LLMs can be run by anyone

It doesn't take a huge company to run LLMs. It might have taken big team to develop the frameworks we use today, but now anyone can run a modest-sized LLM on the hardware they have at home.

Will it be as broad & capable as the likes of Claude & ChatGPT? No.

Can you instead choose a specialised model for a specific purpose? Yes.

Will it run locally, where you might feel more comfortable granting it access to your files? Yes.

The open source models are a big deal. They're already capable of doing really impressive things, like transcription, image generation, and natural language-based data transformation, running on commodity hardware. I run several models on the laptop I'm typing this on – a computer that doesn't even have a GPU.

Cory Doctorow (https://pluralistic.net/2025/10/16/post-ai-ai/)

I don't believe LLMs are a technology different from any other that started off in the hands of a few, eventually becoming so mainstream so as to be barely acknowledged. Whilst these tools are publicly accessible by anyone with an internet connection today, my thinking is focused on openly accessible forms of the technology & not the loss-leading services big tech companies run today in hopes of one day turning a profit.

Onion routing (TOR) & GPS, just to name two, started off as US Military projects & today are used by countless numbers of people, directly & indirectly. I doubt that it would have been dreamed of that any person with a consumer-grade phone in the pocket would one day have access to GPS.

So I think LLMs will follow this common trend. Whilst it seems that every tech company is currently loading up any software they can with a AI features, one day soon we will find ourselves with local LLMs (or some other form of AI) that are as democratised as the inventions that came before them.

At which point focus will be on the "next big thing" that we will one day welcome into our everyday lives.