Invest1 publisher2 min readPublished
San Francisco and San Mateo have shed 64,400 professional and information jobs since 2022
Payrolls in both sectors are back at their 2019 levels. The two AI companies headquartered in San Francisco are asking markets for roughly $11 million for every job left in the sectors where their hiring lands.
The Investor · Invest desk
What happened
- Professional, scientific and technical services payrolls in San Francisco and San Mateo County fell to 198,800 in August, the lowest since March 2019 and below the lockdown lows, per BLS data.
- By Wolf Street's count the sector has shed 35,500 jobs since its June 2022 peak, which the publication puts at 13.2% of its employment.
- Information employment in the two counties fell to 102,530, the lowest since October 2019, erasing the pandemic hiring boom that had peaked in August 2022.
- Losses in financial activities accelerated, taking employment in that sector to its lowest level since 2014.
Compiled by The InvestorSomething wrong?How this is made
Why it matters
- constraint The survey assigns each job to a business location regardless of where the worker lives, so it cannot identify who rents or buys in the two counties. That weakens the housing inference built on it.
- contradiction With no AI industry code, the hiring meant to offset these losses is unmeasurable in this data, so the boom case and the bust case rest on the same unknown quantity.
- cost Information carries 4.6 times the local payroll weight it carries nationally, so a sector-wide contraction is paid for by two counties' wage and tax base.
Divide the 198,800 jobs by the 17.6% of area payrolls they represent and the two counties come to about 1.13 million jobs [1][3][1]. Professional services and information together were 301,330 of those in August [2]. Anthropic's $2 trillion IPO target plus the midpoint of OpenAI's $1.2 trillion to $1.5 trillion range is $3.35 trillion [12][3]. That is about $11.1 million for every job in the two sectors where, Richter writes, AI employment is concentrated [14][4].
The professional-services figures in the Wolf Street account do not quite reconcile. A loss of 35,500 from a peak that leaves 198,800 puts June 2022 at 234,300 and the drawdown at 15.1%, wider than the 13.2% given [1][2][5]. The information numbers do reconcile: 102,530 plus 28,900 is 131,430, and 28,900 of that is 22% [5][6][6].
Richter wrote that the local economy and housing market "have become immensely dependent on unsustainable AI magic-money" [16]. The series he cites counts jobs by business location. A worker who commutes 60 miles to an office in San Francisco counts as a San Francisco job, and so does a remote employee assigned to a San Francisco office who lives far away [15].
Pay is what connects a $2 trillion valuation to a rent roll, and the payroll release counts employment while leaving wages out [15]. Richter describes OpenAI and Anthropic as having "rewarded their employees with huge salaries and stock compensation packages whose values have been rocketing to the moon" [17]. That pay, spread across a workforce back at 2019 size, bids for the same houses at higher prices.
If the two sectors track the dotcom drawdowns Richter cites, 28% for professional services and 46% for information, another 30,100 and 31,600 jobs go from the August levels, about 61,700 on top of the 64,400 already gone [4][7][7][8][9][10]. Information is 9.2% of payrolls in the two counties against roughly 2% nationally, about 4.6 times the national share [8][11]. If the wage bill has instead held up while headcount fell, rents stay firm and the job count keeps understating local income. Financial activities, at its lowest employment since 2014, sits outside the AI story in either direction [9].
On this evidence I don't think the boom is paper thin. The job counts show a 2019-size workforce; they cannot show 2019-size pay, and that difference is the whole housing question. Total covered wages for the two counties would settle it. Where the growth did land is visible: health care and private education, the only major private-sector category going from record to record [11], and leisure and hospitality, rising briskly and still far below pre-pandemic [10].
What to watch
- Quarterly covered-wage data for San Francisco and San Mateo County would show whether pay per head rose while headcount fell.
- Whether Anthropic prices an IPO anywhere near its $2 trillion target, converting stock compensation into cash that can clear escrow.
- Whether the September payroll print takes Information below 102,530 and professional services below 198,800.