The Sane Asylum
The author, who has worked with turbopuffer on marketing for six months, argues that a handful of profitable, non-VC-funded AI-native companies offer an alternative to the burnout of the venture-backed race. The piece is a close look at turbopuffer's culture as a model for building at scale without raising money or hiring heavily.
- Silicon Valley workers at frontier labs, AI startups, and pre-AI companies all describe anxiety, burnout, and relentless pressure to grow, with unicorns minted daily and 311 so far in 2026.
- A small group of companies, including Midjourney, Increase, and turbopuffer, are profitable, lean, and largely free of venture capital pressure, which the author sees as a healthier model.
- turbopuffer reports more than $100mn in annualized run rate and turned down VC interest, with CEO Simon Hørup Eskildsen saying "What do I need venture capital for? We've got customers!"
The author's core claim is that Silicon Valley has become psychologically unsustainable, and he describes the strain across three groups. Workers at frontier labs and decacorns, the so-called chosen ones, are missing family time, putting weddings and kids on hold, and dealing with sharp elbows and security warnings about wearing company swag. Startups competing with those firms face hiring and compute shortages and the pressure to hit a billion dollars in ARR or get cooked. Pre-AI companies growing 30% a year now feel like they are dying, with investors dropping them from homepages, teams bleeding staff, and friends in group chats suggesting they just sell. The author ties this to the pace itself: a new frontier model monthly, 311 new unicorns by day 300 of 2026, and Cursor going from zero to $100 million in a year. Everyone is working 996 or 007 schedules, and he calls the whole thing a Icarus-style burnout.
His alternative is a small set of companies he calls elite and profitable: Midjourney, Pierre Computer, Increase, and turbopuffer. He says they are AI-native, have stacked teams, raving customers, and avoid venture capital, and that their employees are proud to wear the company's swag. What he finds striking is that these firms seem indifferent to the decacorn-exit playbook and motivated by the work itself as much as the outcome. He points to historical precedents like Linear, Basecamp, Valve, Obsidian, and early GitHub, and argues that AI tooling now makes it possible to build a highly profitable company at scale without raising much money or hiring a lot of people. He also suggests that people who have already made money find this path easier to take.
The piece is a first-person essay from the author's own vantage point, and it is more a mood piece than a data-driven analysis. The evidence for the burnout is anecdotal, relying on overheard conversations and friends' reports, and the case for the profitable-alternative model rests on a handful of named examples without financial figures for most of them. Read it as a useful snapshot of how some people in the Valley are feeling and as a pitch for a different company model, but treat the claims about these companies' health and the broader trend toward them as unproven. The article also cuts off mid-sentence at the end, so the promised account of his cultural exploration is missing.