AI-мания ломает корпоративные решения: стратегии на миллиарды пишут без ChatGPT

Consultant Nik Suresh's essay "AI Mania Is Eviscerating Global Decision-Making," about how the hype around artificial intelligence is destroying decision-making at large companies, went viral on Hacker News and reached developer Simon Willison on July 19, 2026. Suresh's central thesis: billion-dollar AI strategies are often written by executives who have never once opened ChatGPT themselves.

A $2 Billion Strategy Without ChatGPT

Nik Suresh, who consults for large corporations, cites an extreme case: an executive admitted he had never in his life used ChatGPT or any other AI tool — right after releasing a technical strategy, built entirely around AI, for an organization with revenue exceeding $2 billion. The essay is drawn from anonymous stories that engineers and managers inside "overhyped" companies tell the author.

"In one extreme case, an executive admitted that he had never in his life used

ChatGPT or any AI tool — right after preparing a technical strategy for an organization with revenue exceeding $2 billion, built entirely around artificial intelligence." — Nik Suresh, essay "AI Mania Is Eviscerating Global Decision-Making"

Key stories cited by Suresh:

• An AI strategy for $2 billion+ in revenue, written by a person without a single query to a model
• "Token leaderboards" — metrics that encourage employees to spend more requests on AI
• Client claims of "100x" productivity gains that no one dares to challenge
• Engineers running AI on knowingly meaningless tasks just to check a box

Why No One Argues With the Hype

The silence rests on fear of losing a contract, not just on marketing. Suresh asked a skeptical vendor executive why absurd promises get repeated without objection. The answer turned out not to be about "sales noise": executives on the client side themselves had publicly claimed 100x productivity gains.

"If someone from the vendor's side said such a gain was implausible,

it would undermine the client executive's authority, be seen as an attack — or heresy — and could lead to the termination of the corporate contract." — from Nik Suresh's essay

The logic is simple: torpedoing a corporate contract over a dispute that doesn't affect the company's mission is a fast way to get fired. That's why an honest assessment of AI's returns becomes unprofitable for everyone in the deal at once — both vendor and client.

What's Happening With Engineers

Hype pressure trickles down and turns into an imitation of work. One engineer described how, at a company with a "token leaderboard," he pulls a parallel copy of a working Go repository and orders AI to rewrite it entirely in the Zig language — while he himself works on another task — just to formally pad the metric and keep his job.

This scenario is a direct consequence of metrics that measure activity rather than results. When a KPI is tied to the number of tokens spent rather than to working code, employees rationally start burning resources for nothing: the system rewards the volume of requests to the model, not the value they produce.

What It Means

The problem isn't the models themselves, but how organizations make decisions under the pressure of a trend. When a $2 billion strategy is written without experience, criticism is punished with the loss of a contract, and metrics reward token spending — a company ends up optimizing for hype, not results. Suresh's essay is a reminder that an honest conversation about AI's real returns is still losing to the fear of looking like a skeptic.