Powering the AI Boom: Data Centres Shift to Dedicated Gas Generation
These articles are AI-generated summaries. Please check the original sources for full details.
Power markets affecting the next generation of data centres
A consortium led by Babcock & Wilcox, Base Electron, and Applied Digital is constructing 1.2GW of gas-fired generation in North Dakota. This output is comparable to an average nuclear reactor and requires 5.6 million cubic metres of natural gas daily.
Why This Matters
The technical reality of AI infrastructure requires constant, high-load electrical capacity that existing local grids cannot sustain. While renewable energy is expanding, its weather-dependent nature often fails to meet the consistent computational demands of next-gen data centres, forcing a shift toward heavy industry models where tech firms must finance and operate their own dedicated power and transmission infrastructure.
Key Insights
- Global data centre electricity consumption is forecast by the IEA to more than double from 460TWh in 2024 to over 1,000TWh by 2030.
- Babcock & Wilcox is designing four gas turbine units for North Dakota AI campuses, a project valued at $2.4 billion for 1.2GW of capacity.
- JLL reports that North America currently leads data centre construction with over 35GW of capacity underway.
- Natural gas is expected by the IEA to supply approximately 26% of data centre power demand by the end of the decade.
- Major tech firms including Amazon, Google, and Meta have signed pledges to finance new generation and transmission upgrades to avoid grid strain.
Practical Applications
- Use case: Comstock Resources delivering gas directly to NextEra Energy plants to stabilize fuel supply for AI centres. Pitfall: Relying on the public grid for massive AI loads can trigger public policy backlash and price hikes for domestic users.
- Use case: Viking Data Centres converting a former Goodyear factory in Akron into a 150MW AI and crypto facility using local gas. Pitfall: Underestimating the heavy industry nature of energy procurement can lead to revenue-limiting capacity constraints.
References:
Continue reading
Next article
Self-Hosting a Production SaaS on Mini PCs: A High-Availability Infrastructure Guide
Related Content
Google's $15B Data Centre Expansion in India: Scaling AI Infrastructure in Asia
Google is launching a massive US$15 billion, 1GW data centre hub in Andhra Pradesh to support AI workloads and regional data residency requirements across Asia.
AI Infrastructure Spending Forecasted to Reach US$650 Billion by 2026
Major tech firms are projected to invest US$650 billion in AI infrastructure by 2026, shifting industry focus from software to physical hardware like photonics and GPU clusters.
Microsoft's $9.7 Billion Data Centre Partnership with IREN and Nvidia Chips
Microsoft secures access to Nvidia chips via a $9.7 billion deal with IREN, expanding AI infrastructure capabilities while addressing global computing power shortages.