aziz_sunderji

The future of AI is being built in one place. The Bay Area has more technical PhDs than any other metro, and the gap is growing. Eventually the cost of concentrating so much talent in such an expensive city will push the AI economy outward. Clearly, we're not there yet.

Bay area employees of OpenAI, Anthropic, and Perplexity are using secondary share sales to fund home purchases, fueling a rebound in the local housing market. In mid-2023, the San Jose metro area—home to Silicon Valley—ranked 3rd-worst out of the 100 largest US metros for year-over-year home price growth. In the second half of 2025, it ranked 20th. This map (see link) shows the hottest Bay area ZIP codes are those with the highest proportion of tech workers.

For one, my model allows for refinancing.

Weak home construction was hurting the economy last year—not anymore!

During the GFC real estate volatility was mainly driven by big cities. Not this time: even the sleepiest small towns got in on the action.

Buyers waiting for discounts in nominal terms could be frustrated—price falls measured this way are rare and short-lived.

Credit is the lifeblood of the economy. That's why the Fed changes the prices of credit (interest rates) to speed or slow growth. Sadly, it's not very useful for making investment decisions: credit data tends to be coincident with market moves. If only there was a way to predict credit growth. Well, there is! Yesterday's loan officer survey from the Fed tells us where credit, the economy, and markets are headed. (A 3 min read at https://sunderji.substack.com/p/banks-grow-less-willing-to-l...)

The argument for limited deposit insurance rests on an assumption that depositors can detect bad bank behavior. This is falsified by what happened at SVB. If depositors can't restrain bank risk-taking, there is no moral hazard in insuring them fully.

Bank stocks continue to fall, signaling problems with the financial system that run deeper than deposit flight at SVB and other regional banks.

A careful look at SVB's balance sheet shows why deposits were so flighty, and why bankruptcy was in the cards by the end of last year.

One hundred years ago, an American randomly plucked from a lineup of workers would as easily have been a farmer, as a factory worker, or a service-worker (say, a blacksmith). Today, there is an 80% chance that the random pluck would give you a service worker (likely a truck driver or a cashier, but also maybe a teacher, nurse, and slightly less probably, a lawyer or software developer).

The transition from farming and manufacturing to services explains so much about our world today—from growing inequality, to the soaring cost of education, to deepening political polarization.

I wrote about this today at sunderji.substack.com

His post about vaccinated vs unvaccinated death rates is a tour de force of statistical trickery.

I get so frustrated when I read piece-meal, overly specific explanations for the enormous cost of healthcare and education.

There is a much more compelling and systematic answer: Baumol's cost disease. I wrote a quick thing about it at sunderji.substack.com.

Curious to know whether this community agrees or not.

Folks, I wanted to share my experience during the financial crisis...I was a bond strategist at a major investment bank in London. I almost lost a tonne of money...by some insane luck I just escaped with a lesson in market efficiency.