AI and LNG Could Compete for Gas, But It’s Just One Scenario
If the US manages to combine smarter infrastructure and faster financing, then there will be enough natural gas to power both AI and US LNG exports.
I am just back from the Sunshine State, where I managed to carve out a bit of time to admire the Art Deco houses, but the main purpose of the trip was to share my views on The AI Revolution and the Energy Ecosystem, a topic that will likely keep me engaged for the next several years.
In this note, I’ll share only my high-level analysis:
1. The Nightmare Scenario of Blackouts is No Longer Unthinkable
America will learn from the Iberian Peninsula’s blackout. What’s clear is that both private and public sectors must spend more time designing the grid of the future to integrate a more diverse mix of power supply while balancing AI’s growing electricity needs.
Data centers are silently becoming some of the largest "cities of electricity demand" in America.
You’ve probably heard some analogies: In 2023, US data centers consumed more electricity than Sweden (135 TWh). By decade’s end, their consumption could be equivalent to adding a new Florida to the national grid (250 TWh).
In hubs like Northern Virginia, Texas, and Ohio, data centers are already straining local capacity. These grids were never designed for this magnitude of 24/7 industrial-scale energy load.
2. AI-Driven Energy Demand: The Only Direction Is Up
Although there’s no consensus on the exact speed and magnitude of energy demand growth linked to AI, US data center electricity consumption could double or triple by 2030, reaching 6–9% of the national grid (from ~4.4% today).
Every chatbot, cloud computation, and AI model is a massive electricity consumer.
Natural gas-fired power generation in the US is soaring and expanding, driven in part by the escalating electricity demand of data centers and AI applications. In 2024, the US produced a record 104.4 bcf/d of natural gas and consumed 90.3 bcf/d, also a record — driven in part by AI and data center needs.
3. The US is Energy-Blessed, but Infrastructure is the Real Challenge
We need to bet on diversity-- an “all-of-the-above’ energy mix: natural gas, nuclear, wind, solar, geothermal, and next-gen technologies like SMRs (small modular reactors). The US’s energy bonanza is the envy of the world.
Fossil fuels are still indispensable because of their unmatched ability to deliver reliable, dispatchable, large-scale energy. They can become cleaner too thanks to new technologies that capture carbon or detect methane detection technologies.
But abundance does not equal access. The real challenge is energy and electricity infrastructure: turbines, pipelines, transmission, substations — the arteries of a reliable AI-powered economy. We don’t have enough of them.
4. New Investment Opportunities Will Open Up
Let me highlight just one here: blockchain technology. As energy systems become more decentralized, blockchain can bring transparency to electricity trading — and even allow data centers to become active grid participants, selling surplus energy without touching the public grid.
This was also the focus of my recent Energy Vista podcast episode with Nicola De Blasio.
5. AI and LNG Could Compete for the Same Natural Gas Feedstock, But It’s Just One Scenario
Technological breakthroughs may help other energy sources meet rising demand (e.g. giant batteries, geothermal). Nuclear has long lead times, but there are also bottlenecks and shortages in the natural gas supply chain.
Also, AI models may become more energy-efficient faster than expected. That would lower pressure on infrastructure.
If the US manages to combine smarter infrastructure and faster financing, then there will be enough natural gas to power both AI and US LNG exports.
Key questions that should animate the public debate?
In the event of a major grid imbalance or blackout, who will be the villain? Tech companies for overwhelming the grid?
Who will pay for new transmission lines? Tech giants building private infrastructure? End-users paying higher rates? Or will governments and regulators step in?
Are investors prepared to finance a new wave of long-term energy infrastructure — to ensure that the US retains its edge in AI?