Could All These AI Data Centers Be Worthless? Let's explore.

Based on a post by @BelannF on X - I came across it and could not stop thinking about it. The comparison stuck with me, so I wanted to write it out and explore where it leads.

There is a concept that post explained that I had not heard anyone articulate before. It is abstract synthesis versus concrete synthesis. That sounds heavy, I know. But once you get it, you will understand why there is a serious question about whether the biggest AI buildout in history might be heading the same way as the NFT boom.

Chess robots and soccer children

The simplest way I can explain this: robots have been beating us at chess since the 1980s. Thirty years. But there is not a robot in the world that can outplay a clumsy child at soccer.

Chess is abstract. Soccer is concrete. That is the entire difference.

Abstract synthesis can become infinitely more efficient. The best example is the telegraph versus the telephone. The telegraph needed Morse code, shortened words, paper printouts, a person running the message across the office. Very inefficient.

Then Alexander Graham Bell invented the telephone. Same wires. Same infrastructure. Immeasurably more efficient.

That is abstract synthesis.

Why Moore’s Law only applies to made-up things

You have probably heard the expression: if airplanes developed like computers, they would fly around the sun and give you a massage. That is a fun thought, but it misses something. Airplanes are concrete synthesis. You have to build them, test them, crash them, rebuild them. It is linear.

Computers are abstract synthesis. They can double in efficiency every two years because it is all made up. We are just inventing better ways to invent.

A lot of people assume Moore’s Law applies to everything. It does not. And that distinction matters more now than ever.

What is being built right now

These companies are building data centers the size of 60 square miles. They are laying down concrete and wiring, essentially trying to make the telegram faster.

The question I keep coming back to: is this the right approach, or are they building the telegraph while someone else is about to invent the telephone?

DeepSeek was a warning shot. You cannot put a trillion dollars on the line without someone racing to beat the system. I think someone will figure out how to make an AI model as powerful as ChatGPT that runs entirely on your iPhone. I do not know when. But it feels inevitable.

If that happens, the hardware being built right now becomes obsolete almost overnight. The same way NFTs went from everywhere to irrelevant.

I am not saying this will happen. I am saying it might. And it is worth thinking about.

If you want to follow the hardware side of this story, our piece on the AI hosting inflection point covers where the buildout actually stands.

What this means for real businesses

If I am right, the skills that hold real value are the ones that cannot be abstracted. Hard trades. Plumbing. Electrical. Building. Things that require hands, judgment, and local presence.

That is exactly the kind of business Underdog Digital exists to market.

And for the businesses that do need to be online, the fundamentals have not changed. Clear writing. Real expertise. Local authority. Those things do not become worthless when a new model drops.

The honest answer

I do not know for certain. Nobody does. But the theory is worth exploring. The post by @BelannF made me think about it seriously, so I wanted to write it out and share it.

Full credit to them for the comparison. Read the original on X and form your own view.

---

Written by Oliver Worthington Underdog Digital