The Survey Nobody in Silicon Valley Wants to Talk About

Mercer just dropped a number that should stop every working American cold.

According to Mercer's 2026 Global Talent Trends report — which surveyed 825 C-suite executives and 1,650 HR leaders — 99% of corporate executives are planning AI-related job cuts within the next two years. Not some. Not most. Ninety-nine percent.

Nearly as many — 98% — said they're also planning major organizational design changes in that same window. This isn't a fringe concern from a doomer blog. This is the people who sign your paycheck telling a major consulting firm what's coming.

ZeroHedge flagged the Mercer data, and to their credit, the numbers are real and sourced. The facts are alarming enough on their own.

Meanwhile, Cognition's Scott Wu Just Raised $1 Billion

This week, Cognition — the two-year-old startup behind AI coding agent Devin — closed a $1 billion funding round at a $26 billion valuation, according to TechCrunch.

CEO Scott Wu told TechCrunch, quote: "We've never thought about it as replacing humans. I know it's like a scenario, folks have said these things. It has never been our view."

Wu is a child prodigy turned founder — reportedly winning a nationwide math competition for seventh-graders as a second-grader, per a Colossus profile. He started coding at nine. He genuinely loves the craft.

Cognition's own announcement states that 89% of code committed by its engineers was committed by Devin. Not assisted by Devin. Committed by Devin. The humans are handling 11%.

Wu keeps a Devin teddy bear on his desk and calls it "my buddy that helps you build more." It's also a $26 billion company where the AI is doing nearly nine-tenths of the technical work.

The Token-vs-Headcount Trade-Off Is Real and Accelerating

CNBC got two enterprise AI CEOs to describe what's actually happening inside Fortune 500 boardrooms right now, and the picture is brutal.

Arvind Jain, CEO of enterprise AI company Glean, told CNBC: "The number one topic for every enterprise right now is overblown AI budgets." Companies, he said, are burning through annual AI budgets in one to two months.

AI costs are NOT going down the way buyers expected. According to Jain, each new frontier model release costs roughly twice as much per token as the one it replaced. That's the opposite of Moore's Law.

So what are CFOs doing? They're making the math work by cutting people instead.

"This is the first time ever that I can remember that technology costs the same as people, and you're making that comparison: choose tech or people," Jain told CNBC. "We've never had that conversation historically, because tech is a fraction of the overall cost of any operating business."

Matan Grinberg, CEO of Factory AI, told CNBC companies have moved through three distinct phases in roughly a year: first, boards demanding CEOs "do something about AI"; then "tokenmaxxing" — using AI by any means necessary regardless of cost; and now a hard-nosed resource allocation phase where leadership is choosing between employees and AI spend per employee.

This is no longer theoretical. It's a budget line item.

What the Coverage Is Missing

TechCrunch gave Wu a largely uncritical platform to spin a feel-good narrative about AI as a creative partner. They buried the 89% statistic. That number deserved a follow-up question, not a paragraph near the end.

CNBC's piece is solid on the macro economics but stays safely abstract — no names of specific companies making the cuts, no hard job-loss numbers, no workers quoted. It's enterprise-CEO-to-enterprise-CEO conversation that conveniently skips the people getting displaced.

ZeroHedge surfaces the Mercer data but wraps it in economic collapse language that makes readers tune out. The survey is legitimate. It doesn't need the theatrics.

The AI industry's public messaging and its operational behavior are in direct contradiction. The same CEOs attending conferences to say AI "augments" workers are signing off on org charts with 20% fewer humans in them.

Meta, according to ZeroHedge's sourcing, is cutting 1,400 highly paid workers in Washington state as part of its AI overhaul. Those aren't line workers. Those are engineers.

Just 32% of CEOs Think This Transition Goes Smoothly

The Mercer survey buried another telling number: only 32% of the CEOs surveyed believe their workforce can combine human and machine capabilities in an optimal way.

Translation: two-thirds of the executives planning these cuts don't actually believe the "humans and AI working together" story they're selling publicly.

They know what's coming. They're just not saying it out loud.

What This Means

Scott Wu is a genuinely talented guy who built a genuinely impressive product. His intentions may be sincere. But good intentions don't override what a $26 billion valuation built on autonomous code generation actually means for the labor market.

Ninety-nine percent of CEOs are planning cuts. AI costs are rising, NOT falling. CFOs are solving the math by reducing headcount.

The teddy bear on Wu's desk is cute. The Mercer survey is not.