The era of cheap AI is over

Over the last months many treated AI as abundant. Cheap tools, broad access, almost unlimited capacity. That's over.

GitHub started rate-limiting and charging more for advanced models. Others will follow. It's not pricing strategy — it's physics: compute is the bottleneck, and bottlenecks become price.

Three facts

• AI is not going away.
• AI will not get cheaper in the short term.
• AI will be progressively rationed.

Whoever used the cheap window to learn, integrate and redesign their processes now operates at another level. Juniors are more productive, vibe coding democratized creation, the way we work changed. But the edge is no longer in access — it's in usage maturity.

As with the steam engine, electricity and the internet — the gain was not distributed evenly. It was captured by those who anticipated.

It's not about using AI. It's about being structurally ready to operate with it when its cost stops being irrelevant.