The AI market is often presented as a battle between model companies. But there are actually several businesses hiding inside the AI economy. They build foundation models.
The AI market is often presented as a battle between model companies.
But there are actually several businesses hiding inside the AI economy.
Model companies
They build foundation models.
Their advantage may come from:
research,
data,
compute,
talent,
capital,
and scale.
Infrastructure companies
They provide:
compute,
cloud,
data centers,
networking,
storage,
security.
They don't necessarily need to win the model race.
They can benefit from multiple model companies growing simultaneously.
Application companies
These take AI and embed it into a workflow.
A legal system.
A hospital.
A factory.
A bank.
A design platform.
A logistics company.
The customer may not care which model powers it.
They care whether the product works.
Data companies
Then there are organizations whose advantage comes from proprietary information.
If the model becomes commoditized, proprietary data may become more important.
Distribution companies
And finally:
companies that already own customer relationships.
They can integrate AI into existing products and immediately distribute it to millions of users.
This creates a fascinating possibility:
The companies best positioned to capture AI value may not be the companies that invent the smartest AI.
They may be the companies that control the path between intelligence and customers.
What this means
This article is editorial analysis. Verify consequential claims against primary sources before relying on them as fact.
The question nobody asks
Which parts of this argument are documented fact, and which are analysis or uncertainty?
