The New Normal: Silicon Valley's Descent into the 996 Grind

The era of "rest and vest"—a time when Silicon Valley software engineers could expect six-figure salaries, ping-pong tables, and work-life balance—appears to be officially over. In its place, a grueling new standard has emerged, imported directly from the hyper-competitive tech hubs of Shenzhen and Beijing: the "996" work culture.

As of February 2026, the artificial intelligence sector has fully embraced this controversial schedule, which demands employees work from 9:00 a.m. to 9:00 p.m., six days a week. What was once considered a violation of labor norms in the West is now being touted by founders and venture capitalists as the necessary price of admission for the AI revolution. With the race to Artificial General Intelligence (AGI) reaching fever pitch, companies like OpenAI, Anthropic, and a slew of aggressive new startups are effectively operating on a wartime footing, fundamentally reshaping the social contract of the American technology industry.

The "Speedrun" Mentality

The driving force behind this shift is the palpable fear of being left behind. The development cycles for Large Language Models (LLMs) and reasoning agents have compressed from years to mere months. In this environment, speed is the only metric that matters.

"We are basically trying to speedrun 20 years of scientific progress in two years," noted a research scientist at Anthropic, capturing the prevailing mood. The sentiment is widespread across the Bay Area. Engineering teams are no longer sprinting towards a quarterly release; they are running an endless marathon at sprint pace. The consensus among leadership is that the first entity to achieve AGI will capture the vast majority of the economic value, rendering second-place finishers irrelevant.

This winner-takes-all dynamic has emboldened executives to demand total devotion. The "hardcore" culture famously reinstated by Elon Musk at X (formerly Twitter) in the early 2020s has metastasized across the AI sector. It is no longer an anomaly; it is the default operating system for high-growth AI firms.

Institutionalizing Overwork: From Rilla to OpenAI

While some companies attempt to cloak these expectations in euphemisms like "mission-driven" or "high-performance," others are brazenly transparent. Rilla, a fast-growing San Francisco AI startup, has made headlines for explicitly filtering candidates based on their willingness to work 70+ hour weeks. Will Gao, Rilla's Head of Growth, has publicly compared the ideal employee to an "Olympian athlete," seeking individuals possessed by "obsession" and "infinite ambition."

Job postings for these firms now frequently flaunt the grind as a filter to weed out the uncommitted. But even at established giants, the pressure is crushing. Reports from inside OpenAI indicate that despite the company's massive valuation and resources, staff have regularly clocked 80-hour weeks to meet release deadlines for the latest iterations of their models.

The irony of this situation is stark. Just as China’s Supreme People’s Court declared the 996 schedule illegal in 2021 due to its severe social and health costs, American innovation hubs are adopting it with zeal. The table below illustrates the dramatic shift in expectations for the average Silicon Valley software engineer over the last decade.

Table: The Shift in Silicon Valley Work Culture (2016 vs. 2026)

The Human Cost of the AI Gold Rush

The human toll of this shift is already becoming visible. Burnout rates among AI researchers and engineers have skyrocketed. "Sarah," a 28-year-old engineer at a prominent San Francisco AI unicorn, described a life that has shrunk to the dimensions of her laptop screen. "I wake up, check the training runs, go to the office, work until 10 p.m., and repeat," she shared under the condition of anonymity. "My partner handles everything—groceries, cleaning, bills. I am essentially a guest in my own life."

This intensity is also exacerbating the industry's diversity crisis. The 996 model structurally excludes anyone with caregiving responsibilities, disproportionately affecting women and older workers. Sociologists warn that this regression could undo decades of progress in making the tech sector more inclusive.

Even the companies driving this trend are forced to acknowledge the damage. OpenAI and others have had to institute mandatory "shutdowns" or "wellness weeks"—not merely as a perk, but as a tactical necessity to prevent mass resignations and medical leaves. These breaks are stopgap measures, designed to keep the machine running without burning out the parts entirely.

Economic Drivers and Venture Capital Pressure

The pressure is not just internal; it is being applied top-down by the venture capital community. With interest rates stabilizing but capital still expensive, investors demand immediate, exponential returns. The narrative sold to Limited Partners is that AI is a once-in-a-century transformation, and any startup not working nights and weekends is wasting capital.

"If you are not sleeping under your desk, you are not serious about AGI," one prominent VC reportedly told portfolio founders. This rhetoric creates a coercive environment where founders feel they must enforce 996 hours to secure follow-on funding. The result is a race to the bottom in terms of labor standards, where the "hustle" is commodified and sold as a product feature.

Sustainability and the Future

As we move deeper into 2026, the question remains: is this sustainable? History suggests that crunch periods yield diminishing returns over time, leading to code debt, strategic errors, and a hollowed-out workforce. However, the AI industry is betting that the technology itself will soon alleviate the burden—that AI agents will eventually take over the coding and research tasks, allowing humans to step back.

Until that theoretical singularity arrives, however, the humans building the machine are being ground down by its gears. For the foreseeable future, the lights in San Francisco's office towers will stay on well past midnight, illuminating a workforce pushed to its absolute physiological limits.