The Inktelligence - November 17, 2025

AI's Bottleneck: Not GPUs but the Grid

Author

Hock Ng
November 17, 2025

The Real AI Race Isn't About Models. It's About Megawatts.

Everyone's watching the latest AI capabilities. The real pressure is building in the grid, not the GPU.

If you follow the AI hype cycle, it's all GPUs, benchmarks, and AI agents. But there is another angle to look at the AI boom. It’s the power grid. The land. The water. And the communities wrapped around all of it.

The actual constraint on AI's growth is physical infrastructure.

This creates two things that we should pay attention to: data-center-adjacent infrastructure, and a public-impact story that's becoming too large to ignore.

1. AI Infrastructure as the Next Major Asset Class

AI demand isn't growing linearly. It's exploding.

Global data-center electricity demand is projected to increase up to 165% by 2030. U.S. data centers consumed 183 TWh in 2024—over 4% of national electricity—and could hit 426 TWh by 2030. McKinsey estimates global data-center build-out through 2030 could reach $6.7 trillion, with $5.2 trillion of that for AI-optimized centers.

Here's the kicker: power access, not land, is now the primary limiting factor in top data-center markets.

Compute demand is outpacing infrastructure supply. If you’re an investor, that gap creates investable themes:

  • Grid-proximate land with interconnection potential

  • On-site generation (solar, storage, gas peakers, advanced nuclear)

  • Liquid cooling and water-efficient cooling technologies

  • Transmission projects and grid-services markets

  • Data-center REITs and private-equity roll-ups

This is one of the largest physical build-outs of the decade. It comes with major constraints.

2. The Power Problem

Data centers can be built in 18–30 months. Transmission lines take 7–10 years.

That mismatch is now visible. U.S. spare generation capacity fell from 26% to 19% in five years. If AI demand continues, spare capacity could drop below 15%—a level utilities consider "critically tight."

Some AI-focused facilities now require hundreds of megawatts to multiple gigawatts. Proposed campuses are approaching 5 GW.

The bottleneck isn't GPUs. It's electrons.

The Simplest Way To Create and Launch AI Agents

Imagine if ChatGPT, Zapier, and Webflow all had a baby. That's Lindy.

With Lindy, you can build AI agents and apps in minutes simply by describing what you want in plain English.

From inbound lead qualification to AI-powered customer support and full-blown apps, Lindy has hundreds of agents that are ready to work for you 24/7/365.

Stop doing repetitive tasks manually. Let Lindy automate workflows, save time, and grow your business.

3. Water, Cooling & Land

Cooling AI clusters takes tremendous amounts of water.

U.S. data centers used 17 billion gallons directly for cooling in 2023, and over 200 billion gallons indirectly through power generation. A single large data center can use up to 5 million gallons of water per day—similar to a town of 10,000–50,000 people.

New AI campuses are increasingly built in water-stressed Western states, deepening resource tension.

This is where the investment story intersects with the public-impact story.

4. Who Pays?

Air Quality and Health

Communities near EPA-permitted data centers face higher NO₂, diesel particulates, and PM levels than national averages. As backup generators, truck traffic, and increased grid reliance scale up, air-pollution-related health damages from data centers reached ~$6B in 2023. That could reach $10–20B annually by 2030.

Grid Upgrades

Data-center build often requires massive transmission investment. Unless regulators intervene, those costs are generally spread across all ratepayers.

One example: In parts of the PJM market, data-center capacity additions contributed to $16–$18/month increases in residential electric bills.

Land Use and Noise

Communities across the U.S. (Virginia, Oregon, Arizona, Texas) are raising concerns about 24/7 generator noise, loss of rural and agricultural land, traffic and construction disruption, reduced air quality, and water table impacts.

The Opportunity and the Warning

The infrastructure behind AI isn’t hidden. It’s just overlooked. And it’s already reshaping the places we live — our power grids, water systems, industrial corridors, and even neighborhood health.

The communities and municipalities that navigate this moment well will be the ones that understand both sides: AI is a huge opportunity, and it comes with real-world impacts that need to be managed.

This isn’t about taking sides. It’s about paying attention.

Sources

  1. Goldman Sachs Research – AI to drive up to 165% increase in global data-center power demand by 2030

  2. Pew Research (2025) – U.S. data center electricity demand: 183 TWh (2024) to 426 TWh (2030), regional bill impacts, and national consumption share

  3. McKinsey & Company (2025) – Global data-center build through 2030: $6.7T total, $5.2T AI-optimized

  4. CBRE (2024/2025) – Power access as primary constraint in top global data-center markets

  5. Goldman Sachs Research (2025) – U.S. spare generation capacity decline from ~26% to ~19%, projected <15% by 2030

  6. Deloitte Insights (2025) – AI campuses modeled at 2 GW with proposed clusters approaching 5 GW

  7. Lawrence Berkeley Lab estimates; summarized by Pew – 17B gallons direct cooling water, 200B gallons indirect water (2023)

  8. Environmental & Energy Study Institute – Large data-center water usage up to 5M gallons/day

  9. Stanford's And the West – Data centers expanding into water-stressed Western regions

  10. IEEE Spectrum – Estimated $6B annual health damages from data-center pollution, potential $10–20B by 2030

From Boring to Brilliant: Training Videos Made Simple

Say goodbye to dense, static documents. And say hello to captivating how-to videos for your team using Guidde.

1️⃣ Create in Minutes: Simplify complex tasks into step-by-step guides using AI.
2️⃣ Real-Time Updates: Keep training content fresh and accurate with instant revisions.
3️⃣ Global Accessibility: Share guides in any language effortlessly.

Make training more impactful and inclusive today.

The best part? The browser extension is 100% free.

I’d like to hear from you!

If you would like provide me with any feedback, suggestions, thoughts, discoveries, or just say hello, please write to me (I promise that I will read every single email): [email protected]

Till next time,

Hock