OK, what does that mean exactly?

WTF is a token

A token isn't a word.

Think of it more as a Lego block.

Sometimes a token is a whole word ("dog"). Sometimes it's part of a word ("un" in "unprecedented"). Sometimes it's just punctuation.

On average, one token → 0.75 words in English.

Roughly:

• 1,000 tokens → 750 words → 2-3 pages

• 100,000 tokens → 75,000 words → a 300-page novel

• 1,000,000 tokens → 750,000 words → 7 novels

WTF is context

Context is the model’s working memory, or everything it can "see" while responding.

It isn't just what you typed, but the entire conversation history that’s accessible to the model all at once.

The real magic happens through something called "attention".

The model isn’t reading left-to-right like you do. It’s using attention to access the entire context at once BEFORE generating each word.

Hitting the context limit means it literally can't remember anything more, and that determines whether the model can see your entire problem or just parts of it.

Why It Matters

As context limits grow, there’s one important to keep in mind:

Bigger context windows make prompting harder, not easier.

Yes, models are now able to analyze an entire codebase at once, but as a result it needs to work harder to filter out noise to generate good responses.

The winners in this new world of more context won't be those with the most data, but those who can craft prompts that take advantage of how attention works at scale.

OK, what does that mean exactly?

WTF is a token

A token isn't a word.

Think of it more as a Lego block.

Sometimes a token is a whole word ("dog"). Sometimes it's part of a word ("un" in "unprecedented"). Sometimes it's just punctuation.

On average, one token → 0.75 words in English.

Roughly:

• 1,000 tokens → 750 words → 2-3 pages

• 100,000 tokens → 75,000 words → a 300-page novel

• 1,000,000 tokens → 750,000 words → the entire Harry Potter series

WTF is context

Context is the model’s working memory, or everything it can "see" while responding.

It isn't just what you typed, but the entire conversation history that’s accessible to the model all at once.

The real magic happens through something called "attention".

The model isn’t reading left-to-right like you do. It’s using attention to access the entire context at once BEFORE generating each word.

Hitting the context limit means it literally can't remember anything more, and that determines whether the model can see your entire problem or just parts of it.

Why It Matters

As context limits grow, there’s one important to keep in mind:

Bigger context windows make prompting harder, not easier.

Yes, models are now able to analyze an entire codebase at once, but as a result it needs to work harder to filter out noise to generate good responses.

The winners in this new world of more context won't be those with the most data, but those who can craft prompts that take advantage of how attention works at scale.

The biggest hurdle to the one-person billion dollar company is not AI capability, but founder pain tolerance.

Sam Altman is betting on it in his private group chats.

Dario Amodei predicted it'll happen in 2026.

These geniuses think the one-person billion dollar company is inevitable, and unicorn teams are getting smaller, but a one person?

That’s not going to just happen.

One person is an intentional choice by an individual with a particular set of skills.

It's Alex Honnold free-soloing El Capitan.

It's David Goggins running an ultramarathon on broken feet.

The unicorn founders I know can’t even manage their email on their own, so who are we talking about here?

The who

The person likely closest to the milestone today is Heather Cox Richardson, who could easily replace her two copy editors with Claude.

Her newsletter reportedly brings in $12 million / year and a revenue multiple of 2-4x typical for newsletters values her business at ~$50 million.

A far cry from 1 billion.

Touching that B will likely require an app with recurring revenue and network effects, that can justify higher revenue multiples.

If it’s doable, it will probably be by:

1. Someone with a Flappy Bird product that goes vertical and sells at the top, or

2. Someone with an Instagram product who both “calls their shot” from the beginning and has the desire + technical ability to follow through on it.

Anything outside of consumer is highly unlikely. Enterprise money comes with enterprise headaches which will absolutely destroy a solo founder.

The reality

But let’s be honest:

If you’re lucky enough to build something many people love, you’ll want to hire if for no other reason than to make the product as great as it can be.

Maybe there’s a way to do it by only hiring agencies, which technically are a form of automation, but not in the spirit of Sam and Dario’s prediction.

Let’s see! The future is exciting 🚀

This Substack is a community for learning about AI, automation, and building lean businesses.

I'll be dropping actual code, real numbers, and hopefully not too many failures.

Subscribe to embrace our AI future while everyone else is having meetings about meetings!

Pull up a chair, Revenue doesn't mind the company.

Let's build together.

—Marc