On June 19, shares of Accenture, the world's largest IT consulting firm, crashed 18% in a single day.

Market capitalization shrank from $200 billion to under $80 billion; $120 billion evaporated overnight. The trigger was the CFO lowering full-year revenue guidance on the earnings call -- but what really made Wall Street vote with its feet was the subtext behind those words: AI Agents are completely upending the consulting industry's "person-day billing" model.

The Funeral of the Person-Day Model

The consulting industry's business model is brutally simple: you have a problem, I send in a team (partner + manager + consultant + analyst), billing per person per day. An SAP implementation project: 50 people working 18 months, the invoice full of "senior consultant 8000/day, manager 12000/day, partner 30000/day."

This model ran for a hundred years.

But Agents don't play that game. A configured Agent cluster can work 24x7 without travel expenses, overtime pay, or team-building events. More fatally, it bills by "task completion," not by "time invested." When a process optimization project that used to take 20 consultants three months can now be delivered by 3 engineers plus a fleet of Agents in 2 weeks, who's still willing to pay by the person-day?

Oracle's 47 to 3

Accenture isn't the first, and won't be the last.

Oracle disclosed in its latest earnings: 21,000 layoffs in the past fiscal year, 13% of headcount, with scaled AI deployment as the core reason. One widely cited internal data point: a team of 47 database administrators was reduced to just 3 people for oversight after AI went live. 47 to 3 -- that's not efficiency improvement; that's job evaporation.

Salesforce, Meta, Microsoft, Google... tech industry layoffs in 2026 have totaled nearly 120,000 people; almost every company's internal memo contains the same phrase: "AI restructuring." Amazon CEO Andy Jassy put it more bluntly in an internal memo: generative AI will "substantially reduce" white-collar jobs in the coming years.

This layoff wave is different from previous ones. The 2001 dot-com bubble burst cut overexpansion; the 2008 financial crisis cut cyclical contraction -- but this time, it's cutting jobs permanently replaced by technology. These jobs won't come back when the economy recovers, because Agents have permanently occupied them.

Bidirectional Talent Flow

But the other side of the coin is that top AI talent prices are being pushed to absurd heights.

Transformer architecture co-founder Noam Shazeer moved from Google to OpenAI -- Google had just recruited him back in 2024 for a staggering $2.7 billion. Sam Altman publicly accused Meta of trying to poach OpenAI core engineers with $100 million signing bonuses; Meta responded by acquiring 49% of Scale AI for $14 billion.

On one side, ordinary white-collar positions are being compressed at a 10:1 ratio; on the other, top AI talent valuations inflate at 10x speed. The K-shaped divergence plays out vividly in the job market: those who understand AI earn tens of millions annually; those who don't face re-employment.

This Isn't the End, Just the Beginning

The Accenture crash is just one signal. It tells us: AI's impact on the job market has spread from "manufacturing blue-collar" to "knowledge-economy white-collar," and penetrated from "entry-level positions" to "senior positions."

Consulting, law, accounting, auditing, financial analysis, code development, customer service, content creation -- these knowledge-intensive industries once thought "hard for AI to replace" are being restructured by Agents one by one. If not tomorrow, then today.

Jensen Huang only said half of it. The complete version should be: you won't be replaced by AI, but you will be replaced by people who use AI -- and if people who use AI only need one-tenth the original number, you'd better make sure you're that tenth, or the person commanding that tenth.

See you tomorrow.