The 100x Engineer Left. Build Anyway.

TL;DR: The best engineers Canada trains keep leaving for Silicon Valley, and that brain drain is real and mostly permanent. It matters far less than it used to, because AI agents turned elite engineering output into something you rent instead of something you hire.

In my last piece I said the worst handicap of building outside Silicon Valley is the one everyone fixates on: your best people leave. I want to take that apart now, because it is true and it matters less than it used to.

The 100x engineer is real, and Toronto keeps losing them

Quick answer first. Yes, the 100x engineer exists, and Canada has a long habit of training them and then watching them go.

I used to think the term was hype. No such thing, I figured, just 10x people having good weeks. I was wrong. Some individuals are worth a hundred ordinary engineers. Here is the painful part for anyone building in Toronto. We grow them. Then we lose them.

Look at the track record. Andrej Karpathy did his undergrad at the University of Toronto, then left for Stanford, OpenAI, and Tesla. Ilya Sutskever came out of Geoffrey Hinton’s lab at U of T and ended up at OpenAI. Hinton did the foundational work right here. A big share of the modern AI industry was seeded by people Canada educated and then watched board a flight to California. That is not bad luck. It is gravity.

Why the Valley pulls the very top

The top wants to be near the top. The capital is there. The frontier labs are there. The other geniuses are there too, and density compounds, one brilliant person drawing in the next. The equity math works in a way it rarely does here. If you are the kind of engineer who can move the frontier, every incentive you have points at the same few square miles. AI only made that pull stronger, because the labs building AI sit right in the middle of it.

So the brain drain is real. It is more or less permanent. Nothing I am about to say makes it disappear.

Here is what changed anyway.

Leverage moved from headcount to orchestration

For my whole career, the unit of leverage in software was headcount. The question was always how many strong engineers you could put in one building, and the Valley won that question by gathering the most talent in the smallest space. That was the whole game. It is not the game anymore.

The unit of leverage now is orchestration. A strong 10x engineer who can architect a system, direct a fleet of AI agents, and review what they produce ships work that used to take a full team. The output of a 100x engineer is no longer something you hire. It is something you rent. You point good people at good tools and you get results that would have demanded a roster of stars five years ago.

Who the brain drain actually hurts

The answer is narrower than you would think.

If you are trying to build a frontier AI lab in Toronto, losing your Karpathy is fatal. You are in a talent war at the absolute top end, and you will lose it from here. Do not try.

But almost nobody is building a frontier lab. The rest of us are building products on top of the models those labs ship, and for that kind of company the math is completely different. You were never going to keep a Karpathy, and you do not need one. You need a few strong engineers, which Toronto keeps in abundance, pointed at the tools the geniuses left town to build.

Read that again, because it is the whole argument. The 100x people moved to California to build the leverage that you now run from Ontario at half the burn. Their departure is the reason your small team can do so much. You are not competing with them. You are their customer.

The catch, because there is one

AI multiplies whoever it touches. A 10x engineer with agents becomes a 100x engineer. A mediocre engineer with agents becomes a machine for producing mediocre code faster. The multiplier rewards the people who were already good. So the play is not to wait for a genius to show up. The play is to hire the best 10x engineers you can find, and they are findable here, then arm them properly and get out of the way.

I am living this

I am one person building a full mobile app, document pipeline and all, with AI agents working across the entire product and engineering lifecycle. Twenty-eight years ago that was a funded team of a dozen people. Today it is me and a stack of tools, sitting in Ontario, shipping. That is not a prediction about the future. It is what my Tuesday looks like.

And it lines up exactly with what I keep writing about in the markets: agentic AI is quietly rewriting who has an advantage in software.

The genius still gets on the plane. Let him. He is going to California to build the thing that lets the rest of us stay.