The AI Hunger Games: How Data Centers Are Breaking the Energy Market
The world's appetite for artificial intelligence and cloud computing is coming at a steep price — and it's being paid at the power plant.
Natural gas power plant construction costs have surged 66% in just two years, according to new data, nearly doubling the price tag on one of the world's most relied-upon sources of electricity generation. On top of that, projects are now taking 23% longer to complete, stretching timelines that were already straining under growing demand.
The culprit? Data centers — the vast, humming warehouses of servers that power everything from your Netflix queue to the AI chatbots reshaping how we work.
An Electricity Hunger Unlike Anything We've Seen
For decades, electricity demand in developed economies was relatively flat. Efficiency gains in appliances, lighting, and industry largely offset population and economic growth. Grid planners could forecast needs with reasonable confidence.
That model is now broken.
The explosion of generative AI — tools like ChatGPT, Google Gemini, and the thousands of enterprise applications being built on top of them — requires enormous amounts of computing power, which in turn requires enormous amounts of electricity. Training a single large AI model can consume as much energy as hundreds of transatlantic flights. And inference (actually running AI models at scale, billions of times a day) is proving even more power-hungry than many analysts initially predicted.
Major hyperscalers like Microsoft, Google, Amazon, and Meta are racing to build new data center campuses across North America, Europe, and Asia. Each facility can draw anywhere from 100 to 500 megawatts — enough to power tens of thousands of homes.
Why Natural Gas? And Why Is It Getting So Expensive?
Natural gas remains the go-to for new dispatchable power generation in many markets because it can ramp up and down quickly to match demand — unlike solar or wind, which are intermittent. As data centers demand reliable, always-on power, utilities and developers have been rushing to permit and build new gas-fired plants.
But that rush is running into the same supply chain constraints that have plagued construction industries globally since the pandemic. Specialized equipment like gas turbines is on tight allocation. Skilled labour — welders, electricians, pipefitters — is in short supply. Permitting and interconnection queues at grid operators have ballooned.
The result: costs are spiking and timelines are slipping. A project that might have been budgeted at $800 million two years ago could now run $1.3 billion or more.
What This Means for Energy Prices and the Climate
Higher construction costs inevitably mean higher electricity prices — costs that get passed along to consumers and businesses. It also raises questions about whether the buildout can happen fast enough to meet demand without causing reliability issues on existing grids.
From a climate perspective, the trend is troubling. Many jurisdictions had been on track to reduce reliance on fossil fuels for power generation. The data center boom is threatening to reverse those gains, locking in decades of new natural gas infrastructure at a moment when the world needs to be moving in the opposite direction.
Some tech companies are investing heavily in nuclear power — both existing plants and next-generation designs — as a longer-term clean alternative. But those projects face their own cost overruns and multi-decade lead times.
For now, the gap between AI's energy appetite and the grid's ability to feed it is only growing wider.
Source: TechCrunch
