996 and the fetishization of overwork

Sometime in 2025, a job ad for a San Francisco AI startup made the rounds for the wrong reasons. It asked, with no apparent embarrassment, for candidates willing to work “996” — and a chunk of the founder internet nodded along, as if the phrase were a badge rather than a warning. Around the same stretch, a parade of prominent voices took to their feeds to celebrate the 80-hour, 90-hour, no-weekends grind as the natural price of building anything that matters. The message, stripped down, was simple: if you are not destroying yourself, you are not serious.

It is worth pausing on what “996” actually names, because it travels as a vibe and lands as a schedule. The term comes from China’s tech sector and means nine in the morning to nine at night, six days a week — about 72 hours of work in a week that has 168 hours in it. That is not a metaphor for hustle. It is a literal timetable that leaves you, after sleep and a commute, with very little week left to be a person in.

For a datarekha reader — someone building a data, ML, or engineering career, often in a culture that already romanticizes the all-nighter and the heroic on-call save — this is not abstract. The glorification of brutal hours is creeping back into exactly the rooms you work in. So this piece does something deliberately unglamorous: it puts the worship of long hours next to the health data and lets the gap speak. The short version is that hard work and long hours are not synonyms, and the evidence that the second one quietly kills people is far stronger than the founder feeds would like you to believe.

What 996 actually is, and why it is illegal where it was born

Start with the legal fact that rarely survives the meme. In China, where 996 was born and named, it is not some accepted local norm — it is against the country’s own labor law. The standard workweek there is capped well below 72 hours, and courts and regulators have repeatedly said so, including a high-profile 2021 ruling that called the schedule a serious violation. The practice persisted anyway, especially through the tech boom, because enforcement is uneven and ambition is contagious. After the crackdowns it softened in name, but “invisible overtime” — the kind that lives in late Slack messages rather than official timesheets — never really left.

That detail matters because of how 996 is now being imported. When a Western founder invokes it admiringly, they are borrowing the aesthetic of a system its place of origin formally treats as illegal and that its own workers largely endured rather than chose. The grind is repackaged as a competitive edge, shorn of the context that it was a thing people fought to get out from under.

The 2025 import, and the founders who pushed back

The clearest sign that this had stopped being a China story arrived when its elements began showing up in US tech startups through 2025 — the 996 framing in hiring, the open expectation of six-day weeks, the quiet assumption that evenings belonged to the company. And the pushback that followed is the line worth keeping.

Several European founders, asked about the trend, declined to treat it as visionary. They described it instead as a fetishization of overwork rather than smart work — a phrase that does a lot of honest labor in a few words. It draws the exact distinction the hype erases: between effort aimed at outcomes and effort aimed at the appearance of effort. A team working intensely on the right problems for a sane number of hours can beat a team grinding 72-hour weeks on the wrong ones. Duration is an input; output is what you were supposedly optimizing for, and the two come apart faster than the all-nighter mythology admits — your tenth hour of debugging is not your first hour’s equal, and everyone who has shipped a regression at 2am knows it in their body.

None of this argues against working hard. It argues against confusing the clock with the contribution. The fetish is the belief that the hours themselves are the virtue — that visible exhaustion is a moral signal — and that belief is where the damage starts.

The reality check: overwork is a measurable killer

Here is the part the celebratory posts never include, and it is not soft or vibes-based. It is mortality data.

In a joint study, the World Health Organization and the International Labour Organization estimated that long working hours caused roughly 745,000 deaths in a single year, 2016, from heart disease and stroke — split as something like 398,000 stroke deaths and 347,000 from ischemic heart disease. That made long hours, by their accounting, one of the largest occupational health hazards on the planet. And the trend was the wrong direction: those deaths represented about a 29 percent rise since 2000, even as many other workplace risks were falling.

The mechanism is the dose-response curve the founders skip. The same body of research found that working 55 or more hours a week is associated with roughly a 35 percent higher risk of stroke compared with a standard 35-to-40-hour week. Now hold that next to 996: a 72-hour week is not flirting with the 55-hour danger threshold, it is sitting far past it, week after week. The schedule being lionized as the cost of greatness is, in the cleanest occupational-health terms available, a schedule that raises your odds of a stroke and quietly shows up in mortality tables.

Treat those figures as directional rather than surgical — they are modeled estimates across whole populations, with all the assumptions that implies, and the report’s authors say as much. But the direction is not in doubt, and it does not point anywhere good. When the headline number is hundreds of thousands of deaths a year and the risk curve bends upward exactly where 996 lives, “directional” is more than enough to take seriously.

Hours diverge wildly by country — and so do the outcomes

One reason overwork gets normalized is that people compare themselves only to whoever is grinding hardest in their own bubble. Widen the frame and the supposed universality of the long week falls apart. Working hours diverge enormously across countries, which means none of this is a law of nature; it is a choice some labor cultures have made and others have refused.

The numbers behind that chart are worth saying plainly. ILO data for 2024 put American workers near 1,976 hours a year and still climbing — divide that across a working year and you land around 38 hours a week, already on the higher side for a rich economy. At the other end, the Netherlands sits lowest among tracked countries, around 31.6 hours a week, a culture that has institutionalized part-time and protected-evening norms without falling into the sea. Between those poles, South Korea is a useful cautionary case: even after reforms, roughly a quarter of its employees work 50 or more hours a week, against an OECD average closer to one in nine — a reminder of how heavy the long-hours tail can get before a country decides to pull it in.

Set 996 against any of these and the absurdity is geometric, not rhetorical. A 72-hour week is nearly twice the American average and more than double the Dutch one. The grind being sold as the frontier of ambition is, by the numbers, an extreme outlier that most of the developed world has spent decades legislating and culturally negotiating its way down from.

India and the lionizing of the 70-to-90-hour week

The clearest recent example of overwork being talked up as virtue did not come from a startup ad at all. In India, several prominent business leaders publicly championed 70-hour and even 90-hour workweeks as what dedication ought to look like — framing the long week not as a regrettable necessity but as an aspiration, the thing committed people should want.

Indian workplace researchers have flagged exactly this as a cultural driver of strain: the public glorification of those hours by figures with reach normalizes overwork as a proxy for loyalty and blurs the line between work and life until it stops existing. One survey found only about one in ten Indian corporate employees has access to professional mental-health care — so the cultural push toward more hours arrives in a system with little to catch the people it breaks. The point is not to single out one country; it is that when admired leaders make a 90-hour week sound noble, they are broadcasting the fetish at scale, to an audience that will pay the cardiovascular bill.

The honest distinction: hard work is not the fetish

So where does this leave someone who genuinely is ambitious, who does want to build something, who is not looking for permission to coast? Exactly where the European founders’ phrase pointed: at the difference between hard work and a fetish for long hours.

Hard work is doing the difficult, important, often uncomfortable thing the goal actually requires. It spikes sometimes — a launch week, an incident, a real deadline will cost you evenings, and that is normal and not what anyone is warning about. The fetish is different: treating duration as the point, performing exhaustion as proof of seriousness, and sustaining 72-hour weeks as a permanent identity rather than a rare sprint. The first is how things get built. The second is how people get strokes, and the mortality data does not care which story you told yourself about why the hours were noble.

For a data or engineering career specifically, the smart-work framing is not a slogan — it is the actual job. The work that compounds is the well-chosen problem, the design that prevents three weeks of firefighting, the test that catches the regression before it ships, the judgment about what not to build. Almost none of that is improved by the sixtieth hour of a week; most of it gets worse, because tired people make the expensive mistakes. The founders lionizing 996 are, in effect, optimizing the one input — raw hours — that the evidence says is both the least reliable predictor of good output and the most reliable predictor of cardiovascular harm.

The reframe to carry out of here is small and stubborn. When someone tells you long hours are the price of mattering, ask what they are actually counting. Real dedication shows up in the quality and difficulty of the work and in the willingness to sprint when it genuinely counts — not in a permanent 72-hour schedule that the country that invented it made illegal, that European builders call a fetish, and that the WHO can connect, in plain numbers, to hundreds of thousands of deaths a year. Work hard. Just refuse to confuse the clock with the contribution, because your heart is keeping a more honest count than your feed is.