Ai Data centers a scam? do we really need that much compute architecture? nope

Hinzugefügt: 2026-05-16

[00:00:00]There are big battle lines being drawn right now with the people that are pushing and fighting for AI and these giant data centers and the ones that are opposed to it. And what's interesting to me is that there's a very important element completely being skipped in all of this. The big need for AI data centers has to do with compute power. AI is heavily dependent on infrastructure.

[00:00:24]They need a lot of processes to do the things and it's broken into two primary categories, right? One of them is the learning part which is building a new model or keeping a model up to date by reading and learning and consuming information to become aware or smart whatever they want to call it. And the other side is the inference which is the AI model being able to handle all of the requests that people are making of it.

[00:00:49]your little chat GPT question or giant enterprise integrations with AI. These models are working. They're workers.

[00:00:57]They're doing stuff. Both of those need big giant data centers to handle that work. The opposition argues against this shift towards AI because of the damage that these big data centers do. But they're fighting the inevitable. It's like arguing against the steam engine.

[00:01:13]We know it's here. You can't get away from it. And there's a lot of extremely powerful uses for AI. But here's the part that I think that everybody's missing. The people that argue you need those giant data centers to handle the workload are lying to you because you don't. You'd be surprised to realize that 50 to 70% of all of the work they do, the tasks could all be handled locally using a different architecture using things like what I'm developing the cognitive world model which is a state driven state first runtime that basically contemplates what do I really need to send to the big AI engine.

[00:01:49]Another part of the problem is the speed of adoption. People are integrating and moving towards these AI models without understanding it, without realizing you don't have to send everything to the big model. The big corporate oligarchs that own these AI infrastructures, these models benefit from everybody staying ignorant to the fact that you don't really need those big models to do most of the things you're currently asking them to do. There's a better way to do it, a more efficient architecture. You can still use AI, benefit from all of the new technology, but not send all your damn money to these organizations.

[00:02:28]Policymakers shouldn't be arguing against AI as a whole. They should be regulating and forcing the efficient use of data AI. And this isn't theoretical.

[00:02:39]I've literally built a model that proves this out as has as as have others. I wrote a whole thesis about it. It's on my substack. Go check it out if you're interested. Reach out. If you have any questions, happy to talk about it. And if you're so inclined, subscribe to my Substack or maybe donate to my GoFundMe to help support my work.