The $99 Savings Trap

This year, for the first time, Wall Street will put a price on the “intelligence” part of AI.

Anthropic confidentially submitted its draft S-1 to the SEC, announced June 1, 2026 (Anthropic).

Anthropic isn’t compute infrastructure. It’s the intelligence layer. And the market is about to face a new challenge: how do we put a price on intelligence?

Founders and Investors are pricing AI wrong

Most founders and investors are still pricing AI as a discount on labor.

“We replaced our data analyst with AI agents. We’re saving $120K a year.”

They’re not lying. The math checks out.

A data analyst runs about $60 an hour (BLS). Researching one company (search the web, read the site, pull the data, write the report) takes an hour or two. Call it $100 a company.

An agent does the same work for about $1, maybe less (from my own experience.)

You save $99 on every company research report.

The $99 Savings Trap

The $99 Savings Trap is measuring AI by the cost of the human task it replaces. The most you can ever save is what that task used to cost you.

That $120K you saved? That was the cost of the work, no more. That’s the ceiling. The day you automated the task, you hit it.

Why does everyone measure this way? Because it’s the only ruler they have today.

It’s the ruler from the outsourcing era. The cloud era.

Every wave before AI was about getting at least the same result for less money.

AI’s value is in the Intelligence

Years back, I worked with a VC firm in New York, building a CRM for VCs.

The CEO had big ideas – most of them too expensive to build. My job was to keep engineering focused on what mattered and what was actually possible to build.

One big module was investment opportunities, which, as you might guess, meant researching companies.

Back then, the best we could do was automation: write a script, scrape the data, dump it in a database, extract what matched the rules. Scripts work, and they’re dumb.

An AI agent is different. It reads the messy data, makes the judgment, writes an informed decision… in minutes, for about a dollar.

That’s not a script. That’s analyst work.

The $99 Savings Trap looks at that and says: “Great. We saved on analysts.”

August is when the math breaks in public

Until now, founders and investors ran this savings math in stealth mode: they played with AI, found it could do some human’s work, and they fired the person or stopped hiring.

On August 2, 2026, the EU’s AI Act enforcement kicks in for general-purpose AI. The AI Office gains the power to fine providers up to €15 million or 3% of global annual turnover, whichever is higher (EU AI Act, Article 101).

If you are deploying AI, whether as a chatbot or as AI-generated content, and you are serving EU-based users, you need to disclose it (Article 50).

That means if you are using AI, you need to be direct about it. And once you are direct about it, users will start asking: are you using AI in the right way?

Some founders are already sensing this and asking themselves the question: if any competitor can leverage AI, what’s the advantage for us? (defensibility, 2026).

The savings math is falling apart.

Your Next Move

So as a founder or investor, you have:

1. Other founders and investors finding more creative ways to leverage AI beyond cheap work.
2. Wall Street about to put a public price on intelligence itself.
3. The EU (and sooner or later the US) to expose how you are using AI.

If you keep measuring AI the way you do today, the clock is running against you.

You need to change the question.

The wrong question: “What does this save us?“
The right question: “What couldn’t we do yesterday that we can do today… with AI?“

The first question has a ceiling. The second one doesn’t.

I’ve watched AI agents build tools I’d never have had the time or knowledge to make myself, and I’ve been coding for a long time.

There is a more detailed answer for you in this book: Founders and Investors Will Be Replaced by AI.

To prove the ceiling is gone, I wrote its manuscript at 2am and built an agent pipeline that turns it into a living book, one that improves itself over time. A book a legacy author and a publisher couldn’t have made.

Don’t ask what it saved me. Ask what it made possible.