For three decades the trade was simple. If you were the best math student in your year, you went to a hedge fund. The pitch has changed. As one former Jane Street trader described the labs' offer, they are now selling the chance to "build the machine god."

The defection is real and it has names.

And the people already running the labs got there the same way. OpenAI's chief research officer Mark Chen and Perplexity co-founder Johnny Ho both carry quant backgrounds. The trade is not new. The direction reversed.

The money is loud. Per reporting, OpenAI has dangled packages worth around $3 million at junior quants. Anthropic runs dinners aimed squarely at high-frequency-trading researchers, plus quant "social hours" in New York and London, with OpenAI and Perplexity running parallel campaigns. The funds are answering with higher bases, shorter vesting, and equity they used to keep locked away.

The work converged first

The people followed the tools. Jane Street committed several billion dollars to CoreWeave for compute and describes its own research as training large, complex models on noisy data, a line you could lift straight from an Anthropic press release. A CoreWeave executive called Jane Street a shop that operates "like a frontier lab." When the math, the compute, and the engineering are the same, the only thing left that differs is what the model points at: a trading book that earns a spread, or a product that ships.

There is a career-shape argument underneath it too. Quant careers can be U-shaped. Junior researchers tend to produce far more than they are paid, while senior ones eventually run into the ceiling of what the market itself will give up. A lab offers the same hard problem with more option value and a better story attached.

One number cuts against the panic. Citadel took 108,000 internship applications this year, up about a fifth. The funnel into quant is bigger than ever. The labs are not draining the pipeline. They are skimming the very top of it, which is precisely the slice the funds most need to keep.

The arbitrage stopped being compensation a while ago, because the funds can match three million dollars. It is narrative and optionality. A twenty-five-year-old picks the offer with the bigger option value and the better story, and right now that is the lab. The labs did not build a new sourcing funnel. They tapped the cleanest pre-screen in existence, because a standing quant offer already certifies the exact profile they want. The funds will respond not by paying more, which they can, but by manufacturing mission, which they cannot fake. Watch who wins that part.