AI Compute Power By Region -- Global -- Over The Years -- August 12, 2025

Locator: 48839AI.

So many story lines.

Link here. Click on the link and watch the graph change from January 2019. For AI, the inflection point for the US / China, was early 2022. This is one explanation why.

The inflection point for the US / China was early 2022. For an explanation, click here.

The "compute power" comparisons, at a minimum, need to reflect regional populations. The tea leaves suggest, all things being equal, that China's relative percent will continue to fall, the EU will increase.

The big question is whether the gap between the US and the EU will widen or narrow. The population of the US is 330 million, and the population of the EU is 450 million.

Other questions/observations:

Japan's relative percentage.

The fact that Russia, South Korea, and Canada are not broken out separately is also noteworthy.

It will be interesting to see to what extent the relative percents change between the US and China.

US and China: four big factors differentiating the two countries with regard to AI:

• capitalism vs socialism;
• open society vs closed society;
• energy abundance; and,
• entrepreneurial spirit. 

Garbage in / garbage out. 

I'm not quite sure how to articulate this but folks might be able to understand what I'm trying to say.

Large Language Models (LLMs) are a type of AI model, specifically designed to understand and generate human-like text. They are trained on massive amounts of text data. Generally speaking, the data on which American LLMs are trained are generally considered fairly accurate and devoid of government interference. I would imagine the data that LLMs train on in China is much more heavily "influenced" or populated by government propaganda. Taken to the extreme, "all" data in North Korea would be heavily vetted by the central government before being released, and subject to manipulation. It's safe to say that in North Korea there's no chance that AI will amount to anything.