Extreme Work Intensity in AI Startups: A New Standard?

Patterns of extreme work commitment, such as the „996“ schedule (from 9 AM to 9 PM, 6 days a week), were once common in Chinese tech companies until they were officially banned in 2021. In Western countries, similarly exhausting working hours have long been rejected, and in Europe, they are even illegal, as excessive work demonstrably leads to burnout syndrome and other long-term health issues.

Despite this, a number of AI startups are resorting to similar practices. An example is Cognition, which expects its employees to work over 80 hours per week. CEO Scott Wu spoke about this demanding company culture without apology, stating that they make it clear during recruitment. Some teams work in the office on weekends and late into the night, and some employees even literally live at work.

This is not an isolated phenomenon. Lovable states „long hours, high pace“ in its job descriptions, citing an approaching AGI timeline. Replit's CEO posted photos of the team working in the office at midnight. An engineer from AI startup xAI stated that the team „grinded in the office day and night,“ and CodeRabbit was finishing work in San Francisco at 2:30 AM. The founder of Icon, a company focused on ad creation, shared that they only hire „top 0.01% engineers with no life“ and expect them to work 7 days a week. Even Sergey Brin, co-founder of Google, reportedly told employees in the AI division that „60 hours a week is optimal for productivity.“

This extreme commitment is often justified by the race to achieve Artificial General Intelligence (AGI) within months. Many AI professionals believe that achieving AGI will mark the „endgame“ for most companies in this segment, establishing a new status quo where AGI will be capable of self-improvement with more resources. Companies with AGI would then dominate the industry. This vision is a strong motivator to use all available means – including extreme employee work commitment – to rapidly achieve AGI. Although the author personally remains skeptical of such simplified predictions regarding AGI, it is clear that this is a driving force for many founders under commercial pressure.

It has been almost three years since the release of the conversational model, and there are still no signs of AGI, although large language models (LLMs) are constantly improving. The question remains whether exhausting work patterns, intended for just a few months, will not stay in effect for years and become the standard.

Requiring or even demanding long working hours leads to a long-term slowdown in work pace and to burnout syndrome. In many countries, regulations exist to establish reasonable working conditions, and unions advocate for them. Another reason why 80-hour weeks are not accepted elsewhere is that the negative impacts of long hours are soon apparent: employees take more sick leave, productivity decreases, and more employees leave for places where they work fewer hours. Furthermore, demanding long hours automatically excludes many strong potential candidates who have family responsibilities, live far from the office, or prefer personal activities outside of work.

Of course, if a company hires young professionals, burnout won't occur as quickly, and some individuals can work this way for years. Moreover, if they don't have a family or a rich social life, they might be satisfied with such demanding working hours.

Another strong motivation for AI startups to create a culture of long hours is the promise of generational wealth. For most people, the tempting prospect is that after 1-3 years of intense work, for example, 6 days a week and 80+ hours, they could leave with a reward in the range of $10 million. The incredible growth in the AI sector makes such a scenario achievable for many. Windsurf serves as an example, where just 10 months after launching their IDE, 40 team employees were acquired by Google. The founders likely made hundreds of millions, and some engineers could have received rewards exceeding $10 million. For founders, who can earn many times higher sums, such work commitment and pressure on the team are entirely worth it.

Based on this, we can expect founders to continue pushing employees to spend all their waking hours working or thinking about work. As long as the promise of great success with AI startups exists, these working hours will likely persist. Founders in AI startups face pressure to quickly bring products to market so they aren't outpaced by competitors. Speed is key to winning in AI, and the most obvious way to get things done faster is to make people work more.

However, long hours alone do not guarantee success, and there is evidence for this. While Cognition boasts its „extreme performance culture,“ its company culture has not led to commercial success: the Devin tool was among the least mentioned AI tools in surveys. Another sign that exhausting hours do not automatically generate success is the company Icon. Despite requiring seven-day work weeks from employees, the company appears to be transforming into a typical advertising agency, which are good businesses, but rarely fast-growing, multi-billion dollar projects.

It can be expected that the situation will return to normal in a few years, once every tech company also becomes an „AI company“ and it's a common practice. For now, however, the Fear of Missing Out (FOMO) across the industry is so strong that very long work weeks will likely spread. Everyone's situation is different, and it is up to the individual to consider whether they want or need to work extreme hours. However, AI startups will intensify pressure to weed out applicants who resist very long work weeks. Taking such a position can be an amazing career boost; yes, long hours have many drawbacks, but they also allow for better alignment with colleagues, and relationships forged under pressure can last decades. Of course, all the stress can also lead to burnout and health problems. It is important to realize that this trend is real and likely to persist. If your workplace applies a „more normal“ working regime, it's worth knowing that this is not necessarily a given throughout the entire technology industry.

Long hours are not new in technology and can also be about the enthusiasm for new technologies. A retired software engineer, after reading a similar summary, reminisced about how they did almost the same thing 50 years ago: „Thanks for that short trip down memory lane. Of course, this current generation works exhausting hours; this technology is new and exciting. My generation did the same with computers in the 70s... Storing crates of beer in the cooling system under raised floor tiles. Eating burgers at 7:00 AM after an all-night work session following system crashes. I suppose it's all new to you youngsters!“