The Data
What the Numbers Say
OpenAI's chief global affairs officer Chris Lehane sent a filing to the Office of Science and Technology Policy laying out the stakes in stark terms. The US and China are not in the same race. They are not even in adjacent stadiums.
These are not theoretical projections. This is what was actually built last year. Every AI model, every data center, every inference run requires electrons — and electrons require infrastructure that takes years to plan and build. The electron gap is not a future problem. It is already baked into the next decade.
"The nation should add 100 gigawatts in new energy per year to help close the 'electron gap' with China — this would both account for baseline demands and fuel American AI dominance, ensuring that US companies have the power they need."
The Stakes
Why Electrons Are the New Oil
Every previous era of geopolitical competition had a defining resource. Coal for the industrial revolution. Oil for the 20th century. Semiconductors for the 1990s–2010s. The AI era has a new scarcity: reliable, abundant, cheap electricity at massive scale. Whoever builds it fastest wins.
AI models are not compute-constrained in the way people imagine. The bottleneck is not chips. It is the power to run the chips. A single large AI training run can consume as much electricity as a small city for weeks. Inference at scale — billions of queries per day — requires gigawatts of continuous power. Data centers are power plants with servers inside them.
OpenAI, Oracle, and SoftBank's joint infrastructure investment to build US AI data centers and close the electron gap.
The US leads in count — but count means nothing without electrons. Every new data center needs a power plant behind it.
OpenAI estimates its own infrastructure plans will require 20% of the entire existing workforce in specialized electricians and mechanics.
Doubling US annual energy additions. Not for climate or industry. For AI. The electron gap reframes energy as national security.
AI inference must be instant. Latency is downstream of power availability. Whoever has electrons closest to users wins the AI product war.
Power plants take 5–10 years to permit and build. The decisions made in 2025 will determine the AI power map of 2035.
The Road Ahead
How the Electron Gap Gets Closed
OpenAI's filing outlines a specific roadmap — not just a goal, but a set of policy levers the federal government must pull to make 100 GW per year achievable.
OpenAI coins "electron gap" in a White House filing. The framing shifts AI policy from a compute/chip conversation to an energy and infrastructure conversation. The term enters the national security lexicon.
Tightening FERC Order 2023 timelines, creating fast tracks for shovel-ready projects, and accelerating grid interconnection approvals. The bottleneck is not generation — it is transmission and permitting.
Six Stargate sites already underway in Texas, New Mexico, Ohio, and Wisconsin. Each requires dedicated power infrastructure. The skilled trades workforce gap — electricians, mechanics, ironworkers — must be closed in parallel.
If the US closes the electron gap, American AI companies retain compute and inference dominance. If not, AI workloads migrate to where electrons are cheap and abundant — and geopolitical leverage migrates with them.
Domain for sale
Own the domain that names the defining infrastructure contest of the AI century. As "electron gap" becomes policy shorthand, media vocabulary, and investment thesis — this domain sits at the center of every search.
Acquire or Inquire
// Serious inquiries only. Response within 24 hours.