Digital Anarchy
Why it's worth getting into your own hardware, local AI, and decentralized compute
What I'm seeing
More and more, I hear people talking about pulling parts of their digital life back under their own control: moving off the cloud onto a NAS, running automations on a Mac mini, turning their devices into small personal servers — even mini-farms. Some people are running AI agents on Mac minis and physically hauling them into data centers to get good internet, stable power, and infrastructure more reliable than what they have at home.
At first this surprised me. The thought that popped into my head was: it's like digital anarchy is creeping in. Cyber anarchy. But then I thought — what's actually strange about it? Every reasonably large structure already operates this way. Companies try to keep their data in-house and control their own servers and infrastructure. Governments do the same — they want their data on their soil and don't want to be fully dependent on someone else's systems. For companies and governments this is just normal. But for a regular person it still somehow looks like something unusual. And that's the thought that got me hooked.
What I'm worried about
The other question that got me thinking: what happens if Big Tech keeps cranking up prices? Cloud storage is more or less clear — if iCloud, Google Drive, Dropbox or whatever gets too expensive, you can in theory buy a NAS and run your own storage. But what if AI, large language models, and the strong AI tools keep drifting deeper into the closed-off side? That stuff isn't ours. None of it is guaranteed to us.
Big companies can close models, change access rules, restrict regions, turn off features, raise prices, or leave the strongest capabilities only for enterprise customers. This is especially obvious with AI agents. If you want more than a chatbot — a real agent that runs for a long time, walks through tools, analyzes data, reads documents, takes many steps and works through complex tasks — it starts burning a lot of tokens. And that can get very expensive. So the question becomes: what can you actually count on? You can't just assume we'll always get the most powerful models at a reasonable price, on convenient terms, for any task you want.
I got curious about what can actually be mine. What can you run yourself? Which models can you use locally? How smart and how heavy are they? What does it cost to run them? What hardware do you need? What can you set up on a Mac mini, Mac Studio, mini-PC, NAS, home server, or in a data center? I'm not trying to pretend that one person can replace the entire infrastructure of OpenAI, Google or Anthropic. But there's a different question that matters: which parts of your data, tools, and AI workflows can realistically live on your own hardware.
What I'm hoping for
Another piece that got me interested is decentralized and distributed compute projects. Recently platforms have started showing up where people can offer their hardware remotely for compute.
The idea looks like this: if you have hardware, it doesn't have to sit idle. In theory you can plug it into a network, share some of its compute, get paid or earn tokens, and then spend them when you actually need them. It feels like these kinds of projects could be one of the possible answers to the threat of big companies jacking up prices, closing models, or making strong AI available only to a narrow group of customers.
What I'm planning
I don't just want to watch this from the sidelines — I want to dig into it step by step: what's possible, what's expensive, what's pointless, and what could actually give you more freedom. I'm planning to start small: set up a NAS during my vacation this summer, and see where it goes from there.
What do you think about all this? Is this kind of topic interesting to you? Do you have your own NAS, home server, Mac mini, mini-PC, or other hardware you use as more than just a regular computer? Or does all this still feel like unnecessary complexity?