Myth-busting Green IT & sustainable data centers: separating hype from reality

Data centers consumed an estimated 460 TWh of electricity globally in 2025, roughly 2% of total global electricity demand, and this figure is projected to exceed 1,000 TWh by 2030 under current growth trajectories. The industry has responded with ambitious sustainability claims: 100% renewable energy commitments, carbon-neutral pledges, and water-free cooling promises. But the gap between marketing narratives and operational reality remains substantial, and policy makers, compliance officers, and procurement teams must separate genuine progress from greenwashing to make informed decisions.

Why It Matters

The regulatory landscape for data center sustainability is tightening rapidly. The European Union's Energy Efficiency Directive (EED), effective since September 2024, requires data centers with IT capacity above 500 kW to report energy performance metrics including Power Usage Effectiveness (PUE), water usage, renewable energy procurement, and waste heat recovery. The EU's Corporate Sustainability Reporting Directive (CSRD) mandates that companies using significant data center capacity disclose scope 2 and scope 3 emissions associated with digital infrastructure.

In the United States, the SEC's climate disclosure rules require reporting on material energy consumption and emissions, which increasingly includes cloud computing and data center operations for technology-intensive companies. California's SB 253 captures data center operators with revenues exceeding $1 billion, requiring comprehensive greenhouse gas reporting across all scopes.

Beyond regulation, financial materiality is driving scrutiny. Data center electricity costs represent 30 to 40% of total operating expenses. Water consumption for cooling has become a flashpoint in drought-prone regions, with local communities and regulators increasingly challenging new facility permits. Understanding what sustainability claims actually mean, and where they fall short, has become essential for any organization making purchasing, investment, or policy decisions involving digital infrastructure.

Key Concepts

Power Usage Effectiveness (PUE) measures the ratio of total facility energy to IT equipment energy. A PUE of 1.0 would mean all energy goes to computing; real-world values range from 1.10 for hyperscale facilities to 2.0 or higher for older enterprise data centers. The global average PUE in 2025 was approximately 1.58, according to the Uptime Institute, though hyperscalers report averages of 1.10 to 1.20.

Renewable Energy Certificates (RECs) and Power Purchase Agreements (PPAs) represent fundamentally different approaches to renewable energy procurement. RECs are tradeable certificates that represent the environmental attributes of one megawatt-hour of renewable generation, purchasable separately from actual electricity. PPAs contract directly for renewable generation, often from specific facilities. The distinction matters enormously for actual emissions impact.

Water Usage Effectiveness (WUE) measures liters of water consumed per kilowatt-hour of IT energy. Evaporative cooling systems achieve excellent PUE values but consume 1.8 to 7.6 liters per kWh, creating significant water stress in arid regions. Air-cooled and liquid-cooled systems use minimal water but may consume more energy.

Carbon-Free Energy (CFE) Matching tracks whether renewable energy is consumed at the same time and in the same grid region as data center electricity demand, rather than on an annual net basis. Google pioneered hourly CFE matching and reported 64% average 24/7 carbon-free energy across its global fleet in 2023, revealing that annual 100% renewable claims significantly overstate actual carbon impact.

Myths vs. Reality

Myth 1: "100% renewable energy" means data centers run entirely on clean power

Reality: The vast majority of "100% renewable" claims rely on annual energy matching using unbundled RECs, meaning the data center purchases certificates equivalent to its annual consumption but continues drawing electricity from the local grid, which may be predominantly fossil-fueled. A data center in Virginia purchasing wind RECs from Texas has zero impact on the coal and gas generation actually powering its servers.

Google's 24/7 CFE initiative exposed this gap. Despite purchasing 100% renewable energy on an annual basis since 2017, Google's actual hourly carbon-free energy percentage ranged from 21% (Singapore) to 97% (certain Nordic locations) in 2023. The 100% annual matching claim obscured the fact that the company's data centers were still drawing substantial fossil-generated electricity during nights, cloudy periods, and low-wind hours.

Microsoft's 2025 sustainability report similarly acknowledged that its Scope 2 emissions increased 30% year-over-year despite maintaining 100% renewable energy matching, driven by the massive electricity demands of AI training workloads. Annual matching with RECs does not equate to actual decarbonization.

Myth 2: Hyperscale data centers are inherently more sustainable than enterprise facilities

Reality: Hyperscale facilities achieve superior PUE (1.10 to 1.20 versus 1.50 to 2.00 for enterprise data centers) and benefit from greater purchasing power for renewable energy. However, this efficiency advantage obscures the absolute scale of their environmental impact. A hyperscale campus consuming 500 MW at PUE 1.15 still uses more total energy than 100 enterprise data centers consuming 2 MW each at PUE 1.80. The aggregate environmental footprint of hyperscale expansion, including embodied carbon from construction, supply chain emissions from server manufacturing, and water consumption, often exceeds what efficiency metrics suggest.

The IEA reported that global data center energy consumption grew by approximately 70% between 2022 and 2025, driven almost entirely by hyperscale expansion for AI workloads. Efficiency improvements have been real but insufficient to offset the explosive growth in computing demand.

Myth 3: Liquid cooling eliminates data center water consumption

Reality: Direct liquid cooling (DLC) to the chip reduces or eliminates the need for evaporative cooling, significantly lowering on-site water consumption. However, the electricity required to reject heat through dry coolers or cooling towers still exists, and the water footprint shifts upstream to electricity generation. Thermal power plants (coal, gas, and nuclear) consume 1.5 to 4.0 liters of water per kWh generated. A data center using zero on-site water but powered by a gas-fired grid may consume more total water per kWh of IT load than an evaporative-cooled facility powered by solar.

Microsoft's 2024 environmental report disclosed that its global water consumption increased 34% year-over-year to 6.4 billion liters, primarily driven by data center expansion. While newer facilities use more efficient cooling, the sheer growth in capacity overwhelmed per-unit improvements.

Myth 4: PUE improvements mean the industry is becoming more sustainable

Reality: PUE measures the overhead efficiency of cooling and power distribution but says nothing about the total energy consumed, the carbon intensity of that energy, or the water, land, and materials required. A facility can achieve PUE 1.05 while consuming 200 MW of coal-fired electricity. Furthermore, PUE improvements face diminishing returns. Moving from PUE 2.0 to 1.5 saves 25% of total energy; moving from 1.2 to 1.1 saves only 8%. The industry has largely exhausted easy PUE gains, and continued focus on this single metric distracts from more impactful sustainability dimensions.

The Uptime Institute's 2025 survey found that average global PUE improved by only 0.02 points from 2023 to 2025, compared to improvements of 0.10 or more per year a decade ago.

Myth 5: Carbon offsets make data centers carbon neutral

Reality: Several major data center operators and cloud providers claim carbon neutrality through combinations of renewable energy purchases and carbon offset credits. A 2024 analysis by Carbon Market Watch found that 68% of the offsets used by technology companies came from forestry and avoided-deforestation projects with permanence and additionality concerns. When accounting methods exclude embodied carbon from server manufacturing (representing 30 to 50% of lifecycle emissions), Scope 3 supply chain emissions, and upstream fuel extraction for grid electricity, the "carbon neutral" label becomes misleading.

Amazon Web Services disclosed in 2025 that its total carbon footprint increased 48% since 2019 despite purchasing renewable energy equivalent to 100% of its electricity consumption, because Scope 3 emissions from hardware manufacturing, logistics, and customer use of services were growing faster than operational efficiency gains.

Key Players

Google Cloud leads in transparency with its 24/7 carbon-free energy initiative, publishing hourly CFE percentages by region and committing to 100% 24/7 CFE by 2030.

Equinix operates 260+ data centers globally and has committed to climate neutrality by 2030, with published PUE data and science-based targets validated by SBTi.

Microsoft Azure reported 24/7 CFE matching goals alongside its broader 2030 carbon-negative commitment, though disclosed emissions increases have prompted scrutiny.

Digital Realty published industry-leading WUE data and invested in direct liquid cooling infrastructure for AI workloads, targeting water-positive operations by 2030.

Schneider Electric provides EcoStruxure IT infrastructure management that enables real-time sustainability monitoring across colocation and enterprise data centers.

Action Checklist

• Require cloud and colocation providers to disclose hourly carbon-free energy percentages, not just annual renewable matching claims
• Demand WUE reporting alongside PUE, including upstream water consumption from electricity generation
• Evaluate embodied carbon data for IT hardware procurement, including server manufacturing and logistics emissions
• Verify carbon offset quality by reviewing project types, vintages, and third-party validation reports
• Align data center sustainability requirements with CSRD, EED, and SEC disclosure obligations in procurement contracts
• Request Scope 3 emissions reporting from cloud providers covering hardware manufacturing and supply chain impacts
• Benchmark provider sustainability claims against publicly available data from CDP, sustainability reports, and regulatory filings
• Include sustainability performance clauses in service-level agreements with measurable, auditable metrics

FAQ

Q: How should procurement teams evaluate competing "100% renewable" claims from cloud providers? A: Ask for hourly or sub-annual carbon-free energy data broken down by region. Providers using 24/7 CFE matching with local, time-matched renewable generation deliver genuinely lower emissions than those relying on annual REC purchases from distant projects. Also request disclosure of whether renewable claims cover only Scope 2 electricity or also address Scope 1 on-site generation and Scope 3 supply chain emissions.

Q: What metrics beyond PUE should data center sustainability assessments include? A: A comprehensive assessment should track WUE (liters per kWh), Carbon Usage Effectiveness (CUE, kilograms CO2 per kWh), embodied carbon per rack or per kW of IT capacity, e-waste recycling rates, and grid carbon intensity during actual operating hours. The EU's EED reporting framework provides a useful baseline set of metrics.

Q: Are there credible third-party certifications for data center sustainability? A: The most rigorous certifications include the EU Code of Conduct for Energy Efficiency in Data Centres, LEED and BREEAM for facility design, and ISO 50001 for energy management systems. Science-Based Targets initiative (SBTi) validation provides credibility for emissions reduction commitments. Proprietary "green" labels from industry associations should be evaluated against these established frameworks.

Q: How does AI workload growth affect data center sustainability claims? A: AI training and inference workloads consume 3 to 10 times more energy per compute unit than traditional cloud workloads. The rapid expansion of GPU-dense infrastructure is driving both total energy consumption and cooling intensity higher, challenging existing sustainability commitments. Organizations using AI services should request workload-specific carbon intensity data from providers rather than relying on facility-average metrics.

Sources

• International Energy Agency. (2025). Data Centres and Data Transmission Networks: Tracking Report. Paris: IEA Publications.
• Uptime Institute. (2025). Global Data Center Survey: Energy Efficiency and Sustainability Metrics. New York: Uptime Institute.
• Google. (2024). 24/7 Carbon-Free Energy: 2023 Progress and Methodology Report. Mountain View, CA: Google LLC.
• Microsoft. (2025). Environmental Sustainability Report 2024. Redmond, WA: Microsoft Corporation.
• Carbon Market Watch. (2024). Corporate Climate Claims in the Technology Sector: An Assessment of Offset Quality and Accounting Practices. Brussels: Carbon Market Watch.
• European Commission. (2024). Energy Efficiency Directive: Data Centre Reporting Requirements Implementation Guidance. Brussels: European Commission.
• Amazon Web Services. (2025). Amazon Sustainability Report 2024. Seattle, WA: Amazon.com Inc.