If you're worried America could lose influence over technology that changes your working life, I think pulling back is the wrong answer. You don't get a stronger Popeye (AI or Super Intelligence) by starving it of resources...
That is how I think about America's AI future. Popeye is the AI. His spinach is the frontier research, the powerful models, the electricity, and the data centers that let us train and run useful systems. Better engineering matters too: getting more useful work from the same resources is better spinach, not just a bigger can.
My position is that America should keep building. I would rather have a leading role in shaping powerful technology under laws we can influence than depend on decisions made somewhere we have less say.
This is not an argument that another power plant automatically makes a model smarter. It is an argument that capability needs physical support. A brilliant model that people cannot afford to use, or cannot get enough computing capacity to run, has a different practical value from one they can deploy.
The IEA's 2026 analysis forecasts all data-center electricity use rising from 485 TWh in 2025 to about 950 TWh in 2030. That includes non-AI uses. It is a forecast, not a measurement of the future.
Imagine a small business that could use an assistant to organize job photos and prepare estimates. The owner needs more than an impressive demonstration. The service must be available, dependable, and affordable, and the estimate still needs checking. That is where the spinach becomes something useful to a person.
The objection deserves a serious answer: a race to build can produce expensive mistakes, while neighbors bear costs they never agreed to. A proposed data center is not automatically a good project because someone puts AI in its description. Who pays for added capacity? What local effects need addressing? Those are legitimate questions, not proof that the person asking hates progress.
Nor is the lead guaranteed. The 2026 Stanford AI Index describes U.S. and Chinese models trading leads. My concern is that advantages can build on themselves; I cannot prove that falling behind once means falling behind forever.
I want America's laws and democratic choices to have real influence over this technology. An American address alone does not guarantee that. How do we give people meaningful ways to challenge what powerful systems do?
So I am arguing for continued building and capable oversight, not a blank check for every project or a claim that all rules are bad. Slowing ourselves is not the same as slowing the world. The question is how to build the strongest useful Popeye while preserving a real public say over what he does.
AI-assisted editing. Published with Rob Price's approval.