In partnership with
Artificial intelligence is becoming more efficient per task while the computing systems behind it are using more electricity. The International Energy Agency reports that data center electricity demand grew 17 percent worldwide in 2025. Electricity use at AI-focused data centers grew 50 percent that year. Those figures describe related but different parts of the infrastructure.
Turn AI into Your Income Engine
Ready to transform artificial intelligence from a buzzword into your personal revenue generator
HubSpot’s groundbreaking guide "200+ AI-Powered Income Ideas" is your gateway to financial innovation in the digital age.
Inside you'll discover:
A curated collection of 200+ profitable opportunities spanning content creation, e-commerce, gaming, and emerging digital markets—each vetted for real-world potential
Step-by-step implementation guides designed for beginners, making AI accessible regardless of your technical background
Cutting-edge strategies aligned with current market trends, ensuring your ventures stay ahead of the curve
Download your guide today and unlock a future where artificial intelligence powers your success. Your next income stream is waiting.
The distinction matters because data centers run many services besides AI. Their total electricity includes cloud storage, websites, business software, streaming, and other computing. AI systems add demand within that broader base. The IEA estimates that data centers used 485 terawatt-hours in 2025. Its central projection reaches 950 terawatt-hours in 2030, or around three percent of global electricity use. AI-focused centers are projected to grow faster and triple their electricity use across that period.
These are estimates and a forecast, not a meter reading for future years. The IEA says bottlenecks across chips, power equipment, and grid connections constrain near-term growth. Its outlook could change as projects advance, financing shifts, or new energy-intensive AI uses spread. A forecast describes a scenario built from evidence and assumptions, not an assured future outcome.
Per-task efficiency and total demand can move in different directions. The IEA reports that efficiency gains have cut electricity use per AI task by at least tenfold annually in recent years. Simple text queries now typically use less electricity than a television running for the same period, according to its analysis. This does not mean every AI request has the same small footprint.
Most AI Tools Weren't Built for This
B2B customer issues move across teams and systems, not through a single chat window. A new Harvard Business Review Analytic Services briefing paper, sponsored by Front, breaks down where AI tools fall short in B2B service and what to ask before you invest.
Video generation, reasoning, and agentic tasks can require hundreds or thousands of times more energy per query than simple text generation, the agency says. Task type, model capability, output format, hardware, and use volume all influence the total. A number for one prompt therefore cannot stand in for the electricity consumed by an entire service or data center.
The physical footprint also varies by location. A 2025 Lawrence Berkeley National Laboratory review found workload-level data center water use could vary by more than 10,000 times. It linked that spread to server efficiency, the water intensity of electricity generation, utilization, cooling systems, climate, idle equipment, and replacement cycles. This study examined data center workloads generally. It does not establish a single water cost for an AI prompt.
A 2026 United Nations University report also treats carbon, water, and land as separate dimensions of AI’s energy footprint. It argues that electricity sources and infrastructure locations shape these effects. A lower-carbon supply does not automatically mean lower water or land use. The report’s framing widens the accounting question, but it does not make every workload’s local impact identical.
Measurement remains uneven. The IEA says more systematic energy disclosures would improve projections, while monitoring and frequent updates remain important. Public comparisons should identify the task, system boundary, and measurement method. A training estimate cannot automatically represent later inference. Nor can a short text task represent image or video generation. These boundaries matter when figures from different reports appear to describe the same thing.
For readers, the key distinction is between efficiency at one task and growth across millions of tasks. Both affect the system, alongside grid capacity and where infrastructure is built. This daily note is a limited view of an evolving evidence base; later disclosures and measurement may sharpen the estimates without making them universal.
Your AI prompts are leaving out 80% of what you're thinking.
When you type a prompt, you summarize. When you speak one, you explain. Wispr Flow captures your full reasoning (constraints, edge cases, examples, tone) and turns it into clean, structured text you paste into ChatGPT, Claude, or any AI tool. The difference shows up immediately. More context in, fewer follow-ups out.
Used by teams at Microsoft, Amazon, and Notion. Try Wispr Flow free on Mac, Windows, iPhone, and Android.
