Solving For Step 2
The trillion-dollar question mark in the AI revolution
On a school night in South Park, four boys sit in the bedroom of their jittery classmate Tweek, scrambling to finish a last-minute assignment on big business. Tweek keeps insisting that tiny gnomes sneak into his room at 3:30 a.m. to steal his underpants. The others roll their eyes.
Then a rustle comes from the dresser.
A small, helmeted gnome is wrestling away Tweek’s last clean pair.
When confronted, the gnome is offended. This isn’t random theft, he explains. It’s a serious business operation. If the boys want to understand how corporations really work, they should come and see for themselves.
They follow him through a hidden tunnel beneath the house and emerge into a vast underground cavern. Hundreds of underpants gnomes swarm across rickety catwalks, pushing mine carts and dumping endless loads of stolen underwear into towering piles. It looks like an industrial operation. Busy. Organized. Relentless.
Eager to impress their guests, the gnomes gather around a chalkboard.
When asked to explain what they are doing, one of the gnomes proudly explains their business plan.
Step 1: Collect Underpants
Step 3: Profit
The boys are confused ... What’s Step 2?
So they ask another gnome.
The gnome replies
Step 1: Collect Underpants
Step 3: Profit
... and so it goes.
Right now, much of the conversation around AI feels like this episode.
Step 1: Scale LLMs.
Step 2: ???
Step 3: Infinite Productivity
We are very confident about Step 1.
We are euphoric about Step 3.
And strangely vague about the part in between.
We might laugh at the cartoon because it seems absurd.
It’s not ... It is happening with AI Capex right in front of our eyes. A lot of activity. A lot of motion. A lot of confidence. A lot of data center capacity built. But no real explanation as to how all of this profitability and productivity is going to follow.
Unlike the gnomes though, we aren't just stealing laundry; we are pouring trillions into concrete and silicon, betting that Step 2 will simply 'emerge' from the scale.
Most importantly, those trillions are coming at the expense of going into other areas that might badly need them … ie building more housing.
Unfortunately, if we don’t get clear on what Step 2 actually is in the AI boom, we may end up with “mountains of underpants and very little profit to show for it.”
Just like the gnomes did ...