Why I Don't Worry About AI's Environmental Impact

I only learned about the environmental impact of AI companies a few weeks ago when my brother brought it up. It’s not that I don’t care about the environment. I’ve just been heads-down building Taproot and hadn’t been paying much attention to the issue.

Still, it got me thinking: should I be concerned?

Honestly, no. The reason comes down to how I think about memory.

Most people experience AI memory through their conversations with AI software. They return to previous discussions, pick up where they left off, and rely entirely on AI companies to store all of that context. That dependency is what makes the environmental issue so hard to solve from the inside. These companies need massive, warehouse-sized data centers just to store and process the information users continuously feed into their systems.

There will probably be solutions down the road. But I don’t think the problem has to be solved by AI companies at all.

Our computers are already memory banks. So why can’t we optimize our own systems more effectively? There has to be a better way to store these conversations ourselves, on our own machines, rather than depending on AI infrastructure to hold them indefinitely.

That’s the beauty of Markdown files.

Whether you’ve never heard of Markdown or just haven’t used it effectively, it solves this problem cleanly. Instead of expecting AI companies to permanently store every conversation, you can save those conversations directly on your own computer with minimal storage impact. Markdown files are extremely lightweight, often just a few kilobytes, which makes them a far more practical way to preserve context without adding strain to large-scale data infrastructure.

That’s why I built Taproot the way I did. Your vault lives on your machine, in plain text, in formats that any tool can read. The memory problem isn’t just a hardware problem. It’s an architecture problem. And the alternative already exists. It just needs to be used.