Citrini co-author Alap Shah: White-collar jobs are down 8% from 2022 peak — and corporate AI adoption hasn't even started

Summary

White-collar employment has fallen 8% from its 2022 peak, and Alap Shah, co-founder of Sentieo and co-author of the Citrini report on AI's labor impact, expects the decline to accelerate. Corporate AI adoption has not yet begun at scale, but when it does, job displacement will flow through the economy in ways most institutions are unprepared for.

Shah's concern is structural. White-collar workers drive discretionary spending, fund mortgages and consumer credit, and generate tax revenue for health care and education. If AI displaces 4–5% of white-collar jobs annually, the effects cascade upstream: lower payroll tax receipts cut government spending, reduced consumer demand weighs on asset prices, and loans underwriting the system lose collateral.

Information-sector jobs are already down 8% from 2022 despite minimal corporate adoption. Shah distinguishes between announced layoffs framed as efficiency and a deeper shift driven by agentic AI. Agents reduce friction to near-zero, making 1990s predictions about frictionless commerce finally feasible. Within six months, Shah expects agents to autonomously purchase goods on your behalf using your credit card, turning commerce into an API call.

DoorDash and network effects

Network effects and supply aggregation will erode as competitive moats. DoorDash's value rests on customer lock-in through saved payment methods and habit. Agents optimize for price and delivery time, not loyalty. When Gemini and ChatGPT become the customer acquisition channel, new entrants need only supply-side logistics, not billion-dollar driver marketing budgets. Margins compress as the platform and the agent each take a cut. Shah expects the vig to shrink below today's 15%.

Hosts pressed on supply. How do you onboard drivers at scale without massive marketing spend? Shah's answer bypassed the constraint. Smaller driver networks and third-party logistics services already exist and will gain share as friction drops.

Geography and timing

Blue-collar jobs are healthier than white-collar jobs today, but Shah treats the labor market as unified. Displaced white-collar workers without jobs in their sector will flow into gig and blue-collar work, pressuring wages across the economy. India and the Philippines show no unemployment spike yet because white-collar work is a smaller share of their economies. But consulting outsourcers in those countries face pressure. Enterprises with $1B+ budgets get white-glove deployment from OpenAI. Those with $10M budgets turn to offshore providers. The outsourcing sector will initially boom as companies adopt AI, then collapse as the technology matures.

The political case

Critics argued the piece echoes Marx—that capitalism devours itself when competition diffuses productivity gains, workers lose purchasing power, and demand collapses. Shah does not reject the parallel. He argues the contradiction is fixable through policy: redistribute via taxation so GDP growth benefits broadly, not just capital. Without redistribution, economic crisis slows AI progress anyway. It's not ideology; it's math.

Shah expects Anthropic to go public within three to six months, shifting investor focus from incumbent tech giants to AI labs themselves. Google remains best-positioned due to customer lock-in and ability to finance inference losses. AI labs could be the largest winners if solutions are structured correctly.

What follows

Leisure industries will boom if labor displacement is managed. Reindustrialization and hardware face the same AI-driven efficiency pressures as software. High-agency people using tools will do more work, not necessarily create more jobs. Shah plans to release a solutions piece within days.

The core tension: Shah presents a plausible chain from AI efficiency to systemic financial stress, backed by current labor data and grounded in how agents change commerce. The transcript reveals real friction in the argument, especially around supply-side aggregation and whether AI actually collapses customer lock-in as completely as Shah claims. Hosts credibly pushed back on whether a 15,000-person town will have enough driver supply to make a startup delivery app viable. Shah's response leaned on time and competitive benefits rather than defending the mechanism itself.