The more confused investors are about AI, the better it is for you.
I bought AMD at $4.2, Tesla at $13, Palantir at $7. All during peak confusion.
Confusion creates mispricing. Mispricing creates asymmetric opportunity.
You only have to get it right once.
Here's the mental model you need.
You can't predict the future, but you can bet on companies that handle uncertainty well. They:
-
Are obsessed with end customers.
-
Iterate faster than competitors.
-
Are willing to disrupt themselves.
These qualitative properties tend to result in revenue rising much faster than costs over the long term, even with big speed bumps along the way.
Why? Because they compound goodwill.
And compounded goodwill eventually becomes a gravitational force.
Customers don't just prefer them.
They can't imagine going anywhere else.
Revenue ends up outpacing costs and that gap flows directly into free cash flow.
And stock prices track free cash flow per share over the long term.
No matter how crazy the market gets, if a share generates more free cash flow, the price goes up long term.
If free cash flow per share grows exponentially, so does the stock price.
See Palantir's case below.
Valuations in tech are a loser's game.
The only real margin of safety is buying companies you believe can grow free cash flow per share far beyond what the market is currently pricing in.
People thought Palantir was a horribly overpriced consulting company at $7. I leaned into the confusion.
This is how you win big.
How to tell when a company is about to exponentiate free cash flow per share?
By deeply understanding its operational blueprint.
How it creates value in a way that others can't replicate.
Enter The Costco Algorithm: the father of all multi-baggers.
Costco employees wake up every day completely obsessed with delivering more value to end customers per dollar spent on their behalf.
This compounds goodwill, which puts more capital into Costco's hands (because customers love it), which Costco uses to get more efficient and lower real prices for customers.
Nick Sleep coined this as Economies of Scale Shared.
The Costco Algorithm works, as you can see below.
But it has a speed limit and AI removes it.
Modern versions of the Costco Algorithm unlock network effects.
They build massive, highly engaged user bases. Fast.
This yields proprietary data moats: access to behavioural data at a scale no competitor can replicate.
This allows them to train an AI model no one else can train.
In this world, embracing AI is not optional.
The only way to compete and win is by delivering more value per token.
Two forces compound on each other:
-
As AI gets smarter, the value delivered to end customers per unit of proprietary data rises exponentially.
-
As value rises, engagement increases non-linearly, which drives even more proprietary data back into the model.
Ontology Velocity is the end result.
If your Ontology moves even 0.1% faster than a competitor's, you win.
You deliver that tiny bit more value per token, which makes your product exponentially more appealing.
Which attracts more data. More data trains a better model. A better model delivers more value.
The gap compounds until it becomes uncrossable.
Here is where the market is confused.
Generic AI models will be very useful. But the market assumes they'll disrupt every software player equally.
Many will die.
But those that have achieved Ontology Velocity will thrive.
Because end customers always default to the option that delivers better outcomes per dollar. Even if the difference starts small.
The "wide moat" businesses Warren Buffett talks about are obvious to us now — because the market has had decades to catch up with his thinking.
Modern wide moats elude the untrained eye. Just like they did back then.
A lot of modern Costco Algorithm companies look like toys at the beginning.
Tesla looked like a niche EV shop for wealthy people.
By obsessing over value per dollar, they became a highly efficient EV printer.
Now, as those EVs harvest driving data at scale, Tesla is positioned to become an Autonomy OS.
Spotting moats isn't easy. But there's a rule of thumb:
Do something hard that gets you lots of users. Lots of users get you lots of data. Lots of data lets you outrun every generic AI model.
The more components the hard thing has, the harder it is for anyone else to reach the point where Ontology Velocity can be unlocked.
-
How hard is "hard"? Count the components: user experience, regulatory, manufacturing, distribution, trust. The more boxes checked, the deeper the moat.
-
Has the company scaled far enough to accumulate more proprietary data than anyone else in its category?
-
Is the company actively training its AI on that data or leaving it on the table?
-
Is the resulting product experience already pulling ahead of generic models in ways users can feel?
The ultimate litmus test is a cult user base.
If a company ticks all the above boxes, it becomes what I call a Singularity Scaler.
A company that accelerates free cash flow per share growth as we move towards the Singularity.
Because it has abstracted away enough value chain complexity to become exponentially more powerful — at near-zero marginal cost — as AI capabilities compound.
This is in turn the basis for the mental model I believe will create vast fortunes in the coming decade: Singularity Asymmetries.
Just like the speed of value creation is unprecedented, so is the speed at which false narratives spread.
This makes it more likely than ever that the market will unanimously declare AI has killed a company, precisely when that company has achieved Ontology Velocity.
The result: exponentially rising earning power. Exponentially decaying stock price.
That gap is the opportunity.
The math is extraordinary. If a stock goes down 90%, you can buy 10 times more shares per dollar than you could before. If you were planning to allocate $10,000, it's as if someone handed you $90,000 for free.
At $1,000,000, that's $9,000,000 in extra purchasing power handed to you by the market's confusion.
At $10,000,000, that's $90,000,000 in extra purchasing power.
At $100,000,000, that's $900,000,000.
Nearly a billion dollars in extra purchasing power — handed to you for free. Because competition in the AI space is won at the infinitesimal edge.
And the average investor doesn't get it.
The more confused the market is, the bigger the gift. You only have to get it right once.
This essay is for informational and educational purposes only. Nothing here is financial advice. I own positions in some of the companies mentioned. Always do your own research.