In April, the startup AfterQuery raised 30 million dollars (around 26 million euros) at a valuation of 300 million. Five months later, according to published information, the same firm is valued at 3.2 billion dollars - around 2.8 billion euros. That is more than a tenfold jump in less than half a year.

The founders are 22 and 23. Eighteen months ago they were in Y Combinator's winter batch. Gustaf Alstromer, a partner at the accelerator, says this is the fastest run to a billion in that house's history. The wording is accurate and it is also an advertisement - Y Combinator sells exactly this kind of promise to every subsequent cohort of founders.

What does the firm actually do? It does not build a model and it does not sell an app. It hires doctors, lawyers and other professionals to train models - not to answer questions correctly, but to reproduce the working patterns of a person from the profession. The company describes this as "encoding the patterns, decisions and reasoning of the world's best practitioners." In simpler words: paid experts hand over their own craft to software that will then be sold without them.

From San Francisco they announced that annual revenue has already reached a run rate of 100 million dollars and that they work with the big labs. Named clients include Nvidia, Legora and South Korea's Motif Technologies. Mercor and Scale already operate in the same niche - which means this is not an isolated phenomenon but an entire layer of the industry built on having people with degrees train the models.

The relationship between the two numbers is worth pausing on. Revenue of 100 million, a valuation of 3.2 billion - that is a price thirty-two times the annual intake. Investors are not buying revenue, they are buying the expectation that revenue will multiply. If that expectation holds, somebody made an excellent move tonight. If it does not hold, whoever came in last pays the difference.

For a region where a qualified engineer looks for work paying a few hundred euros, the most interesting detail is not the number but the business model: a professional's knowledge is bought once and sold endlessly. The question that calculation raises is not whether it will reach the Balkans, but what they will offer here for the same work.