233,000 tech layoffs: AI as alibi for a class decision

According to Le Figaro, which speaks of an "annus horribilis" for employment in American tech, every major group in the sector has drastically cut its headcount this year. As of 30 October, nearly 233,000 workers had been laid off, and that figure does not include the expected departures at Microsoft. The paper evokes, behind these plans, the shadow of artificial intelligence. The dispatch in front of us does not detail the breakdown by company, the real share of AI in each decision, or the trend in profits. We will not invent it, then.

What follows is our reading, not an established fact.

First observation: "AI" is presented here as a grammatical subject. AI "threatens", AI "replaces". A neat sleight of hand: a machine fires no one. Boards and shareholders decide to turn a tool into a reduction of the wage bill. Calling this a technological phenomenon is commodity fetishism applied to the algorithm: a social decision takes on the appearance of a force of nature.

Second observation, more down to earth. If a machine allows the same output with less living labour, the productivity gain does exist. The question is who gets it. In the logic of profit, the labour saved is not redistributed as free time or wages: it becomes surplus for capital. A laid-off worker receives nothing more; those who stay produce more, for the same pay, under the threat of the colleague at the next desk who was let go. Layoffs after a good quarter, productivity gains that never reach the payslips: this is the relation of surplus value made visible. The dispatch does not say whether the groups concerned are posting such results; that is the first thing to check.

Third point: 233,000 more people on the labour market, often highly qualified, is a brand-new industrial reserve army. Marx wrote it in the chapter of Capital devoted to the general law of accumulation: accumulation progressively produces a relative surplus population, an industrial reserve army (our translation). It also serves to discipline those who remain: it is hard to demand a raise, remote working or a say over the tools when the queue outside the door grows longer. Tech engineers believed themselves sheltered because they were well paid; political economy makes no exception for Silicon Valley badges.

Finally, one may ask whether this wave is not also part of the race to invest, in which everyone spends massively so as not to fall behind, then squeezes labour to restore their margins. The dispatch does not allow us to assert this. But the mechanism is classic: the crisis is always resolved by shifting the bill downward.

AI is therefore not the subject. The subject is ownership: who owns the machines, who decides how they are used, and who collects the difference. As long as the answer stays the same, every technical advance will look like a redundancy plan.