Dear Reader,
What a great day it is to be alive.
A lot of you asked how my son did in his game.
It was a championship basketball game this weekend.
They lost.
He played great, and his team lost.
The coolest part was what he said afterward.
He said he wants to do better next year, and that he doesn’t like this feeling of losing.
That’s the grit we want.
When life hits you in the face, you get back up and go after it.
I’m very proud of him.
He should be proud of himself too, the way he played and the way he showed up for his team.
Well anyways, today I want to talk about Anthropic.
The Price Keeps Falling
Anthropic is going public this month.
That’s the plan.
Anthropic makes Claude, and OpenAI makes ChatGPT.
Those are the two independent labs most people interact with, though a lot of you are using Google’s Gemini more and more.
Anthropic was expected to go public at a $2 trillion valuation.
With the shakeout in the market, even before the election really, that came down to $1.5 trillion.
Then people started talking about $1.25 trillion.
Now they’re talking about $1 trillion, or raising a smaller amount in the IPO.
A lot of you have asked what I think.
This Isn’t SpaceX
We covered SpaceX earlier this year, and it’s a different animal.
As a rocket carrier, SpaceX was an incredibly profitable business.
It had a dominant competitive advantage.
Something like 80% to 90% of US payload sent to space went up on a SpaceX rocket.
It had pricing power and high gross margins, and you could see it right in the income statement.
Anthropic’s story is very different.
It’s a scale business.
It costs a lot to start, and the plan is that as you gain critical mass, your per-unit costs go down and your margins expand.
That’s the idea.
But a few things flag me about this IPO.
And I wouldn’t be surprised if some of you start getting calls from your broker.
“Hey, I’ve got a great deal for you. Want in on the Anthropic IPO?”
They’ll push it out to everybody they can think of.
The Insiders Can Sell Right Away
In the S-1, the plan allows insiders to sell their shares very quickly.
That concerns me.
I like lockup periods.
I don’t want a company to go public and immediately use it as an exit event for insiders.
It’s like you’re getting rich at our expense.
What makes me happy is when the company takes the money and expands the business, and the insiders are largely locked in.
That shows confidence in the business.
Here, they have permission to sell.
The Commitments
Anthropic now has about $1.3 trillion in debt and other contractual commitments for data centers and compute.
It’s chicken and egg.
They can’t grow without those contracts.
Without all that compute capacity, Claude would be slow and not very smart.
So they’re locked in.
Who’s Liable?
Say an agent of theirs acts on your behalf.
You let it plug into your bank accounts or credit cards to find and cancel the subscriptions you don’t use.
If that agent does something really shady inside American Express or Visa or Mastercard, who’s liable?
You?
Anthropic?
Ultimately it has to be them, or the product won’t work.
I see a lot of liability with wayward agents.
More Squeeze Than Juice
Because they need such big spending commitments, it’s a very unprofitable business today.
That doesn’t mean it can’t be very profitable at scale.
People have argued that, and I understand the point.
But right now it’s a lot more squeeze than juice.
The juice isn’t worth the squeeze, perhaps.
The Moat
This is the one that concerns me most.
Revenue is still growing fast, though the pace of that growth has slowed.
But token pricing is going down.
Revenue growth isn’t really the problem.
The problem is the price per unit, what they can charge.
That tends to happen with companies that work like this.
SpaceX had pricing power.
Here, revenue growth comes only by charging customers less.
So either they pull it off and it’s the greatest thing in the world, or they don’t and it blows up.
Then there’s the durable competitive advantage.
In a world where the most powerful models can do crazy harmful things, they have a disincentive to release them.
So they release models that are powerful enough, but a lot of people can copy those.
That gives others room to catch up in the mass market.
I’m sure they have super powerful models in private.
But they may not be able to release them to the mass market.
What I’d Do
I don’t know how the stock will trade.
Nobody does.
There’s a lot of overhang, and there are going to be a lot of net sellers.
Anthropic and OpenAI are the two killers in the independent AI industry.
We don’t know how it works out, and we don’t know where Google plays.
There’s a lot of debt overhang.
If Claude has become part of your daily life, where you can’t live without it, maybe you give it a shot.
But I don’t see any need to rush or panic buy.
My personal opinion is that you’ll probably get a chance to buy at or close to the IPO price within months of it going public.
Having said that, there’s a case to be made the other way.
They could lower the IPO price enough that it’s undervalued relative to its prospects, and the stock rises.
That’s how they used to do IPOs back in the day.
They’d bring them out a little undervalued, give us little people a taste, and the stock would pop 15%, 20%, 30%.
That doesn’t seem to be the plan here.
But the bankers, probably Goldman and Morgan, would be smart to do it that way.
It gives Anthropic a real pop right when it comes public, everybody’s happy, and it brings the shine back to the company.
That’s my two cents.
I don’t see any reason to rush into the Anthropic IPO.
The people who got in early have already made most of the money.
But there’s another way to own a piece of this story.
One of the biggest companies in the world already owns a large stake in Anthropic, and you can buy it any day of the week.
I put together a briefing on who that is, plus the quiet company every AI chip has to pass through.
Have a wonderful Tuesday.
I’ll see you tomorrow.
“The Buck Stops Here,”
Found this helpful? Share it with others.
Written by Dylan Jovine