Between 2001 and 2003, Santa Clara County shed close to a fifth of its jobs. Whole floors of Sunnyvale office parks went dark and stayed dark for years. The county's median home price gave back single digits and made most of it back inside two years.
I keep that number close, because the story now being told about Silicon Valley real estate runs the other direction: layoffs arrive, prices follow. As of writing, the region posted the weakest home-price growth among the 40 largest U.S. markets — Homes.com data reported through CoStar — on a median sitting near $1.6 million. Meta had cut roughly 3,000 roles. The two facts sat next to each other in the coverage, and the causal arrow drew itself.
The verdict, in one sentence: layoffs are a coincident indicator in this region, not a leading one, and what is actually softening Peninsula housing demand is a slowdown in new arrivals that began before the headcount cuts and will outlast them.
The belief has a birthday
Every region carries a folk model of itself. Detroit's was that the plants set the floor. Houston's was that the rig count sets the ceiling. The Bay Area's is that tech hiring is the housing market — that you can read Santa Clara County by reading layoffs.tracker headlines.
Ask where that model came from and almost everyone lands on the same place: 2001. And something did break in 2001. It was rents.
San Francisco rents fell by roughly a third from the 2000 peak, depending on whose index you trust. Landlords ran two-months-free concessions. That collapse was fast, legible, and it happened in the same eighteen months as the job losses, which is why it became the founding memory of the whole belief.
The for-sale market did not do that. It dipped, absorbed a large drop in mortgage rates, and moved on.
So the evidence underneath "tech jobs move Bay Area housing" is one episode, in one segment of the market, running through one mechanism: young workers who had arrived for the boom left or doubled up, and the number of households fell. Rents reprice the marginal household month to month. A four-bedroom in Cupertino reprices through a thin-inventory auction among equity-rich buyers holding a rate lock. Those are two different machines. The belief was built on the first and has been applied to the second for twenty-five years.
Three tests, and the belief did poorly on all of them
2001–2003. Employment collapse, single-digit price dip, full recovery in about two years. Rates did most of the work.
2008–2011. Stockton and parts of Sacramento lost more than half their value. Santa Clara County lost a fraction of that. The crash was a credit event, and it hit hardest where the subprime paper was, not where the layoffs were.
2020–2022. "San Francisco is finished" was consensus for about a year. Rents fell roughly a quarter and then recovered inside about two years. The people who sold on the layoff-and-exodus thesis got the direction and the duration wrong at the same time.
Three tests. In each one, the layoff count told you less than the flow of people did.
The variable nobody puts on a dashboard
Household formation. How many new households form here in a year, and where they come from.
Foreign-born residents make up roughly 40% of Santa Clara County — one of the highest shares of any large county in the country. They arrive on H-1Bs, on OPT after a master's, on L-1 transfers, on family petitions. They rent first, in Santa Clara and Milpitas and Sunnyvale, and a meaningful share of them buy five to eight years later.
Which means the pipeline has a lag welded into it. Reduce arrivals and nothing happens to prices this quarter, or next year. The people who landed in 2019 are buying now. The people who never landed in 2024 do not buy in 2031, and no one writes that story, because a purchase that never occurs generates no data point.
I am not an economist. I sent 419 applications over eleven months and got four callbacks, and what that year gave me was a close view of who could wait it out and who was on a clock. Two friends went home. Not because they lost an argument about their skills — because a lottery went against them. Neither was ever going to buy in Sunnyvale now. Neither of them appears in a single layoff tracker.
San Francisco is the control group
If this were the sector dying, San Francisco would go first. It has more concentration, thinner diversification, and a decade of people predicting its funeral.
Instead the city has held up better than the valley south of it, because AI capital concentrated in the city rather than along 101. Same industry, two outcomes, forty miles apart.
That divergence is the most useful thing in the whole data set. It rules out "tech is over" cleanly. Whatever is pressing on valley home values is geographic and demographic, not sectoral.
How I would actually read the next few years
Rents before prices. Rents mark the marginal household within months. Prices lag the same signal by years. Asking-rent trends in Santa Clara and Fremont are the earliest honest read available.
Arrivals, not exits. Visa issuance volumes, international graduate enrollment at San Jose State and Berkeley and Stanford, new-lease formation. These are leading. Layoff counts are not.
The entry tier, specifically. Days on market for Peninsula condos and townhomes in the roughly $1.1M–1.6M band. That is where the first immigrant purchase happens. It softens before the $3M single-family tier does, and it tells you something the top tier cannot.
Sector geography over sector headlines. "Tech layoffs" is not a variable. Which campus, which city, which tax base.
Layoff counts last. Loud, coincident, and mostly noise for housing.
Who this changes something for, and who it doesn't
If you own a Peninsula property and intended to hold it a decade, the cyclical-versus-structural question does not change your action. It changes your rent-growth assumption, which is a smaller and more boring adjustment than the headlines suggest.
If you are underwriting new entry-tier inventory on the premise of a self-replenishing supply of first-time buyers, that premise is the load-bearing wall of your model, and it is the one worth stress-testing at half strength.
If you write policy, the honest note is that you do not control layoffs and never did. You have some influence over housing supply and almost none over federal immigration throughput, which is the actual driver here — an uncomfortable division of labor, and pretending otherwise costs years.
And the honest negative on my own read: structural forces move slowly enough that I cannot prove this from two years of price data, and I won't pretend I can. If arrivals return to trend within a few years, this looks like an ordinary cycle in hindsight and I will have overweighted the demographics. I would rather be early on the right variable than precise on the wrong one.
The myth is that Silicon Valley housing is soft because tech is cutting jobs.
The more accurate version is that it is soft because fewer people are arriving to become the next decade's households, and the layoffs are only the loudest thing in a room that has gone quiet.
