Between Promise and Reality: Cloud Computing in Meeting Environmental Sustainability Challenges

Cloud Computing and the Environment: The Carbon Footprint Between Ambitions and Challenges

Cloud computing is a backbone of the digital economy, yet it simultaneously poses a growing environmental challenge, especially amid the AI boom that increases demand for computing power. This article, based on the latest academic research, International Energy Agency (IEA) reports, and data from major cloud service providers, examines the quantitative and qualitative dimensions of cloud computing’s climate impact, reviews innovative mitigation strategies from efficiency improvements to AI-based solutions, and provides a precise analysis of the gap between environmental ambitions and operational reality.

First: The True Scale of the Carbon Footprint — Figures That Put the Picture in Context

Media reports raise concerns about “doubling” and “tripling” of data center energy consumption, but placing these figures in their global context reveals a more accurate picture.

IEA data indicate that data centers are currently responsible for only 0.5% of global CO2 emissions from fuel combustion (about 180 million tons out of a total of 35,000 million tons). Even in the accelerated growth scenario, data centers’ share is expected to reach only 1.4% of global combustion emissions by 2030. By contrast, aviation emissions account for more than 2% of global emissions annually.

The real warning, however, lies in the direction of change: data centers are among the few sectors whose emissions are increasing through 2030, alongside road transport and aviation, while most other sectors are seeking to reduce emissions. The impact is also highly concentrated geographically: in the United States, data centers may account for half of electricity demand growth by 2028, and in Ireland they already consume 21% of the country’s electricity.

On the other hand, quantitative analysis shows that global data center consumption will rise from 300 TWh in 2020 to 945 TWh by 2030 in the IEA’s base scenario—equivalent to Japan’s current consumption. But this growth of 530 TWh represents only 8% of the total expected increase in global electricity demand by 2030, compared to 838 TWh for electric vehicles and 651 TWh for air conditioning.

Second: AI — The Fastest Driver of Demand

The Goldman Sachs report describes AI as the most influential driver of growing demand for computing capacity, projecting a 160% rise in global demand for computing power by 2030. In another IEA report, AI is estimated to be responsible for 5–15% of data center consumption currently, with this share expected to rise to 35–50% by 2030.

One root cause of this growth is that GPUs used in training and running AI models consume far more energy than traditional processors and generate higher heat requiring denser cooling systems.

Third: Mitigation Strategies — From Efficiency to Innovation

1. Improving Power Usage Effectiveness (PUE)

Power Usage Effectiveness (PUE) is the most common measure of data center efficiency, measuring the ratio of total energy consumed to the energy used by IT equipment. Google data indicate it achieved a global average PUE of 1.09 in 2025, compared to an industry average of 1.56. This means Google uses 84% less energy in cooling and supporting infrastructure compared to the global average.

2. Advanced Cooling Technologies

A study published in Nature (2025) revealed significant environmental benefits of advanced cooling systems. A life cycle assessment (LCA) conducted by Microsoft showed that liquid cooling technologies (cold plate and immersion cooling) can achieve:

• A 15–21% reduction in greenhouse gas emissions

• A 15–20% reduction in energy demand

• A 31–52% reduction in blue water consumption

This is because traditional air cooling can account for up to 40% of total energy consumption in a data center. With energy density increasing due to AI processors, heat loads are expected to exceed the capacity of current air cooling systems, making liquid cooling an operational necessity rather than merely an environmental choice.

3. Smart Task Scheduling Using AI

Academic research is investing in developing energy-aware cloud platforms. A 2025 IEEE study presented a model combining Long Short-Term Memory (LSTM) networks to predict energy demand and carbon intensity, and Deep Reinforcement Learning (DRL) for carbon-aware task scheduling, achieving:

• A 12.6% reduction in carbon emissions

• A 9.3% improvement in scheduling efficiency

Another study in IEEE Access (2025) developed a cloud framework using real-time carbon intensity data from APIs such as Electricity Maps to schedule energy-intensive tasks during periods of low carbon intensity (i.e., when the electricity grid is cleaner). An architectural study in ACM in 2025 concluded that designing low-carbon cloud servers using components such as energy-efficient processors and reused DRAM memory can achieve a 28% reduction in emissions per processing core.

4. Clean Energy and Hybrid Infrastructure

Cloud service providers are turning to renewable and nuclear energy sources. IEA estimates indicate that the electricity mix for data centers may shift from about 60% fossil fuels and 40% clean energy today to 60% clean energy and 40% fossil fuels by 2035. An IDC study also showed that transitioning to hybrid multicloud infrastructure helped organizations reduce physical server space by up to 80% and energy consumption by 60–80%.

5. The Role of Regulatory Governance

Regulators have begun monitoring the cloud’s environmental footprint. The French regulatory authority Arcep announced in January 2026 the expansion of environmental data collection to include cloud computing service providers, with indicators covering greenhouse gas emissions, energy consumption, and water use. This type of regulatory transparency drives tangible improvements in the sector’s environmental performance.

Fourth: The Great Paradox — Cloud Computing as Part of the Solution

Research reveals a fundamental paradox: despite its rising carbon footprint, cloud computing and AI may contribute to emissions savings far exceeding what they consume. The IEA estimates that current AI applications in the energy, industry, transport, and buildings sectors could achieve emissions reductions of up to 1,400 million tons of CO2 by 2035—equivalent to 3 to 4 times the total emissions of data centers in the accelerated growth scenario.

Among the most promising applications:

• Improving fossil power plant efficiency by more than 2%

• Improving transport efficiency by 5–10% through smart routing and driving optimization

• Improving building energy consumption with smart control systems saving about 10% of energy use

Fifth: Conclusion and Future Outlook

Cloud computing presents a complex model of the relationship between technology and the environment. While its rising energy consumption raises real concerns, the figures show that its overall share of global emissions remains limited, and the potential environmental benefits of AI applications may far outweigh their costs.

Regarding the future outlook, five main pathways can be identified:

1. Continuous efficiency improvements: PUE improvements are expected to continue with the spread of advanced cooling technologies, with Microsoft aiming for a PUE of 1.12.

2. Transition to clean energy: The energy mix for data centers is expected to shift from a fossil-majority to a clean-majority by 2035.

3. Low-carbon server engineering: Designing servers using reused or energy-efficient components can achieve a 28% reduction in emissions per core.

4. Carbon-aware smart scheduling: Using AI to schedule tasks at times and locations with low carbon intensity.

5. Governance and transparency: Growing regulatory pressure for cloud service providers to disclose their environmental footprint.

Sources and References

1. ScienceDirect. (2024). Cloud: sustainability as a driver of cloud technology innovations.

2. IEEE Xplore. (2026). Creating an Energy-Aware Cloud Platform to Optimize Carbon Footprint of Large-Scale Computational Environmental Models.

3. IEEE Access. (2025). A Green Cloud-Based Framework for Energy-Efficient Task Scheduling Using Carbon Intensity Data for Heterogeneous Cloud Servers.

4. Google Data Centers. (2025). 2025 PUE Report.

5. Carbon Brief. (2025). AI: Five charts that put data-centre energy use – and emissions – into context.

6. Nature. (2025). Using life cycle assessment to drive innovation for sustainable cool clouds.

7. Arcep (French Regulatory Authority). (2026). Environment – Data collection expansion to cloud computing service providers.

8. arXiv. (2025). A User-centric Kubernetes-based Architecture for Green Cloud Computing.

9. International Energy Agency (IEA). (2025). Energy and AI – AI and climate change.

10. Elsevier / ScienceDirect. (2025). Energy efficiency sustainability framework for cloud and quantum data centres.

11. Nutanix / IDC. (2025). Hybrid Multicloud Deployment Choices Can Increase ROI Via Sustainability Benefits.

12. ACM / ISCA. (2025). Designing Cloud Servers for Lower Carbon.