Why Stripe is betting $7B it won't matter which AI you use

Here’s what we’re reading and thinking about this week:

Stripe is reportedly paying more than $7 billion for a company called OpenRouter.

OpenRouter does one thing. It lets you switch which AI model does your work, without redoing the work. It hooks you up to more than 400 of them.

That's the whole product. Seven billion dollars.

So here's how we read that. The biggest AI deal of the year is a bet that the model is the part you throw away.

Think of it like an oven

Say you have a great recipe for banana bread. The recipe is yours. The oven is just an oven.

If the oven breaks, you buy a new one and you make the bread again. You don't lose the recipe. The recipe was the valuable part the whole time.

AI works the same way. The model is the oven. Your playbook is the recipe.

Stripe just paid $7 billion for an easy way to swap ovens.

The model is the cheap part

Three months ago OpenRouter was worth $1.3 billion. This week it sold for five times that.

AI models didn't get worse in those three months. They got better, and cheaper, like they do every quarter.

So what went up in value? Not the models. The freedom to stop caring which one you use.

Think about what you'd have to believe to pay that. Nobody pays $7 billion to switch between models if they think one model is about to win. You pay it because you think they'll all end up about the same, and because whoever is ahead keeps changing.

Meanwhile, most of the internet is still fighting about which model is best.

🌶️ Our take: the model doesn't matter. Your playbook does.

When AI does something well for you, it feels like the model did it. It didn't.

You decided what good looked like. You fixed it eleven times. You told it what to never say, which numbers need a second set of eyes, and how your team talks.

That's not the model. That's a process. The model just ran it.

So the valuable thing was never the model. It's the written-down version of how the work gets done. The playbook!!

That's the whole idea behind own the playbook, rent the tech. You own the recipe. You rent the oven.

Most teams never write the recipe down

Here's what we keep running into.

A team spends eight months teaching a custom GPT to do real work. Client updates, case studies, whatever it is.

By month eight it's genuinely good! It knows the tone. It knows what's off limits. It knows the three things legal always flags.

But none of it is written down. It lives inside the tool, in the tool's memory, and in one person's chat history.

Then something changes, because something always changes. The model goes away, or the person quits, or the price goes up.

So somebody opens up those instructions for the first time in a year...

and finds forty rules nobody remembers writing. Nobody can explain half of them. Nobody knows which ones are holding it together, and which ones somebody typed in once when they were annoyed.

That's not a recipe. That's a really good cook who left.

"But Claude is just better at this"

Fair. We hear it all the time. Here's our honest answer.

If your bread only comes out right in one oven, the problem usually isn't the oven. Your recipe is missing steps. The oven has been covering for you.

Good recipes travel. We call it the smart intern check. Could a sharp person who has never done this job follow your written steps and get it right?

If yes, a good model can too. If no, you never had a process. You had a favorite tool and a lot of hope.

UPCOMING EVENTS

Need help recruiting your next (or first) AI Operator?

If you run operations, AI has probably made your job harder, not easier.

Everyone on the team is using it. Nobody is using it the same way. None of it is written down. And you're the one who has to make it all coordinate.

Lani Koontz spent 15 years in operations before she started building AI systems for a living. She's teaching how she does it on August 26.

LINKS

For your reading list 📚

• Target hired its first Chief AI Officer this week, poached from Lowe's. His stated first priority is to listen and learn before building anything. Genuinely the right answer.

• DoorDash's internal agent platform chewed through 130,000 engineering tasks in one month. The detail we can't stop thinking about: it runs on 300+ reusable playbooks, invoked 10,000 times a week.

• Google's Gemini app crossed a billion monthly users on August 11, the fastest-growing product in Google's 28-year history. 63% of them talk to it out loud.

• Spotify is putting an "AI Persona" badge on artists who aren't real people, and keeping them out of your recommendations unless you go looking.

• OpenAI is hiring a power trader. Not an engineer, a trader. … Interesting

That's all!

We'll see you again soon. Thoughts, feedback and questions are much appreciated - respond here or shoot us a note at [email protected]

Cheers,

🪄 The AMP Team (formerly: the AI Exchange Team)