The cost of keeping up
How the cost of learning is landing on four very different groups
I’ve been having the same conversation a lot lately. Not about which AI tools are best, or whether they’re coming for our jobs. It’s about money, and who’s paying for the learning.
It goes something like:
“I’m paying for three subscriptions out of pocket and I’m not sure how long I want to keep doing that.”
“I’m trying to figure out how to work with Copilot, because we’re not allowed to use the Claude Code app.”
“I’m paying 200 a month for Claude Cowork, because the lower tier just isn’t enough for me.”
Four steps
The more I have these conversations, the more I start to see a pattern. People in tech are standing on one of four steps right now, and which one you’re on has less to do with talent or curiosity than you’d think.
The Champion has it covered. Their company pays for the tools, gives them time to learn, and treats it as an investment. They’re building skills on company time with company money.
But even Champions need room to move. The tools change fast, and the ones that work best today aren’t always the ones IT approved. We’re all part of a big experiment right now. The Champions who are learning are the ones whose companies give them the freedom to explore, not just a license to one platform.
The Beggar believes in the tools and wants to use them, but every request for access turns into a business case. They spend more energy justifying a license than actually using it. After a while, the frustration can turn into something that sounds like blame: not enough budget, not enough support. Sometimes that’s an easier story to tell.
The Self-Funder is keeping pace, but on their own money and their own hours. Evenings spent learning what Champions get to learn during the workday.
It’s not always about the money, though. Some Self-Funders could afford the company tools just fine; what they’re paying for is the freedom to move quickly, try things, and find workflows that actually work without waiting for approval. It works, for now. (Let’s not discuss the policy violations here, those are besides the point for now.)
The Stranded can’t get in the front door. They’re resourceful about it: hunting down free tiers, stretching trial periods, stitching together whatever’s available. This takes effort and it shows initiative. But it’s hard to build real, lasting skills this way.
Most people I talk to belong to the middle two.
This story is about access
It would be easy to read all of this as “everyone should be using more AI !!!1!” That’s not what I’m saying.
What I’m seeing is that a professional divide is forming along familiar lines. Access, budget, and employer support. The people who get to build these skills as part of their job are running ahead of the people building them on personal time and personal money. Both groups are pulling ahead of the people who can’t access any of it.
We’ve seen versions of this before: who had a computer at home growing up, or dial up internet, or a “free” version of Adobe Photoshop (looking at you, designers). We’ve seen this before, but with the internet in the mix this is now affecting everyone at the same time.
What’s changing the math
The first wave of AI tools came with flat subscriptions. That was never going to last. Running these models is expensive in a way that traditional software isn’t, and the pricing we’ve been paying has been more of a demo rate than a real one.
OpenAI found this out the hard way with Sora: roughly a million dollars a day in compute costs, with a fraction of that coming back in revenue. They shut it down last month.
That reality is catching up everywhere. Figma started enforcing credit limits on Figma Make this year. Claude Coworkcaps how much you can use based on your tier. Adobe Firefly runs on a generative credit system. The tools that felt unlimited six months ago now have a price per action.
For Champions, this is not something to think about. Their company foots the bill. For Self-Funders, every credit spent is a personal cost-benefit calculation. The Stranded don’t even get there.
I’m not arguing that we should throw more money at AI subscriptions. But it’s worth noticing who’s absorbing the cost of learning right now, and what that means for who gets to build these skills over time.
The question underneath
There’s something else forming here that I keep thinking about. A lot of us have been building up context inside these tools for months. We’ve refined prompts, developed new ways of working, trained our instincts for what to ask and how. It’s starting to look less like “using a tool” and more like a skill.
This is how things used to work: we learned how to do things on the job, and that knowledge lived in us. It was portable. You brought it with you to the next job, and that’s part of what made you valuable as a human being.
But where does this new skill live? If you built it inside a company system, a copy of it stays there when you leave, but it might not be very portable. If you built it on your own account, it’s yours to bring along to your next place of work, but you paid for it yourself.
We’ve never sold our skills to employers, we sold our time and our output. But when the context, the memory, the way you’ve shaped a tool to think the way you think: when all of that lives inside a system someone else controls, the lines get blurry.
I don’t have a clean answer for this, and I’m not sure anyone does yet. But it seems worth paying attention to where you’re building, and what you’d leave behind.