Everything is
getting cheaper

In 2010 my brother gave me an old camera and I started posting photographs to a Facebook page. I had no media background, no training, and no connections. It turned out none of that mattered, because the cost of making and publishing media had just collapsed. A camera, an internet connection, and something to say were enough. A few years earlier, reaching an audience meant getting past a television channel or a film studio. Now a teenager in Pakistan could put his work in front of millions of people without asking anyone's permission. I was that teenager.

I did not understand what had happened to me until much later. The door I walked through did not open because I was talented or lucky, although luck was involved. It opened because the cost of doing the thing fell so far that the gatekeepers guarding it stopped mattering. Once I saw that clearly, I started looking for the same pattern in other places. This essay is the short version of what I found.

It is the most common question I get. You are worried that AI will take your job. Some version of that worry is justified, and I will not pretend otherwise. Work that is repetitive and easy to check is already being automated, and the people it hits first are juniors, the ones doing the routine work that used to be the first rung of a career. That is real and it is already in the data.

But the honest version of the threat is different from the one in your head. The tool does not decide anything. It multiplies whoever is holding it. Which means the question in front of you is not whether AI replaces you. It is whether you are the person holding the tool, or the person competing against someone who is.

AI will not take your job. The person who uses AI will.

Strip any economy down to its parts and it runs on four things: the ability to think, the ability to do physical work, the energy that powers the work, and the raw material you shape. Everything humans have ever made is some combination of those four. For the first time, all four are heading down in cost at once.

• Intelligence falling now↓ The cost of a unit of machine intelligence has fallen roughly a thousandfold in three years.
• Labour falling now↓ Robots are starting to leave the demo videos and enter factories.
• Energy turning↓ Solar is the cheapest electricity in recorded history, and storage is following it down.
• Raw material on the horizon↓ The furthest away, but the cost of reaching space, where most of it sits, keeps falling year after year.

I have to be careful with the word cheaper, because your own bills argue against it. Your rent, your groceries, your electricity are all higher than last year. That is real, and it is not what this claim is about. Two different prices are moving in opposite directions: the price of the things you buy and use up, and the cost of the tools you can build with. Your bills track the first. This essay is about the second.²

One more thing, because it is the part that took me longest to accept. Before a cost falls, it often rises first. Energy is doing this now: AI needs enormous amounts of electricity, demand is surging, and prices are climbing. But a rising price is the signal that pulls in money and builders. Record sums are flowing into solar, storage, and nuclear precisely because power got expensive, and that spending is what eventually makes it cheap.³

The expensive phase is not the opposite of the trend. It is how the trend starts.

So the inputs of everything are getting cheaper. Why should you care? Because every time the cost of making something collapses, the people who were winning stop winning automatically. In media, the studios lost their gate, and the value moved to whoever had taste and could reach an audience. It is happening to software now: a prototype that used to need a funded team and a year can be built by one person in a night.

When building gets cheap, being able to build stops being the advantage. Knowing what to build, and reaching the people who need it, becomes the advantage. Credentials and capital matter less. Judgment and distribution matter more.

The opening is largest for the people the old system kept out, because they have the least to lose from the old rules ending.

This has happened before. Every large leap in history followed the collapse in cost of something essential. Machines collapsed the cost of making goods, and within a few generations, farms emptied, cities filled, and whole new kinds of work appeared. It was painful, it was slow, and the people who positioned early captured most of it. The evidence that we are inside another such moment, and the honest arguments against that view, are laid out in the full version. I am not asking you to take it on faith here. I am telling you what I concluded, and where to check my work.

The move is simpler than the theory.

1. Apply AI to something you already understand

   Not AI in the abstract, not a certificate. Your unfair advantage is the domain you already know: your industry, your city, your community's problems. AI is a multiplier that attaches to knowledge you already have.

2. Build where building just got cheap

   Find one real problem that you or the people around you face. Build a small solution with AI. Show the work publicly, so the people with that problem can find you. Charge for it. Then do it again. You do not need a million users or an investor. You need one real problem and a few people who will pay you to solve it, again and again.

3. Stop competing on being the cheapest pair of hands

   That market is dying, and it deserves to. The work that survives is judgment: knowing what is worth making, and being trusted by the people you make it for.

If you already do freelance or contract work, I know that advice can sound like it is for someone else, someone with time and savings, not someone whose rent depends on the next cheap gig. So here is the narrower version. You do not quit anything. You take one week. Pick a single industry or community you know from the inside, not a generic skill, and find one problem those people complain about repeatedly. Build the smallest thing that solves it, using AI to do the parts that used to need a team. Then show it to ten of the people who have that problem.

That is week one: not a business, not a leap, just the first thing you have ever made for a market instead of for a client. Most of what you build this way will go nowhere, and it will still leave you with more skill and more of an audience than another month of racing strangers to the bottom of a price. The point is not the first thing you build. It is that you stop selling your hours and start owning something that solves a real problem.