I left San Francisco. I tried remote. I worked from cheaper cities with more space and better weather. Then I moved back.
The Density Problem
Remote work is excellent for execution -- shipping features, fixing bugs, running systems. It is terrible for the phase before execution: figuring out what to build, finding co-founders, raising capital, and building the network that compounds over a career.
In most cities, working 80-hour weeks on an obscure technical problem makes you unusual. In San Francisco, you are the median. This density produces serendipity that Zoom cannot replicate -- the impromptu conversation that becomes a collaboration, the dinner where you meet your next co-founder, the ambient pressure to ship.
The Financial Reality
San Francisco is expensive. Brutally. But the cost framing is wrong.
The relevant comparison is not "rent in SF versus rent in Austin." It is the long-term career delta: the network effects, the access to capital, the density of hiring pools, the speed of information flow. At a certain career stage, the cost of not being here exceeds the cost of being here.
That calculation does not apply to everyone. It does not apply at every career stage. But if you are building a venture-scale technology company, the math tilts sharply toward presence.
What Remote Gets Right
Remote-first companies solved real problems: timezone flexibility, access to global talent pools, reduced commute waste. These gains are real and I do not dismiss them.
But they optimized for individual productivity at the expense of collective discovery. The hardest problems in building a company -- finding product-market fit, closing a fundraise, hiring a founding team -- are high-bandwidth, trust-intensive activities that degrade over distance.
The Observation
Every major AI company, every breakthrough startup cluster, every transformative technology effort of the last five years has concentrated in San Francisco. Not distributed across remote-friendly cities. Concentrated.
The remote advocates will continue arguing that geography is irrelevant. Meanwhile, the empirical evidence keeps pointing in one direction.
That is not a coincidence. It is the answer.