Cloud Computing Carbon Footprint: AWS vs Azure vs Google Cloud Emissions Compared

Data centres consumed roughly 415 TWh of electricity in 2024 — about 1.5% of global electricity demand, according to the IEA's 2025 Energy and AI report. For digital businesses, cloud infrastructure is often the single largest source of Scope 3 emissions. Choosing between AWS, Azure, and Google Cloud is not just a technical or financial decision anymore. It is an environmental one.

The problem: all three providers claim to be sustainable. All three publish carbon data. But the methodologies, transparency levels, and actual performance differ substantially. This article cuts through the marketing to compare what we actually know about each provider’s emissions.

The Current State: Carbon Reporting by Provider

Each hyperscaler offers a carbon reporting dashboard for customers. The quality and methodology vary significantly:

The most important distinction is between market-based and location-based accounting. Market-based accounting allows providers to claim zero emissions in regions where they have purchased Renewable Energy Certificates (RECs), even if the grid still runs partially on fossil fuels. Location-based accounting uses the actual grid carbon intensity where the data center operates. The difference can be enormous. For a detailed breakdown of Scope 1, 2, and 3 distinctions, see our complete emissions guide, and for how cloud Scope 2/3 data feeds into statutory disclosure, see our CSRD digital sustainability reporting guide.

AWS: The Scale Leader With Transparency Gaps

Amazon Web Services runs 39 Regions and 124 Availability Zones, and reports an average global PUE of 1.14 for 2025 (1.15 in 2024). AWS committed to matching 100% of its electricity consumption with renewable energy by 2030, and says it hit that goal in 2023, seven years early — alongside a separate commitment to net-zero carbon by 2040 through The Climate Pledge.

What AWS Reports

The AWS Customer Carbon Footprint Tool (CCFT) provides market-based emissions estimates per service, per region. Since an October 2025 methodology update, it also reports location-based emissions and adds Scope 3 estimates covering embodied carbon from server manufacturing. AWS applies renewable energy credits at the regional level, meaning regions with high REC coverage show dramatically lower market-based emissions.

Where AWS Falls Short

• Dual reporting is brand new — AWS only added location-based Scope 2 and Scope 3 data to the CCFT in an October 2025 update, years after Azure and Google Cloud. Historical data before that is market-based only.
• REC quality questions — market-based figures rely on renewable energy certificates and PPAs, which the GHG Protocol Scope 2 Guidance treats as lower quality than on-site generation. AWS does not disclose its certificate mix by region.
• Limited Scope 3 coverage — embodied carbon data is partial and methodology is not fully published.
• Water usage data — AWS publishes aggregate water efficiency metrics but does not provide customer-level water usage data, which is increasingly relevant for CSRD reporting.

Best and Worst AWS Regions for Carbon

AWS regions powered primarily by hydroelectric or nuclear energy have the lowest carbon intensity: eu-north-1 (Stockholm), ca-central-1 (Montreal), and eu-west-1 (Ireland) are among the cleanest. High-carbon regions include those in Asia-Pacific where grid intensity is higher: ap-southeast-1 (Singapore) and ap-south-1 (Mumbai).

Microsoft Azure: Aggressive Commitments, Subscription-Level Reporting

Azure operates across 80+ regions globally. Microsoft has committed to being carbon negative by 2030 and removing all its historical emissions since 1975 by 2050 — the most ambitious climate target among the three hyperscalers.

What Azure Reports

The Azure Emissions Impact Dashboard provides both market-based and location-based emissions. This dual reporting is valuable for companies subject to CSRD, which requires both methodologies. Azure breaks emissions down by subscription, making it easier to allocate carbon costs to specific business units or products.

Where Azure Excels

• Subscription-level dual reporting — the Emissions Impact Dashboard exposes market-based and location-based Scope 2 data by subscription, still the most convenient breakdown for allocating costs, though AWS and Google Cloud have since added the same dual reporting.
• Climate technology investments — Microsoft's Climate Innovation Fund pledged $1 billion in 2020 for climate tech broadly (not removal alone); ~$793 million deployed as of FY2024.
• Internal carbon fee — in place since 2012; $15/tonne CO2e since 2019, with a $100/tonne rate on business air travel since 2022.

Where Azure Falls Short

• Scope 3 growth from AI — Microsoft's 2025 Environmental Sustainability Report (FY2024 data) shows total emissions up 23.4% and Scope 3 up 26% versus its FY2020 baseline — driven largely by data center construction for AI workloads — even as Scope 1+2 emissions fell 29.9% over the same period. Scope 3 is moving in the right direction, but it is still well above the 2020 baseline, and that is the distance Microsoft has to close before 2030.
• Data center construction emissions — the embodied carbon of new data center construction (concrete, steel, equipment) is massive and growing as AI demand drives expansion.
• Backup generator capacity — like other hyperscalers, Azure data centers rely on diesel backup generators for resiliency. Operational hours are low, but the capacity exists and emissions occur during outages and testing.

Google Cloud: 24/7 Carbon-Free Energy Pioneer

Google Cloud operates across 43 regions globally. Google claimed operational carbon neutrality from 2007 through offsets, but said in its 2024 Environmental Report that it stopped maintaining that claim starting in 2023. Its current target is to run on 24/7 carbon-free energy (CFE) everywhere by 2030 — a fundamentally different approach from annual REC matching.

What Makes 24/7 CFE Different

Most renewable energy claims work on an annual matching basis: a company buys enough RECs or generates enough renewable energy over a year to match its total consumption. The problem is temporal: a data center might run on coal-generated electricity at 2 AM and “match” it with solar RECs generated at 2 PM. The grid impact is zero.

Google’s 24/7 CFE approach tracks carbon-free energy on an hourly basis, region by region. A region scoring 90% CFE means that, hour by hour across the year, about 90% of the electricity it consumed was matched with carbon-free generation on the same grid — not that enough certificates exist somewhere to cover annual consumption.

Current 24/7 CFE Scores by Region

Google publishes CFE scores for every region, calculated on an hourly basis. Top performers (2025 data):

• europe-north1 (Finland) — 98% CFE, 30 gCO2eq/kWh (hydro + nuclear + wind)
• northamerica-northeast1 (Montréal) — 97% CFE, 11 gCO2eq/kWh (hydro)
• us-central1 (Iowa) — 88% CFE, 432 gCO2eq/kWh (wind PPAs)
• europe-west4 (Eemshaven, Netherlands) — 82% CFE, 194 gCO2eq/kWh
• europe-west1 (Belgium) — 79% CFE, 126 gCO2eq/kWh

Lower performers:

• asia-east1 (Taiwan) — 15% CFE, 428 gCO2eq/kWh
• asia-south1 (Mumbai) — 20% CFE, 598 gCO2eq/kWh
• asia-southeast1 (Singapore) — 5% CFE, 366 gCO2eq/kWh

These scores are published transparently, which is itself a significant differentiator. For businesses tracking their website carbon footprint, Google Cloud’s regional CFE data enables accurate Scope 3 calculations.

Where Google Cloud Falls Short

• Legacy “carbon neutral” claim — Google’s original “carbon neutral since 2007” claim relied on offsets; the company stopped maintaining it starting in 2023 and now leads with 24/7 CFE metrics instead.
• AI energy demand — like Microsoft, Google’s AI expansion (Gemini training, inference infrastructure) is driving significant energy demand growth that risks outpacing CFE procurement.
• Smaller region footprint than Azure — 43 regions versus Azure’s 80+, which may force some customers to choose higher-carbon regions for latency reasons.

Head-to-Head: Which Provider Is Greenest?

The honest answer: it depends on your specific usage pattern.

For CSRD-reporting companies, all three now expose market- and location-based Scope 2 data; Azure’s per-subscription breakdown is the most convenient to slice by business unit. For grid impact, Google Cloud’s 24/7 CFE approach is more rigorous than annual REC matching. For cost-driven teams, a low-carbon region is still the single highest-leverage choice — grid intensity ranges from 11 gCO2eq/kWh in Montréal to 366-598 g in Singapore and Mumbai.

Practical Steps: Reducing Your Cloud Carbon Footprint

Regardless of provider, these actions reduce cloud emissions:

1. Region Selection

Move workloads to low-carbon regions where latency permits. Google's own region-carbon data shows grid intensity ranging from 11 gCO2eq/kWh in Montréal to 366-598 gCO2eq/kWh in Singapore and Mumbai — tens of times higher. This is the single highest-impact decision.

If your web infrastructure runs on traditional hosting rather than hyperscaler cloud, the same logic applies: choosing a provider with verified renewable energy is the highest-impact sustainability decision. Our 2026 guide to sustainable web hosting covers PUE, Green Web Foundation verification, and the key differences between genuine renewables and unbundled RECs.

2. Right-Sizing and Utilization

Over-provisioned instances waste energy. Use autoscaling, spot instances, and reserved capacity that matches actual demand. Research on data center energy proportionality found that idle servers still draw 50-70% of their peak power, and most production servers run at just 10-50% utilization — meaning a large share of provisioned capacity goes to waste.

The same principle applies at the browser layer. JavaScript can consume 3–5× more energy than equivalent CSS operations — reducing unused scripts cuts both page weight and Scope 3 emissions from user devices.

3. Carbon-Aware Scheduling

Shift flexible workloads (batch processing, ML training, backups) to times when the grid is cleanest. Google’s Carbon-Aware Computing and emerging tools for AWS and Azure enable this. Our guide to reducing website carbon footprint covers additional optimization techniques.

4. Measure and Report

Enable your provider’s carbon dashboard. Export the data monthly. Include it in your Scope 3 inventory. What gets measured gets managed. For a methodology overview, see our measurement guide. For a step-by-step comparison of the best measurement tools, see our complete guide to measuring website carbon footprint. To display your website’s carbon footprint in real time, the Carbon Badge widget gives visitors instant sustainability visibility. For industry benchmarks, the website carbon footprint benchmarks by industry 2026 covers CO2 per pageview across 12 sectors, from media to SaaS.

5. Challenge Your Provider

As a customer, push AWS and Azure to publish hourly, grid-based carbon data the way Google does for its regions — not just monthly market-based aggregates. Ask all three for their underlying Scope 3 methodology. Customer pressure is part of how Google’s 24/7 CFE reporting came to exist in the first place.

The AI Elephant in the Data Center

All three providers are experiencing massive demand growth from AI workloads. Training a large language model can emit hundreds of tonnes of CO2 — Strubell et al. (2019) estimated ~284 tonnes for a large NAS Transformer, and Patterson et al. (2021) put GPT-3 at ~552 tonnes CO2e. Inference has a lower per-query footprint but scales with millions of daily queries.

The uncomfortable truth: cloud computing emissions are likely to increase in absolute terms over the next 3-5 years, even as efficiency improves per unit of compute. AI demand risks outpacing renewable energy procurement. Microsoft’s 26% Scope 3 increase in its 2025 report (FY2024 data) is a preview, not an anomaly.

For businesses choosing a cloud provider with sustainability in mind, the question is not “who is green” but “who is most transparent about the problem and most credible in addressing it.” On that metric, Google Cloud leads in transparency, Microsoft leads in financial commitment to climate technology investment (including carbon removal), and AWS leads in absolute renewable energy procurement. None of them have solved the AI energy problem yet. For a detailed look at how this AI energy surge affects website carbon calculations, see our AI data center energy and website carbon guide.