ö. Kayalica, D. Ünal, A. C. Duman, H. S. Erden, Carbon-aware workload management in data centers: A multi-energy integration approach , in: Pro- ceedings of the 16th ACM International Conference on Future Energy Systems (e- April 14, 2026 Energy ’25), ACM, New York, NY, USA, 2025. doi:10.1145/3679240.3735104. URL https://dl.acm.org/doi/1...

    work page doi:10.1145/3679240.3735104 2026

  6. [6]

    Radovanović, R

    A. Radovanović, R. Koningstein, I. Schneider, B. Chen, A. Duarte, B. Roy, D. Xiao, M. Haridasan, P. Hung, N. Care, S. Talukdar, E. Mullen, K. Smith, M. Körner, R. Schneider, Carbon-aware computing for datacenters, IEEE Transactions on Power Systems 38 (2) (2023) 1270–1280. doi:10.1109/TPWRS.2022.3173250

    work page doi:10.1109/tpwrs.2022.3173250 2023

  7. [7]

    L. Cote, A. Sun, Locational marginal emissions for carbon-aware data center operations in large-scale power grids , arXiv preprint arXiv:2512.18819 (2025). URL https://arxiv.org/abs/2512.18819

    work page arXiv 2025

  8. [8]

    Singla, B

    A. Singla, B. Chandrasekaran, P. B. Godfrey, B. Maggs, The Internet at the speed of light, in: Proceedings of the 13th ACM Workshop on Hot Topics in Networks (HotNets- XIII), ACM, Los Angeles, CA, 2014, pp. 1–7. doi:10.1145/2670518.2673876

    work page doi:10.1145/2670518.2673876 2014

  9. [9]

    Nicolás, The bar derived category of a curved dg algebra, Journal of Pure and Applied Algebra 212 (2008) 2633–2659

    B. Cottier, et al., Advances and challenges in energy and climate alignment of AI infrastructure expansion , Cell Reports Sustainability (Sep. 2025). doi:10.1016/j. crsus.2025.100372. URL https://www.sciencedirect.com/science/article/pii/S266679242500037X

    work page doi:10.1016/j 2025

  10. [10]

    Cruzes, Telemetry and agentic AI: Foundations for optical network automation, IEEE Access 14 (2026) 8800–8838

    S. Cruzes, Telemetry and agentic AI: Foundations for optical network automation, IEEE Access 14 (2026) 8800–8838. doi:10.1109/ACCESS.2025.3649768

    work page doi:10.1109/access.2025.3649768 2026

  11. [11]

    AI Infrastructure Sovereignty

    S. Cruzes, AI infrastructure sovereignty , arXiv preprint arXiv:2602.10900 (2026). URL https://arxiv.org/abs/2602.10900

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  12. [12]

    B. Acun, B. Lee, F. Kazhamiaka, K. Maeng, U. Gupta, M. Chakkaravarthy, D. Brooks, C.-J. Wu, Carbon explorer: A holistic framework for designing carbon aware data- centers, in: Proceedings of the 28th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2, ASPLOS 2023, Association for Computing Machi...

    work page doi:10.1145/3575693.3575754 2023

  13. [13]

    Sukprasert, A

    T. Sukprasert, A. Souza, N. Bashir, D. Irwin, P. Shenoy, On the limitations of carbon- aware temporal and spatial workload shifting in the cloud, in: Proceedings of the Nineteenth European Conference on Computer Systems (EuroSys ’24), ACM, Athens, Greece, 2024, pp. 1–18. arXiv:2306.06502, doi:10.1145/3627703.3650079

    work page doi:10.1145/3627703.3650079 2024

  14. [14]

    Buyya, A

    R. Buyya, A. Beloglazov, J. Abawajy, Energy-efficient management of data center re- sources for cloud computing: A vision, architectural elements, and open challenges, IEEE Transactions on Parallel and Distributed Systems 24 (7) (2013) 1396–1407. doi:10.1109/TPDS.2012.309. April 14, 2026

    work page doi:10.1109/tpds.2012.309 2013

  15. [15]

    Silva, R

    C. Silva, R. Vilaça, A. Pereira, R. Bessa, A review on the decarbonization of high- performance computing centers, Renewable and Sustainable Energy Reviews 189 (2024) 114019. doi:10.1016/j.rser.2023.114019. URL https://www.sciencedirect.com/science/article/pii/S1364032123008778

    work page doi:10.1016/j.rser.2023.114019 2024

  16. [16]

    Hoxha, M

    J. Hoxha, M. Thanasi-Boçe, T. Khalifa, A deployment-aware framework for carbon- and water-efficient LLM serving, Sustainability 17 (23) (2025) 10473

    2025

  17. [17]

    Asadov, V

    N. Asadov, V. C. Coroam, M. Franzil, S. Galantino, M. Finkbeiner, Carbon-aware spatio-temporal workload shifting in edgecloud environments: A review and novel al- gorithm, Sustainability 17 (14) (2025). doi:10.3390/su17146433. URL https://www.mdpi.com/2071-1050/17/14/6433

    work page doi:10.3390/su17146433 2025

  18. [18]

    Sallam, K

    M. Sallam, K. Kaur, Optimized resource forecasting for carbon-intelligent data centers with temposight: A hybrid deep learning approach, IEEE Internet of Things Journal 12 (23) (2025) 49884–49903. doi:10.1109/JIOT.2025.3610895

    work page doi:10.1109/jiot.2025.3610895 2025

  19. [19]

    J. Xin, X. Li, D. Kilper, S. Huang, Load-balance-guaranteed DNN distributed inference offloading in MEC networks interconnected by metro optical networks, IEEE Transac- tions on Network Science and Engineering 13 (2026) 3391–3408. doi:10.1109/TNSE. 2025.3637030

    work page doi:10.1109/tnse 2026

  20. [20]

    R. Gu, Z. Yang, Y. Ji, Machine learning for intelligent optical networks: A compre- hensive survey, Journal of Network and Computer Applications 157 (2020) 102576. doi:10.1016/j.jnca.2020.102576

    work page doi:10.1016/j.jnca.2020.102576 2020

  21. [21]

    Emergent Mind, Carbon-aware scheduling, https://www.emergentmind.com/topics/ carbon-aware-scheduling, accessed: 2026-03-24 (2025)

    2026

  22. [22]

    Fratini, O

    S. Fratini, O. Hine, Digital sovereignty: A descriptive analysis and a critical evaluation of existing models, Digital Society (2024). doi:10.1007/s44206-024-00146-7

    work page doi:10.1007/s44206-024-00146-7 2024

  23. [23]

    S. B. Chetty, D. Grace, S. Saunders, P. Harris, E. E. Tsiropoulou, T. Quek, H. Ah- madi, Sovereign AI for 6G: Towards the future of AI-native networks , arXiv preprint arXiv:2509.06700 (2025). doi:10.48550/arXiv.2509.06700. URL https://arxiv.org/abs/2509.06700

    work page doi:10.48550/arxiv.2509.06700 2025

  24. [24]

    Lehdonvirta, B

    V. Lehdonvirta, B. Wú, Z. Hawkins, Compute north vs. compute south: The un- even possibilities of compute-based ai governance around the globe , Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society 7 (1) (2024) 828–838. doi: 10.1609/aies.v7i1.31683. URL https://ojs.aaai.org/index.php/AIES/article/view/31683

    work page doi:10.1609/aies.v7i1.31683 2024

  25. [25]

    Shehabi, S

    A. Shehabi, S. J. Smith, A. Hubbard, A. Newkirk, N. Lei, M. A. B. Siddik, B. Holecek, J. Koomey, E. Masanet, D. Sartor, 2024 united states data center energy usage report , Tech. Rep. LBNL-2001637, Lawrence Berkeley National Laboratory, energy Analysis April 14, 2026 and Environmental Impacts Division, Contract No. DE-AC02-05CH11231 (dec 2024). URL https:...

    2024

  26. [26]

    R. E. Leenes, Framing techno-regulation: An exploration of state and non-state regula- tion by technology, Legisprudence 5 (2) (2011) 143–169, argues that norms embedded in ICT platforms by private providers have binding regulatory effect on tenants, inde- pendently of state law. doi:10.5235/175214611797885675

    work page doi:10.5235/175214611797885675 2011

  27. [27]

    Swire, K

    P. Swire, K. Ahmad, Encryption and globalization, Columbia Science and Technology Law Review 13 (2012) 416–481, analyzes how U.S. legal jurisdiction over cloud providers extends to tenant data and operations regardless of the tenant’s geographic location. doi:10.2139/ssrn.1960602

    work page doi:10.2139/ssrn.1960602 2012

  28. [28]

    468 at 1550 nm (Nov

    ITU-T, Characteristics of a single-mode optical fibre and cable , Recommendation G.652, International Telecommunication Union, defines geometrical, mechanical, and transmission attributes of dispersion-unshifted single-mode fibre (SMF-28 class); group refractive index n ≈ 1. 468 at 1550 nm (Nov. 2016). URL https://www.itu.int/rec/T-REC-G.652-201611-S/en

    2016

  29. [29]

    H. Liu, X. Hu, R. Wang, J. Hao, Q. Wu, H. Zhang, Green scheduling for LLM workloads with model and data reuse across geo-distributed data centers , Digital Communications and Networks 12 (2) (2026) 236–251. doi:10.1016/j.dcan.2025.11.006. URL https://www.sciencedirect.com/science/article/pii/S2352864825001828

    work page doi:10.1016/j.dcan.2025.11.006 2026

  30. [30]

    Moore, S

    H. Moore, S. Qi, N. Hogade, D. Milojicic, C. Bash, S. Pasricha, Sustainable carbon- aware and water-efficient LLM scheduling in geo-distributed cloud datacenters , in: Pro- ceedings of the Great Lakes Symposium on VLSI 2025, 2025. doi:10.1145/3716368. 3735301. URL https://dl.acm.org/doi/full/10.1145/3716368.3735301

    work page doi:10.1145/3716368 2025

  31. [31]

    Y. Wei, T. Hu, C. Liang, Y. Cui, Communication optimization for distributed training: Architecture, advances, and opportunities, IEEE Network 39 (3) (2025) 241–248. doi: 10.1109/MNET.2024.3449276

    work page doi:10.1109/mnet.2024.3449276 2025

  32. [32]

    Gujarati, R

    A. Gujarati, R. Karimi, S. Alzayat, W. Hao, A. Kaufmann, Y. Vigfusson, J. Mace, Serving dnns like clockwork: Performance predictability from the bottom up (2020). arXiv:2006.02464. URL https://arxiv.org/abs/2006.02464

    work page arXiv 2020

  33. [33]

    R. S. Tucker, Green optical communications—part I: Energy limitations in transport, IEEE Journal of Selected Topics in Quantum Electronics 17 (2) (2011) 245–260. doi: 10.1109/JSTQE.2010.2044266

    work page doi:10.1109/jstqe.2010.2044266 2011

  34. [34]

    2009.A Statistical Learning Perspective of Genetic Programming

    T. Achterberg, R. Wunderling, Mixed Integer Programming: Analyzing 12 Years of Progress, Springer Berlin Heidelberg, 2013, pp. 449–481. doi:10.1007/978-3-642- 38189-8_18. April 14, 2026

    work page doi:10.1007/978-3-642- 2013

  35. [35]

    Gouveia, Multicommodity flow models for spanning trees with hop constraints, Eu- ropean Journal of Operational Research 95 (1) (1996) 178–190

    L. Gouveia, Multicommodity flow models for spanning trees with hop constraints, Eu- ropean Journal of Operational Research 95 (1) (1996) 178–190. doi:10.1016/0377- 2217(95)00337-1

    work page doi:10.1016/0377- 1996

  36. [36]

    R. E. Bixby, A brief history of linear and mixed-integer programming computation , in: Documenta Mathematica Extra Volume ISMP (2012), EMS Press, 2012, pp. 107–121. doi:10.4171/DM/2012. URL https://ems.press/content/book-chapter-files/27357

    work page doi:10.4171/dm/2012 2012

  37. [37]

    M. R. Garey, D. S. Johnson, Computers and Intractability: A Guide to the Theory of NP-Completeness, W. H. Freeman, New York, 1979

    1979

  38. [38]

    Cornuéjols, G

    G. Cornuéjols, G. L. Nemhauser, L. A. Wolsey, The uncapacitated facility location problem, in: P. B. Mirchandani, R. L. Francis (Eds.), Discrete Location Theory, Wiley & Sons, New York, 1990, pp. 119–171

    1990

  39. [39]

    J. F. Benders, Partitioning procedures for solving mixed-variables programming prob- lems, Numerische Mathematik 4 (1) (1962) 238–252. doi:10.1007/BF01386316

    work page doi:10.1007/bf01386316 1962

  40. [40]

    The Benders decomposition algorithm: A literature review.European Journal of Operational Research, 259(3):801– 817, 2017.doi:10.1016/j.ejor.2016.12.005

    R. Rahmaniani, T. G. Crainic, M. Gendreau, W. Rei, The Benders decomposition algorithm: A literature review, European Journal of Operational Research 259 (3) (2017) 801–817. doi:10.1016/j.ejor.2016.12.005

    work page doi:10.1016/j.ejor.2016.12.005 2017

  41. [41]

    Bonami, D

    P. Bonami, D. Salvagnin, A. Tramontani, Implementing automatic Benders decom- position in a modern MIP solver, in: D. Bienstock, G. Zambelli (Eds.), Integer Pro- gramming and Combinatorial Optimization, Vol. 12125 of Lecture Notes in Computer Science, Springer, Cham, 2020, pp. 78–90, proceedings of the 21st International Confer- ence on Integer Programmin...

    work page doi:10.1007/978-3-030-45771-6_7 2020

  42. [42]

    M. L. Fisher, The Lagrangian relaxation method for solving integer programming prob- lems, Management Science 27 (1) (1981) 1–18. doi:10.1287/mnsc.27.1.1

    work page doi:10.1287/mnsc.27.1.1 1981

  43. [43]

    L. A. Wolsey, Integer Programming, Wiley Series in Discrete Mathematics and Opti- mization, Wiley-Interscience, New York, 1998, chapter 10 covers Lagrangian relaxation, duality gaps, and conditions under which the gap is small for structured MILPs

    1998

  44. [44]

    European Commission, 2030 digital compass: The european way for the digital decade , Communication COM(2021) 118 final, European Commission, sets 2030 targets for climate-neutral digital infrastructure, including 10,000 climate-neutral edge nodes and energy/water efficiency requirements for data centers (Mar. 2021). URL https://eur-lex.europa.eu/legal-conten...

    2030

  45. [45]

    European Commission, Fostering a european approach to artificial intelligence, Commu- nication COM(2021) 205 final, European Commission, 2021 review of the Coordinated Plan on AI; sets out national AI strategy alignment and sustainability obligations for April 14, 2026 AI deployment (Apr. 2021). URL https://eur-lex.europa.eu/legal-content/EN/ALL/?uri=COM:20...

    2021

  46. [46]

    contribution of working group I to the sixth assessment report of the intergovernmental panel on climate change, Tech

    IPCC, Climate change 2021: The physical science basis. contribution of working group I to the sixth assessment report of the intergovernmental panel on climate change, Tech. rep., Cambridge University Press (2021). doi:10.1017/9781009157896

    work page doi:10.1017/9781009157896 2021

  47. [47]

    Moret, F

    S. Moret, F. Babonneau, M. Bierlaire, F. Maréchal, Decision support for strategic energy planning: A robust optimization framework, European Journal of Operational Research 280 (2) (2020) 539–554, available online June 2019; citation key reflects online- first publication date. doi:10.1016/j.ejor.2019.06.015

    work page doi:10.1016/j.ejor.2019.06.015 2020

  48. [48]

    Wörman, G

    A. Wörman, G. Lindström, J. Riml, Spatiotemporal management of solar, wind and hydropower across continental Europe, Communications Engineering 3 (1) (2024) 3. doi:10.1038/s44172-023-00155-3

    work page doi:10.1038/s44172-023-00155-3 2024

  49. [49]

    J. W. Chinneck, Feasibility and Infeasibility in Optimization: Algorithms and Compu- tational Methods, Vol. 118 of International Series in Operations Research & Manage- ment Science, Springer, New York, 2008. doi:10.1007/978-0-387-74932-7

    work page doi:10.1007/978-0-387-74932-7 2008

  50. [50]

    J. B. Rawlings, D. Q. Mayne, Model Predictive Control: Theory and Design , Nob Hill Publishing, Madison, WI, 2009, xi + 567 pages. URL https://www.mpc.berkeley.edu/

    2009

  51. [51]

    URL https://www.dmtf.org/sites/default/files/standards/documents/DSP0266_1.21

    DMTF, Redfish specification , Standard Specification DSP0266, Distributed Manage- ment Task Force (2024). URL https://www.dmtf.org/sites/default/files/standards/documents/DSP0266_1.21. 1.html

    2024

  52. [52]

    Electricity Maps, Marginal vs average: Which one to use for real-time decisions? , [Online; accessed <date>] (Mar. 2026). URL https://www.electricitymaps.com/resources/publications/marginal-vs-average- which-one-to-use-for-real-time-decisions

    2026

  53. [53]

    Shakir, A

    R. Shakir, A. Shaikh, P. Borman, M. Hines, C. Lebsack, C. Morrow, gRPC net- work management interface (gNMI) , Internet-Draft draft-openconfig-rtgwg-gnmi-spec- 01, IETF (2018). URL https://datatracker.ietf.org/doc/html/draft-openconfig-rtgwg-gnmi-spec-01

    2018

  54. [54]

    Daskal, Borders and bits , Vanderbilt Law Review 71 (1) (2018) 179–246

    J. Daskal, Borders and bits , Vanderbilt Law Review 71 (1) (2018) 179–246. URL https://scholarship.law.vanderbilt.edu/vlr/vol71/iss1/3

    2018

  55. [55]

    A. Chander, Sovereignty 2.0, Vanderbilt Journal of Transnational Law 55 (2) (2022) 283–??? URL https://scholarship.law.vanderbilt.edu/cgi/viewcontent.cgi?article=2748& context=vjtl April 14, 2026

    2022

  56. [56]

    URL https://digital-strategy.ec.europa.eu/en/library/coordinated-plan-artificial- intelligence-2021-review

    European Commission, Coordinated plan on artificial intelligence 2021 review , Com- munication COM(2021) 205 final, European Commission, Brussels (2021). URL https://digital-strategy.ec.europa.eu/en/library/coordinated-plan-artificial- intelligence-2021-review

    2021

  57. [57]

    Koronen, M

    C. Koronen, M. Åhman, L. J. Nilsson, Data centres in future European energy systems—energy efficiency, integration and policy, Energy Efficiency 13 (1) (2020) 129–

    2020

  58. [58]

    doi:10.1007/s12053-019-09833-8

    work page doi:10.1007/s12053-019-09833-8

  59. [59]

    P. X. Gao, A. R. Curtis, B. Wong, S. Keshav, It’s not easy being green, in: Proceedings of the ACM SIGCOMM 2012 Conference, SIGCOMM ’12, ACM, Helsinki, Finland, 2012, pp. 211–222. doi:10.1145/2342356.2342398

    work page doi:10.1145/2342356.2342398 2012

  60. [60]

    Zheng, L

    Z. Zheng, L. Rao, X. Liu, Green latency-aware data placement in data centers, Com- puter Networks 110 (2016) 45–58. doi:10.1016/j.comnet.2016.09.016

    work page doi:10.1016/j.comnet.2016.09.016 2016

  61. [61]

    Kosar, et al., Carbon-aware end-to-end data movement, in: Proceedings of the Workshop on Sustainable Computer Systems (HotCarbon ’24), 2024

    T. Kosar, et al., Carbon-aware end-to-end data movement, in: Proceedings of the Workshop on Sustainable Computer Systems (HotCarbon ’24), 2024. arXiv:2406. 09650

    2024

  62. [62]

    Qin, et al., Joint energy optimization on the server and network sides for geo- distributed data centers, The Journal of Supercomputing (2021)

    F. Qin, et al., Joint energy optimization on the server and network sides for geo- distributed data centers, The Journal of Supercomputing (2021). doi:10.1007/ s11227-020-03523-4

    2021

  63. [63]

    Keller, C

    M. Keller, C. Robbert, H. Karl, JASPER: Joint optimization of scaling, placement, and routing for network services, in: arXiv preprint, 2017. arXiv:1711.10839

    work page arXiv 2017

  64. [64]

    Souza, S

    A. Souza, S. Jasoria, B. Chakrabarty, A. Bridgwater, A. Lundberg, F. Frančik, A. Ali- Eldin, D. Irwin, P. Shenoy, CASPER: Carbon-aware scheduling and provisioning for distributed web services, in: Proceedings of the 14th International Green and Sustain- able Computing Conference (IGSC ’23), ACM, Toronto, ON, Canada, 2023, pp. 67–73. arXiv:2403.14792, doi:...

    work page doi:10.1145/3634769.3634812 2023

  65. [65]

    Nocedal, S

    J. Nocedal, S. J. Wright, Numerical Optimization, 2nd Edition, Springer, New York,

  66. [66]

    doi:10.1007/978-0-387-40065-5

    work page doi:10.1007/978-0-387-40065-5

  67. [67]

    D. G. Luenberger, Y. Ye, Linear and Nonlinear Programming, 4th Edition, Springer, Cham, 2016. doi:10.1007/978-3-319-18842-3

    work page doi:10.1007/978-3-319-18842-3 2016

  68. [68]

    Lange, Optimization, 2nd Edition, Springer Texts in Statistics, Springer, New York,

    K. Lange, Optimization, 2nd Edition, Springer Texts in Statistics, Springer, New York,

  69. [69]

    doi:10.1007/978-1-4614-5838-8

    work page doi:10.1007/978-1-4614-5838-8

  70. [70]

    16 Dimitris Bertsimas, David B

    D. Bertsimas, M. Sim, The price of robustness, Operations Research 52 (1) (2004) 35–53. doi:10.1287/opre.1030.0065. April 14, 2026

    work page doi:10.1287/opre.1030.0065 2004

  71. [71]

    J. W. Chinneck, E. W. Dravnieks, Locating minimal infeasible constraint sets in linear programs, ORSA Journal on Computing 3 (2) (1991) 157–168. doi:10.1287/ijoc.3. 2.157

    work page doi:10.1287/ijoc.3 1991

  72. [72]

    Ben-Tal, A

    A. Ben-Tal, A. Nemirovski, Robust solutions of uncertain linear programs, Operations Research Letters 25 (1) (1999) 1–13. doi:10.1016/S0167-6377(99)00016-4

    work page doi:10.1016/s0167-6377(99)00016-4 1999

  73. [73]

    Clark, K

    C. Clark, K. Fraser, S. Hand, J. G. Hansen, E. Jul, C. Limpach, I. Pratt, A. Warfield, Live migration of virtual machines, in: Proceedings of the 2nd USENIX Symposium on Networked Systems Design and Implementation (NSDI ’05), USENIX Association, Boston, MA, 2005, pp. 273–286

    2005

  74. [74]

    Voorsluys, J

    W. Voorsluys, J. Broberg, S. Venugopal, R. Buyya, Cost of virtual machine live mi- gration in clouds: A performance evaluation, in: Proceedings of the 1st International Conference on Cloud Computing (CloudCom ’09), Vol. 5931 of Lecture Notes in Com- puter Science, Springer, 2009, pp. 254–265. doi:10.1007/978-3-642-10665-1_23

    work page doi:10.1007/978-3-642-10665-1_23 2009

  75. [75]

    Lindberg, J

    J. Lindberg, J. Hagberg, J. Rydén, M. Engardt, E. Kjellström, Open-data based carbon emission intensity signals for electricity generation in European countries—top down vs. bottom up approach, Smart Energy 4 (2021) 100060. doi:10.1016/j.segy.2021. 100060

    work page doi:10.1016/j.segy.2021 2021

  76. [76]

    Wiesner, O

    P. Wiesner, O. Kao, Moving beyond marginal carbon intensity: A poor metric for both carbon accounting and grid flexibility (2025). arXiv:2507.11377. URL https://arxiv.org/abs/2507.11377

    work page arXiv 2025

  77. [77]

    Cappart, D

    Q. Cappart, D. Chételat, E. B. Khalil, A. Lodi, C. Morris, P. Velikovi, Combinatorial optimization and reasoning with graph neural networks , Journal of Machine Learning Research 24 (130) (2023) 1–61. URL http://jmlr.org/papers/v24/21-0449.html

    2023

  78. [78]

    Telemetry and Agentic AI: Foundations for Optical Network Automation,

    C. Reinert, B. Nilges, N. Baumgärtner, A. Bardow, This is sparta: Rigorous optimiza- tion of regionally resolved energy systems by spatial aggregation and decomposition (2023). arXiv:2302.05222. URL https://arxiv.org/abs/2302.05222 Author biography Sergio Cruzes received his MSc. degree in electrical engineering from the University of São Paulo (EESC) wit...

    work page arXiv 2023