VFF - The signal in the noise
News

AI Data Centers Could Rival Major Nations in Gas Consumption by 2035

Read original
Share
AI Data Centers Could Rival Major Nations in Gas Consumption by 2035

U.S. data centers powered by AI demand could consume more natural gas by 2035 than Germany and Japan combined, according to analysis. The surge reflects the energy intensity of large language models and AI infrastructure expansion. This trajectory raises questions about energy supply, grid capacity, and the sustainability of current AI scaling approaches.

  • U.S. data centers could exceed natural gas consumption of Germany and Japan combined by 2035
  • AI infrastructure buildout is the primary driver of projected energy demand growth
  • Natural gas consumption by data centers would rank them among the world's largest consumers of the fuel
  • Current trajectory reflects the energy requirements of large-scale AI model training and inference

Energy consumption at this scale has direct implications for natural gas markets, grid infrastructure, and climate commitments. Policymakers, utilities, and investors need visibility into whether current energy infrastructure can support continued AI expansion without supply constraints or price volatility.

Companies building or operating data centers face potential energy cost increases, grid access limitations, and regulatory scrutiny. Energy availability and pricing will become material factors in AI infrastructure economics and competitive positioning.

  • Natural gas supply chains and pricing could face significant pressure from data center demand concentration
  • Grid operators may need substantial infrastructure investment to handle peak loads from AI workloads
  • Energy costs could become a limiting factor in AI model scaling and deployment economics
  • Regulatory and policy responses around energy allocation and emissions may emerge

Monitor announcements from data center operators regarding power procurement strategies, renewable energy commitments, and location decisions. Track utility capacity planning and any regulatory proposals addressing data center energy demand. Watch for industry responses around efficiency improvements or alternative cooling technologies.

OneUpAI
OneUp Your Business. Get More Done. OneUp Your Business. Get More Done. OneUp Your Business. Get More Done.
Learn More
Share

Subscribe to the newsletter

The latest stories and analysis, delivered to your inbox.

Free. No spam. Unsubscribe any time.

Related stories

Power, not GPUs, becomes AI's real bottleneck

Power, not GPUs, becomes AI's real bottleneck

As AI infrastructure demands surge, electricity has become the primary bottleneck for data center expansion, with utilities reporting multiyear waits for new grid connections. Industry leaders are pursuing three distinct strategies: optimizing existing grid capacity through demand management, building microgrids as temporary bridges to full grid integration, and developing sites with spare capacity already available. The shift reflects a fundamental realization that the constraint is not total power availability but rather the speed and efficiency of delivering it to servers.

by The Information Partnerships· The Information
Deepgram Acquires Daily to Integrate Real-Time Voice Infrastructure
TrendingNews

Deepgram Acquires Daily to Integrate Real-Time Voice Infrastructure

Deepgram has acquired Daily, the maker of Pipecat, to strengthen its voice AI infrastructure capabilities. The deal gives Deepgram access to Pipecat's real-time voice and video infrastructure technology. The acquisition positions Deepgram to offer more comprehensive voice AI solutions by combining its speech recognition and processing capabilities with Daily's real-time communication infrastructure.

by Ivan Mehta· TechCrunch AI
NVIDIA Releases Nemotron 3.5 Super VL on AWS

NVIDIA Releases Nemotron 3.5 Super VL on AWS

NVIDIA has released Nemotron 3.5 Super VL, a 120B-parameter multimodal model available on AWS infrastructure including Amazon EKS and SageMaker HyperPod EKS. The open model combines reasoning and tool use with text, image, document, and video understanding, activating only 12B parameters per token through a mixture-of-experts architecture. It supports agentic AI workflows for coding, web research, and document intelligence tasks.

by Alex Iankoulski· AWS Machine Learning Blog
Beta Technologies Demonstrates Electric Aircraft Economics at Scale

Beta Technologies Demonstrates Electric Aircraft Economics at Scale

Beta Technologies flew a journalist in its Alia CX300 electric aircraft at Patrick Leahy Burlington International Airport in Vermont, demonstrating a half-hour flight consuming only $9 worth of electricity compared to $210-$255 for a comparable gas-powered aircraft. The company is pursuing a stepwise market entry strategy focused on electric cargo and medical aircraft using conventional runways, backed by a claimed order and option-to-buy backlog of 991 aircraft. Beta's approach differs from competitors Archer and Joby, which are focused on electric vertical-takeoff-and-landing air taxis.

by Tom Clynes· IEEE Spectrum AI