The DOI matches its Crossref record; the references it cites are not checked yet.
Carbon-Aware Scheduling of AI Data Center Workloads Using Environmental and Energy Grid Forecasts
Open accessAbstract
The global surge in demand for artificial intelligence and cloud computing has led to a rapid expansion of data centers. In 2023, U.S. data centers consumed 176 terawatt-hours (TWh) of electricity – 4.4% of the nation’s total. It is projected to account for up to 12% of national usage by 2028. As data centers become more energy-intensive, they contribute significantly to grid instability and greenhouse gas emissions. The purpose of this research is to design and simulate a carbon-aware workload scheduling algorithm that reduces emissions and energy costs by dynamically adjusting when and where jobs run, using environmental and grid data. The methodology involves simulating AI job-level workloads using open-source workload profiles (e.g., MLPerf benchmarks) and combining them with ComStock data to estimate baseline building-level energy use and thermal load patterns representative of data centers in California. These simulated workloads are integrated with time-series datasets from the California Independent System Operator (CAISO) for hourly grid carbon intensity, Open-Meteo for local temperature forecasts, and PG&E’s time-of-use electricity pricing. The scheduling algorithm, developed in Python, prioritizes non-urgent computing tasks during periods of lower carbon intensity, cooler ambient temperatures (to reduce cooling energy needs), and lower electricity rates. It uses a hybrid decision engine that combines rule-based logic with machine learning to forecast optimal scheduling windows up to 24 hours in advance. To evaluate its effectiveness, simulations compare baseline (unscheduled) workloads against carbon-optimized schedules using key metrics such as estimated CO₂ emissions, total energy cost, and cooling demand. Initial simulations suggest the algorithm could reduce carbon emissions and energy costs by 5–10% for flexible workloads, with even greater gains during periods of high renewable output or extreme heat. Scaled across major data centers, this approach could prevent millions of tons of CO₂ emissions annually, ease strain on electrical grids, and support more efficient integration of renewable energy. This research offers a practical, scalable strategy for reducing the environmental footprint of AI computing – one that aligns digital infrastructure with global climate goals.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
10 Own interpretation · 5 Own result
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Significant load shifting away from peak-carbon and peak-cost hours”
- Also stated in: § Weather & Renewable Forecasts · boosted
“Clear load shifting demonstrates the system's ability to avoid high-carbon and high-price periods.”
§ Carbon-Aware Scheduling of AI Data Center Workloads Using Environmental and Energy Grid Forecasts > Weather & Renewable Forecasts
boosteddirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Also stated in: § Weather & Renewable Forecasts · boosted
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“6% average reduction in CO₂ emissions compared to baseline schedules”
- Also stated in: § Weather & Renewable Forecasts · plain
“Simulations show ~6% reductions in emissions and energy costs without impacting throughput”
- Also stated in: § Abstract · hedged
“Initial simulations suggest the algorithm could reduce carbon emissions and energy costs by 5-10% for flexible workloads”
§ Carbon-Aware Scheduling of AI Data Center Workloads Using Environmental and Energy Grid Forecasts > Weather & Renewable Forecasts
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Also stated in: § Weather & Renewable Forecasts · plain
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The carbon-aware scheduler consistently reduces emissions and cost while meeting power and duration constraints.”
§ Carbon-Aware Scheduling of AI Data Center Workloads Using Environmental and Energy Grid Forecasts > Weather & Renewable Forecasts
boosted- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Simulations show ~6% reductions in emissions and energy costs without impacting throughput”
§ Carbon-Aware Scheduling of AI Data Center Workloads Using Environmental and Energy Grid Forecasts > Weather & Renewable Forecasts
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Integrating hourly carbon intensity, temperature-dependent cooling load, and time-of-use pricing into a unified scheduler enables smart temporal shifting of flexible workloads”
§ Carbon-Aware Scheduling of AI Data Center Workloads Using Environmental and Energy Grid Forecasts > Weather & Renewable Forecasts
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption: Grid carbon intensity varies substantially over short time scales.
- Rests on an assumption: Many AI workloads have flexible execution windows.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Even partial shifting of non-urgent compute can deliver grid-scale benefits when applied across large data-center fleets”
§ Carbon-Aware Scheduling of AI Data Center Workloads Using Environmental and Energy Grid Forecasts > Weather & Renewable Forecasts
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption: Grid carbon intensity varies substantially over short time scales.
- Rests on an assumption: Many AI workloads have flexible execution windows.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Carbon-aware scheduling can measurably reduce the environmental footprint of AI workloads using existing operational data streams”
§ Carbon-Aware Scheduling of AI Data Center Workloads Using Environmental and Energy Grid Forecasts > Weather & Renewable Forecasts
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption: Grid carbon intensity varies substantially over short time scales.
- Rests on an assumption: Many AI workloads have flexible execution windows.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Broad adoption could support grid stability, increase renewable utilization, and help meet decarbonization targets”
§ Carbon-Aware Scheduling of AI Data Center Workloads Using Environmental and Energy Grid Forecasts > Weather & Renewable Forecasts
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption: Grid carbon intensity varies substantially over short time scales.
- Rests on an assumption: Many AI workloads have flexible execution windows.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Scaled across major data centers, this approach could prevent millions of tons of CO2 emissions annually, ease strain on electrical grids, and support more efficient integration of renewable energy.”
§ Carbon-Aware Scheduling of AI Data Center Workloads Using Environmental and Energy Grid Forecasts > Abstract
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption: Grid carbon intensity varies substantially over short time scales.
- Rests on an assumption: Many AI workloads have flexible execution windows.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Results support the feasibility of carbon-optimized AI workload scheduling in real-world utility conditions”
§ Carbon-Aware Scheduling of AI Data Center Workloads Using Environmental and Energy Grid Forecasts > Weather & Renewable Forecasts
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- It may state two separate claims as one. 1
- Probably not a substantial claim. 2
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“~6% cost savings achieved using PG&E Time-of-Use (TOU) rates”
- Also stated in: § Weather & Renewable Forecasts · plain
“Simulations show ~6% reductions in emissions and energy costs without impacting throughput”
- Also stated in: § Abstract · hedged
“Initial simulations suggest the algorithm could reduce carbon emissions and energy costs by 5-10% for flexible workloads”
§ Carbon-Aware Scheduling of AI Data Center Workloads Using Environmental and Energy Grid Forecasts > Weather & Renewable Forecasts
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
- Also stated in: § Weather & Renewable Forecasts · plain
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Broad adoption could support grid stability, increase renewable utilization, and help meet decarbonization targets”
§ Carbon-Aware Scheduling of AI Data Center Workloads Using Environmental and Energy Grid Forecasts > Weather & Renewable Forecasts
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
- Rests on an assumption: Grid carbon intensity varies substantially over short time scales.
- Rests on an assumption: Many AI workloads have flexible execution windows.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Broad adoption could support grid stability, increase renewable utilization, and help meet decarbonization targets”
§ Carbon-Aware Scheduling of AI Data Center Workloads Using Environmental and Energy Grid Forecasts > Weather & Renewable Forecasts
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper may give no reason why it holds.
- Rests on an assumption: Grid carbon intensity varies substantially over short time scales.
- Rests on an assumption: Many AI workloads have flexible execution windows.
The paper relies on this without support.
AI growth is rapidly increasing data-center electricity demand.
“AI growth is rapidly increasing data-center electricity demand (176 TWh in 2023, projected to double within a decade).”
The paper relies on this without support.
Cooling needs heighten stress during heatwaves, when prices and carbon intensity spike.
“Cooling needs heighten stress during heatwaves, when prices and carbon intensity spike.”
The paper relies on this without support.
Annual Net Zero claims can still mask high hourly emissions.
“Annual Net Zero claims can still mask high hourly emissions.”
The paper relies on this without support.
Most AI workloads run on fixed schedules, missing cleaner, lower-cost windows.
“Yet most AI workloads run on fixed schedules, missing cleaner, lower-cost windows.”
The paper relies on this without support.
Grid carbon intensity varies substantially over short time scales.
“Grid carbon intensity varies substantially over short time scales, while many AI workloads have flexible execution windows.”
The paper relies on this without support.
Many AI workloads have flexible execution windows.
“Grid carbon intensity varies substantially over short time scales, while many AI workloads have flexible execution windows.”
The paper relies on this without support.
Data centers contribute significantly to grid instability and greenhouse gas emissions.
“As data centers become more energy-intensive, they contribute significantly to grid instability and greenhouse gas emissions.”
Citations
The citation lists haven’t been fetched yet
Cited by 1
Metadata
- Authors
- Bhavsar, A., Sun, T., Flora, J., & Zanocco, C.
- Affiliations
- Bellarmine University (US)
- Stanford University (US)
- Corresponding author
- Bhavsar, Adyant
- Publisher
- Wiley
- Published
- 2026-03-02
- OpenAlex
- W7133220690
- Cited by
- 1
- References
- 0
- Citations by year
- 2026: 1
- Type
- posted content
- Language
- en
- Open access
- green
- doi.org
- Topics
- Physical Sciences › Computer Science › Information Systems › Cloud Computing and Resource Management
- Physical Sciences › Computer Science › Hardware and Architecture › Energy Efficiency in Computing
- Physical Sciences › Engineering › Electrical and Electronic Engineering › Smart Grid Energy Management
- Keywords
- carbon-aware scheduling, AI data centers, workload scheduling, carbon intensity forecasting, grid carbon intensity, time-of-use electricity pricing, temperature prediction, CO2 emission reduction, energy cost reduction, renewable energy integration
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
A language model drew the claims from the paper on 2026-10-11, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-11 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Carbon-Aware Scheduling of AI Data Center Workloads Using Environmental and Energy Grid Forecasts. https://app.cochat.ai/papers/10.22541/essoar.177248651.16591616/v1
- The paper
- Bhavsar, A., Sun, T., Flora, J., & Zanocco, C. (2026). Carbon-Aware Scheduling of AI Data Center Workloads Using Environmental and Energy Grid Forecasts. https://doi.org/10.22541/essoar.177248651.16591616/v1
