Dear Reader,
Anthropic is going public.
This is the maker of Claude and the biggest competitor to ChatGPT.
I think it’s a great chance to unpack some lessons about competitive strategy that can help you in your own investing.
That’s my goal in life, really.
To share the painful lessons I’ve learned so you don’t have to repeat them.
Lord knows I’ve made every mistake there is.
I heard somebody say recently that the people who will run the world will fit in an elevator.
That’s a little crazy when you think about it.
It’s a big bang of intelligence, and it matters for us as humans, as a society, and most importantly for this channel, as investors.
When SpaceX went public, I talked about the stock and the best time to buy the shares.
With Anthropic, I want to talk about the business.
I’m also going to show you the related party transactions, so you can see how insider deals work among that little cabal of Silicon Valley elite.
Cabal is, for lack of a better word, the kindest one I can come up with.
Start With Gross Margins
My biggest problem with these companies is price competition.
I don’t know which of them, if any, will have pricing power.
SpaceX was the only game in town, so it could charge whatever it wanted and had very high gross margins.
Here’s a pro tip.
When your buddy opens a restaurant, or your grandson shows you a business plan and asks you to invest, open the income statement first.
Ask what the gross margin is.
How much does it cost to make what you’re selling, before you get into salaries and everything else?
Superstar companies run gross margins around 80%.
Think Microsoft, Google, or Meta, before all this AI spending.
Microsoft writes the software once and sells it a bazillion times.
Say it cost $1 billion to build.
By the billionth copy, the cost is $1 per copy, and they’re selling it for $300.
Anthropic is in a very different position.
In the second quarter, Anthropic spent $5.6 billion on computing costs, including developing new models.
That was against $11.5 billion in revenue.
Roughly half of every revenue dollar goes straight out the door.
That’s a 50% gross margin.
Good, but not Microsoft.
The Rippling Story
The Information, a great little online magazine, ran a piece by Laura Bratton and Stephanie Palazzolo on the competition between Anthropic and OpenAI.
They used a Silicon Valley company called Rippling as an example.
In May, Anthropic powered 57% of Rippling’s coding work.
OpenAI handled 16%.
Within six months, that flipped.
Now OpenAI handles 52% and Anthropic handles 18%.
That tells you two things, and both are concerning if you care about durable competitive advantage.
The first is that there appear to be no switching costs.
You can run your company on Claude and press a button to move to OpenAI.
That’s scary.
The second is how OpenAI won the business.
It offered great discounts.
Anthropic was smart.
It went after businesses while OpenAI went after consumers.
The business use case turned out to be better for higher-margin work early on.
But OpenAI looked at Anthropic’s numbers, saw that 50% gross margin, and went right after it.
You call up all of Anthropic’s customers and say, “We’ll give you that deal for less.”
A Game of Chicken
If OpenAI keeps doing that, it ends up with horrible gross margins of its own.
It has to underwrite losses just to stay competitive.
At some point its investors are going to ask, “What’s the plan here?
When can you raise prices?”
It’s like playing chicken in two cars.
You drive straight at each other and see who blinks first.
The one with the most cash and the longest runway doesn’t blink without destroying itself.
So the question is whether any of these models has pricing power.
Can anybody with a billion dollars launch an LLM and give it away for free, the way China is doing, and keep prices artificially depressed?
That’s a very big question.
What It Means for the IPO
The people who got in early, the so-called elites who did all those related party transactions, are going to make their money.
The IPO might do well on the first day, or it might not.
I have no idea.
What I do know is that over time the business determines how the stock does.
If this is a business with zero pricing power, where everything comes down to price, it starts to look like insurance.
Geico cuts its price, then State Farm cuts its price.
That’s a 10% to 15% profit margin business, not a 30% margin business.
It’s a very different model with very different characteristics.
It’s still early, and either side can make moves that change this.
Anthropic would be very happy if OpenAI blew up and it could consolidate everybody and raise prices.
We’ll see if that happens.
But Google has deep pockets, and it can drop prices and give everybody a hard time.
If you’re in ChatGPT’s seat, it’s like being a NASCAR driver.
Anthropic cuts you off on the left.
Meta comes out with Muse and moves right into the middle.
And you’re sitting there asking whether to go left or right.
That’s exactly what’s happening.
It’s a fascinating case study.
I’m sorry if I’ve bored you to death with this.
I find it very, very interesting, which is why I do this for a living.
Anthropic and OpenAI are playing chicken, and nobody knows who blinks first.
But there’s one thing both of them need no matter who wins.
Power. Enormous amounts of it, and the public grid is years behind.
So the big AI players rent it from companies that build the power plants right next to the data centers, on contracts that run ten and twenty years.
The price war doesn’t touch those contracts. The meter keeps running either way.
That’s the kind of pricing power I wish the model makers had.
I put together a report on three of these companies.
One of them has raised its payout every year for 27 years, and the next payout cutoff is October 30th.
Get the tickers and full breakdown here>>>
Have a wonderful Tuesday.
I’ll see you tomorrow.
“The Buck Stops Here,”
Found this helpful? Share it with others.
Written by Dylan Jovine