← EasyTool.me

Mitchell Hashimoto: "Entire Companies Under AI Psychosis" — The HN Post That Struck a Nerve

Published: 2026-05-16 Reading: 6 min AI / Software Engineering

On May 15, 2026, Mitchell Hashimoto — the founder of HashiCorp and creator of Vagrant, Terraform, and more recently the Ghostty terminal emulator — posted a tweet that exploded across the tech community:

"I believe there are entire companies right now under AI psychosis."

Within hours, the post hit the top of Hacker News with 1,359 points and 669 comments — making it one of the most-discussed HN threads of the month. The reaction was immediate, visceral, and deeply divided.

What Is "AI Psychosis"?

Hashimoto's phrase captures something that many developers have felt but struggled to articulate: the idea that some companies have become so absorbed in the AI narrative that their decision-making has become detached from reality. It's not about whether AI tools work — it's about organizations that have lost the ability to critically evaluate them.

Symptoms of "AI psychosis" in a company might include:

The HN Comment Section: A Microcosm of the Industry

The 669-comment thread became a fascinating case study in itself. Several distinct camps emerged:

Camp 1: "I Haven't Written Code Since February"

One highly-upvoted comment described working at a big tech company where all microservices use a standardized stack in a monorepo. The commenter claimed they haven't written a single line of code since February 2026, and that incident rates haven't increased while feature output has gone up.

They described their workflow as "comment-driven development" — understanding the codebase, writing detailed comments explaining what needs to happen, then letting AI agents fill in the implementation.

But the replies were brutal. Other users pointed out that the same commenter had previously said their company considers Git "outdated" now that they have agentic coding, and that they don't even write their own commit messages. One reply simply said: "This is next-level trolling, or a serious case of AI psychosis."

Camp 2: The Bot Accusations

Perhaps the most unsettling part of the thread was the accusation that some pro-AI comments were astroturfing by bots. One user with an account from 2008 wrote:

"Vividfrier is a bot. You can see in many threads that if the general opinion does not go the way of AI companies, a completely outrageous pro-AI comment appears and is voted to the top, so that casual readers are tricked into thinking that the fake comment represents the general opinion."

Whether or not this specific accusation is true, it points to a real problem: the discourse around AI in software engineering is now so heavily manipulated that it's becoming hard to trust any opinion — positive or negative.

Camp 3: The Skeptics

Many commenters pushed back on the "AI is mature" narrative, asking a simple question: where are the public examples? If so many companies are successfully shipping AI-written code, why can't anyone point to a public case study beyond a handful of examples (Bun's rewrite, Ladybird's rewrite)?

This is a fair point. The gap between what people claim in anonymous comments and what's verifiable in public is enormous.

Camp 4: The Pragmatists

Some commenters tried to find middle ground, noting that AI tools are genuinely useful for certain tasks (boilerplate, tests, documentation, translation) but that the leap from "useful tool" to "replacing software engineering" is massive and unproven.

Why This Post Resonated So Deeply

Several factors made this particular post explode:

  1. Who said it: Mitchell Hashimoto isn't some random critic. He's a legendary infrastructure engineer who has spent decades building tools used by millions. When he says something is wrong, people listen.
  2. The timing: May 2026 is peak AI hype cycle. Companies are announcing AI-first strategies, laying off engineers, and promising 10x productivity. The backlash has been building for months.
  3. The language: "AI psychosis" is memorable and precise. It names a phenomenon that many people recognized immediately.
  4. The bot problem: The thread itself became evidence of the problem it was discussing — the possibility that some of the pro-AI comments were themselves AI-generated or bot-promoted.

The Real Question: Are We Measuring the Right Things?

Beneath all the drama, there's a genuine technical question that the industry hasn't answered:

What does "AI-written code is working" actually mean?

The companies claiming success typically point to:

But they rarely measure:

These are hard things to measure, and the tech industry has never been great at long-term thinking. The companies that are "doing fine" with AI code today may be setting themselves up for a reckoning in 18 months.

What This Means for Developers

If you're a developer navigating this landscape, here are some takeaways:

The Bigger Picture

Hashimoto's tweet isn't just about AI. It's about the gap between narrative and reality in tech — something that happens in every hype cycle. We saw it with the dot-com bubble, with blockchain, with microservices, and now with AI.

The pattern is always the same:

  1. New technology shows genuine promise
  2. Hype builds beyond what the technology can deliver
  3. Companies adopt it for FOMO, not because it solves their specific problems
  4. Critics are silenced as "not getting it"
  5. Reality eventually reasserts itself

We're somewhere between steps 3 and 4 right now. The question is how long step 5 takes — and how much damage is done in the meantime.

TL;DR: Mitchell Hashimoto (HashiCorp founder) tweeted that entire companies are under "AI psychosis." The post hit 1,359 points on HN with 669 comments, revealing deep divisions in the tech community about AI adoption. The thread itself became a case study in the problem — with accusations of bot astroturfing and a fundamental disagreement about what "AI is working" actually means.

Source: Mitchell Hashimoto on X · Hacker News discussion (1,359 pts)