An engineer’s AI usage could soon cost as much as their salary

Instagram head Adam Mosseri has a striking prediction about where corporate AI spending is headed: within a year or two, an engineer’s AI usage could cost a company as much as their actual salary.

‘I think that you can imagine, at least in a year or two coming, that the burn rate of a strong engineer might be the same as their salary or their cost of employment,’ Mosseri said on a recent episode of ‘Lenny’s Podcast,’ according to Business Insider.

The comment reflects a broader shift in how Mosseri now views AI tokens, the basic units of data AI models process and the basis on which top providers charge businesses. He treats them as a standard corporate resource that needs budgeting alongside GPUs, server storage, RAM and payroll, rather than an open-ended pool for experimentation.

That mindset marks a departure from the early AI boom, when companies including Meta reportedly ran internal leaderboards tracking massive token consumption, a trend known as ‘tokenmaxxing.’ Mosseri now calls that practice ‘a terrible idea,’ and says his own division brought costs down simply by ‘shutting down the silly things that we were doing.’

To keep budgets from spiralling further, Mosseri suggested companies will eventually need to place strict data caps on workers, distributed based on an employee’s proven ability to deliver a positive return on the investment.

Image: Wikimedia Commons/by TechCrunch