<!-- mobian-agent-page publisher="dailydev" canonical="https://daily.dev/posts/why-ai-data-centers-are-running-out-of-power-boxvuei0b" -->

---
title: Why AI Data Centers Are Running Out of Power | daily.dev
description: A video breaks down why AI data centers are straining global power grids, citing that data centers now consume around 565 terawatt hours annually (about 2% of...
canonical: https://daily.dev/posts/why-ai-data-centers-are-running-out-of-power-boxvuei0b
twitter:card: summary_large_image
twitter:site: @dailydotdev
og:type: website
og:site_name: daily.dev
og:title: Why AI Data Centers Are Running Out of Power | daily.dev
og:description: A video breaks down why AI data centers are straining global power grids, citing that data centers now consume around 565 terawatt hours annually (about 2% of...
og:url: https://daily.dev/posts/why-ai-data-centers-are-running-out-of-power-boxvuei0b
og:image: https://api.daily.dev/og/posts/boXVUeI0B.png
og:image:alt: Why AI Data Centers Are Running Out of Power
og:image:width: 1200
og:image:height: 630
og:locale: en
---

> ## Documentation Index
> Fetch the complete documentation index at: https://daily.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Why AI Data Centers Are Running Out of Power

**[Tiff In Tech](https://daily.dev/sources/tiffintech)** · 13 min read · 0 upvotes · 0 comments

## Summary

A video breaks down why AI data centers are straining global power grids, citing that data centers now consume around 565 terawatt hours annually (about 2% of world electricity, up 12% year over year), on par with global aviation. It explains why AI chips like Nvidia's H100 and Blackwell draw far more power than traditional CPUs due to massively parallel core designs, and how liquid cooling is replacing air cooling to manage heat. The video covers the US's estimated 9.3 gigawatt power shortfall tied to AI demand, IEA projections of up to 1,050 terawatt hours by 2030, and company responses including Amazon's nuclear investment in Pennsylvania, Microsoft's renewable energy deal with Brookfield, Google's power-first site selection, and speculative alternatives like Star Cloud's in-orbit AI training and SpaceX's satellite data center plans. It closes with a balanced take acknowledging both AI's benefits and its infrastructure costs.

## Full article

daily.dev links to this article rather than hosting it. Read it at the original source: <https://www.youtube.com/watch?v=qADr2YFPBrA>

## Questions this post answers

### Why do AI GPUs consume so much more power than regular CPUs?

AI GPUs pack thousands of simpler cores that all fire simultaneously to perform the same repeated calculations needed for training AI models, unlike regular CPUs which use a handful of powerful cores for one complex task at a time. This parallel design generates far more heat and draws much more power. Nvidia's H100 chip saw a 75% jump in power consumption over its predecessor within about two years, and newer Blackwell chips can hit 1,200W each.

_Engineers evaluating GPU power budgets for AI workloads can track chip-level power trends on daily.dev._

### How much electricity do global data centers use per year?

Global data centers consume roughly 565 terawatt hours of electricity per year, about 2% of all electricity used on Earth, and this figure is up 12% from the prior year. That puts data centers on par with the entire global aviation industry. The International Energy Agency projects this could climb to as much as 1,050 terawatt hours by 2030 due to rising AI demand.

_Anyone weighing AI's infrastructure footprint can follow energy demand trends on daily.dev._

### What are tech companies doing to address AI data center power shortages?

Companies are pursuing multiple strategies: Amazon invested roughly $20 billion into Pennsylvania to explore small modular nuclear reactors at existing nuclear sites, Microsoft is pursuing a 10.5 gigawatt renewable energy plan with Brookfield, and Google adopted a power-first model that selects data center locations based on available clean power. Better liquid cooling and smarter workload scheduling are also cutting energy per unit of compute.

_Teams planning AI infrastructure investments can track these power-sourcing strategies on daily.dev._

## Similar posts on daily.dev

- [How much energy do data centers and artificial intelligence use?](https://daily.dev/posts/how-much-energy-do-data-centers-and-artificial-intelligence-use--n18jjn9mp) · Hacker News · 1 upvotes · 0 comments
- [Europe should make AI prove its value before giving it priority on the grid](https://daily.dev/posts/europe-should-make-ai-prove-its-value-before-giving-it-priority-on-the-grid-hnv60eiqc) · The Next Web · 0 upvotes · 0 comments

---

Tags: [#gpu](https://daily.dev/tags/gpu), [#ai-infrastructure](https://daily.dev/tags/ai-infrastructure)

[View this post on daily.dev](https://daily.dev/posts/why-ai-data-centers-are-running-out-of-power-boxvuei0b)

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]}
{"@context":"https://schema.org","@type":"TechArticle","headline":"Why AI Data Centers Are Running Out of Power","url":"https://daily.dev/posts/why-ai-data-centers-are-running-out-of-power-boxvuei0b","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/why-ai-data-centers-are-running-out-of-power-boxvuei0b"},"datePublished":"2026-09-18T19:36:15.788Z","dateModified":"2026-09-21T04:32:17.175Z","description":"A video breaks down why AI data centers are straining global power grids, citing that data centers now consume around 565 terawatt hours annually (about 2% of...","image":"https://i.ytimg.com/vi/qADr2YFPBrA/sddefault.jpg","thumbnailUrl":"https://i.ytimg.com/vi/qADr2YFPBrA/sddefault.jpg","isAccessibleForFree":true,"articleSection":"Tiff In Tech","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Tiff In Tech","logo":"https://media.daily.dev/image/upload/s--cw-MSz5l--/f_auto/v1724391415/logos/tiffintech","url":"https://daily.dev/sources/tiffintech"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/why-ai-data-centers-are-running-out-of-power-boxvuei0b","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"gpu,ai-infrastructure","timeRequired":"PT13M","video":{"@type":"VideoObject","name":"Why AI Data Centers Are Running Out of Power","description":"A video breaks down why AI data centers are straining global power grids, citing that data centers now consume around 565 terawatt hours annually (about 2% of...","thumbnailUrl":"https://i.ytimg.com/vi/qADr2YFPBrA/sddefault.jpg","uploadDate":"2026-09-18T19:36:15.788Z","duration":"PT13M","url":"https://api.daily.dev/r/boXVUeI0B","embedUrl":"https://www.youtube.com/embed/qADr2YFPBrA"}}
{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Tiff In Tech","item":"https://daily.dev/sources/tiffintech"},{"@type":"ListItem","position":3,"name":"Why AI Data Centers Are Running Out of Power"}]}
{"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/why-ai-data-centers-are-running-out-of-power-boxvuei0b#faq","mainEntity":[{"@type":"Question","name":"Why do AI GPUs consume so much more power than regular CPUs?","acceptedAnswer":{"@type":"Answer","text":"AI GPUs pack thousands of simpler cores that all fire simultaneously to perform the same repeated calculations needed for training AI models, unlike regular CPUs which use a handful of powerful cores for one complex task at a time. This parallel design generates far more heat and draws much more power. Nvidia's H100 chip saw a 75% jump in power consumption over its predecessor within about two years, and newer Blackwell chips can hit 1,200W each. Engineers evaluating GPU power budgets for AI workloads can track chip-level power trends on daily.dev."}},{"@type":"Question","name":"How much electricity do global data centers use per year?","acceptedAnswer":{"@type":"Answer","text":"Global data centers consume roughly 565 terawatt hours of electricity per year, about 2% of all electricity used on Earth, and this figure is up 12% from the prior year. That puts data centers on par with the entire global aviation industry. The International Energy Agency projects this could climb to as much as 1,050 terawatt hours by 2030 due to rising AI demand. Anyone weighing AI's infrastructure footprint can follow energy demand trends on daily.dev."}},{"@type":"Question","name":"What are tech companies doing to address AI data center power shortages?","acceptedAnswer":{"@type":"Answer","text":"Companies are pursuing multiple strategies: Amazon invested roughly $20 billion into Pennsylvania to explore small modular nuclear reactors at existing nuclear sites, Microsoft is pursuing a 10.5 gigawatt renewable energy plan with Brookfield, and Google adopted a power-first model that selects data center locations based on available clean power. Better liquid cooling and smarter workload scheduling are also cutting energy per unit of compute. Teams planning AI infrastructure investments can track these power-sourcing strategies on daily.dev."}}]}
```

