My life & career
AI services depend on physical grids; technology choices and electricity availability now interact.
YOTTABIT / THE BIG STORIES / Energy
What if AI's biggest bottleneck is electricity?
Better AI relies on real physical infrastructure: chips, buildings, cooling systems and reliable power. That's easy to forget when intelligence arrives through a small screen.
THE BIG PICTURE
The growth of AI could change where companies build, how cities plan, and which energy innovations become strategically important. At the same time, AI may help manage parts of the energy systems it depends on.
Energy forecasts are scenarios, not promises. Grid connections and equipment take years to build.
THE YOTTABIT PERSPECTIVE
What does it mean to me?
Big change is fascinating. Its implications are what matter.
My business
Evaluate the full resource cost of AI infrastructure, including power, cooling and connection delays.
My industry
Utilities, data-center operators and industrial customers face a new balance between digital demand and physical build times.
FROM JIM CARROLL’S WORK
In his 2026 energy keynote, Jim described AI as both the load and the solution—and the grid as a distributed, increasingly software-defined system.
Meet the futurist behind YottaBit ↗WHAT COULD I DO MONDAY MORNING?
Add a speed-to-power question to your next digital infrastructure discussion.
CURIOUS FOR MORE?
Explore the story behind the story.
For those who want the deeper explanation, the research is always available. The fascinating possibilities are only the beginning.
Research, explanations and sources EXPLORE THE DETAILS +
Two different speeds
Models can improve within months. Major power-generation, transmission, interconnection and substation projects can take years. The IEA estimates data centers used about 415 TWh in 2024; a later report edition estimates about 485 TWh in 2025 and projects approximately 950 TWh in 2030. Those report vintages and projection assumptions must be preserved rather than spliced into a false observed series.
Demand and solution
AI increases demand for computing infrastructure, but forecasting, optimization and control could also help grids manage complexity, improve planning and coordinate distributed energy resources. The net effect depends on reliability, implementation costs, policy, local constraints and additional demand triggered by cheaper computation.
The real constraint
Energy, cooling, equipment availability and grid access determine where and when infrastructure can be built. A compelling AI forecast that ignores these constraints is an incomplete forecast of the Yottabit Era.