Training GPT-4 consumed approximately 50 gigawatt-hours of electricity — roughly equivalent to what 120 average American homes use in an entire year. Every single ChatGPT query uses about ten times the energy of a standard Google search. And OpenAI processes over 100 million queries per day.

Do the arithmetic. Then consider that GPT-4 is already two generations behind the current frontier. Each new generation requires substantially more compute. More compute means more electricity. More electricity means more infrastructure, more cooling, more water, more environmental impact.

This isn't a technology story. It's an energy story. And it's about to become your business story, whether you're ready or not.

The Invisible Infrastructure

Every time you save a file to OneDrive, send an email through Exchange, run a query in your CRM, or ask an AI assistant for help — you're consuming resources in a data center. That data center needs electricity to run. It needs more electricity to cool itself. It needs water for its cooling systems. It sits on land that was previously something else.

For most of computing history, this was a rounding error. The energy consumed by your emails was negligible. Cloud storage was efficient enough that nobody needed to think about it.

AI changed the equation. Fundamentally.

The International Energy Agency projects that global data center electricity consumption will more than double between 2022 and 2026, from approximately 460 TWh to over 1,000 TWh. To put that in perspective, 1,000 TWh is roughly equal to the entire electricity consumption of Japan. And the primary driver of this explosion is AI workloads.

The New Arms Race

Tech giants are responding to AI's energy demands with infrastructure investments that would have been inconceivable a decade ago:

Microsoft signed a deal with Constellation Energy to restart the Three Mile Island nuclear reactor — yes, that Three Mile Island — specifically to power its AI data centers. Cost: $1.6 billion.

Amazon purchased a nuclear-powered data center campus in Pennsylvania for $650 million and is investing in small modular nuclear reactors for future facilities.

Google signed the first-ever corporate agreement to purchase electricity from small modular nuclear reactors that don't yet exist, planned for completion by 2030.

Meta is building a 2-gigawatt data center — a facility that would consume more electricity than many European cities. They explored acquiring a nuclear power plant before the deal fell through over environmental concerns.

Read that list again. Technology companies are buying nuclear power plants. They are building facilities that consume as much electricity as cities. They are signing contracts for energy technologies that haven't been built yet.

This is not normal corporate expansion. This is the emergence of technology companies as major players in global energy markets. And the implications cascade in every direction.

The Paradox: AI for Sustainability, Powered by Fossil Fuels

The technology industry promotes AI as a tool for sustainability. AI can optimise energy grids, predict weather patterns for renewable energy planning, reduce waste in manufacturing, and improve logistics efficiency.

All true. And all rendered somewhat ironic by the fact that AI's own carbon footprint is growing faster than its ability to reduce everyone else's.

A single generative AI interaction produces approximately 4.32 grams of CO2 — seemingly trivial until you multiply by billions of queries per day. Google's total emissions increased by 48% between 2019 and 2023, driven primarily by data center energy consumption. Microsoft reported a 29% increase in carbon emissions in 2024, despite its stated goal of being carbon-negative by 2030.

These are the companies with the resources and motivation to invest in renewable energy. They're still going backwards. The smaller cloud providers that your business likely uses are doing worse.

Water: The Hidden Cost

Data centers don't just consume electricity. They consume water — enormous quantities of it for cooling.

Microsoft disclosed that its global water consumption increased by 34% in 2022, to nearly 6.4 billion litres. Google consumed 5.6 billion litres in the same year. Much of this increase was attributed to AI workloads.

In a world where water scarcity is an escalating crisis, the fact that answering your question about a recipe or summarising a quarterly report requires cooling infrastructure that drinks millions of litres of water per day is worth contemplating.

What This Means for Your Business

If you're thinking "interesting, but this is a problem for tech giants, not for me" — you're wrong. Here's why:

Cloud costs will increase, not decrease. The era of steadily declining cloud computing prices is likely over. Energy costs are rising. Data center construction is expensive. Water and cooling systems add overhead. These costs will be passed to you. If your IT budget assumes flat or declining cloud costs, revise your assumptions.

The EU Corporate Sustainability Reporting Directive (CSRD). As of 2024, large EU companies are required to report on their environmental impact, including Scope 3 emissions — which means the emissions generated by their supply chain, including their cloud service providers. If your company falls under CSRD reporting requirements (and the threshold is lowering), you will need to account for the carbon footprint of your IT infrastructure.

The EU Energy Efficiency Directive. Data centers in the EU with capacity over 500 kW are now required to report energy performance data. This doesn't directly affect most businesses, but it creates transparency that will influence procurement decisions. Your shareholders, customers, and regulators will increasingly ask: how efficient is your technology stack?

Green procurement pressure. Government contracts, large corporate supply chains, and ESG-conscious investors are increasingly requiring sustainability disclosures from their suppliers. Your IT choices — which cloud provider, which data center region, which services — are becoming sustainability decisions.

The Greek Context

Greece faces a particular set of challenges and opportunities in this landscape.

Electricity costs. Greek businesses already pay among the highest electricity rates in Europe. Any upward pressure on energy prices from data center competition hits harder here. If your business runs on-premise servers, your energy costs are directly exposed. If you use cloud services, the costs are indirect but real.

The sustainability reporting wave. Greek businesses, particularly those in regulated industries or with European clients, are encountering sustainability reporting requirements with increasing frequency. The ability to report on your IT infrastructure's environmental impact isn't optional — it's becoming a condition of doing business with large enterprises.

The opportunity. Greece has abundant solar energy, significant wind resources, and a strategic geographic position between Europe, the Middle East, and Africa. The country is already seeing investment in submarine cable infrastructure. With the right policy framework, Greece could position itself as a destination for sustainable data centers — turning an energy challenge into an economic opportunity.

But that's a national conversation. For your business today, the question is more immediate: do you know the carbon footprint of your IT operations? Can you report it? Are you ready to be asked?

What You Can Do Now

The transition to energy-aware IT isn't optional, and it isn't coming someday. It's in progress. Here's what practical steps look like:

Audit your cloud footprint. Major cloud providers (Azure, AWS, Google Cloud) now offer carbon footprint dashboards. Know your numbers before someone asks for them.

Choose your regions deliberately. Data center location matters. Nordic regions run on largely renewable energy. Other regions don't. Where your data is processed has environmental implications.

Right-size your infrastructure. Over-provisioned servers, unused virtual machines, and abandoned storage volumes waste energy. Regular infrastructure audits save money and reduce environmental impact simultaneously.

Evaluate AI usage critically. Not every task needs AI. A database query that takes 0.01 seconds and minimal energy shouldn't be replaced by an AI interaction that takes 10 seconds and 100x the energy. Use AI where it adds value, not where it adds complexity and cost.

Prepare for reporting. Even if CSRD doesn't apply to you today, the direction is clear. Start collecting the data now, while it's voluntary and you can learn without regulatory pressure.

Previously in the series: Deepfakes, AI Phishing, and the Perfect Scam: The Nightmare That's Here to Stay

Need help understanding your IT infrastructure's environmental impact and preparing for sustainability reporting? Let's talk about practical steps that satisfy regulators without paralysing operations.