The Largest Opportunity Nobody Can Size
We are not in the middle of the AI story. We are in the first chapter, and the person writing it is still figuring out how long the book is.
I have been sitting with that thought since listening to Jensen Huang's conversation with Lex Fridman. Jensen was asked about NVIDIA's future, about where the ceiling is for a company now worth $5.3 trillion. The most powerful answer he gave was not a number. It was a set of open questions. How many knowledge factories does the world need? How many tokens does the world need? How much is society willing to pay for them? What happens to the global economy when productivity improves at a scale we have never seen before?
Jensen Huang does not ask questions he does not have answers to. He is one of the most deliberate communicators in business. When he frames something as an open question in public, it means the answer is so large it cannot yet be stated. Those are not rhetorical questions. They are the edge of what the most informed person in the room can see.
Today NVIDIA's infrastructure powers every serious AI system on the planet. The chips, the networking, the software stack, the data center architecture. What Jensen calls the knowledge factory: a physical plant that manufactures intelligence the way a semiconductor fab manufactures compute. You feed it power, you feed it data, and it produces tokens. Those tokens are the unit of manufactured intelligence, and they are already being priced accordingly. General purpose reasoning costs fractions of a cent per million tokens today. Specialized intelligence, the kind that makes decisions in drug discovery, financial modeling, legal analysis, and engineering simulation, is heading toward $1,000 per million tokens. That is not one market. That is a tiered commodity structure that has barely begun to form.
NVIDIA is the largest company in the world by market capitalization. It did not get there by accident. Jensen made a bet in the early 2000s that programmable compute would matter beyond gaming. He carried that bet through years of margin compression, skepticism, and a market cap that at one point fell to $1.5 billion. The company that bet looked like a mistake for a decade before it looked like a prophecy.
And yet Jensen is publicly asking how many of these factories the world needs.
The person who built the infrastructure that every AI company on earth depends on, who runs a company that has compounded faster than almost any asset in modern financial history, who has been right about this technology longer than most people have been paying attention, does not yet know how large the demand will be. Not because he lacks information. Because the opportunity is genuinely that large.
Every generational technology shift has had this quality at its inflection point. The builders of the early internet infrastructure could not have told you in 1997 what streaming video, cloud computing, or mobile commerce would consume in bandwidth two decades later. The demand did not exist yet because the applications had not been invented. The knowledge factory changes that calculus for intelligence. When manufactured reasoning becomes cheap enough to use at scale, industries currently constrained by the cost of human expertise will be restructured entirely. Those industries are not small. They are healthcare, law, finance, education, engineering, and government.
Jensen is building the factory. He is telling you openly that he does not know how many the world will need. For an investor, that is not a warning. That is the signal that we are in the earliest innings of the biggest shift of our lifetime.
Bashar Aboudaoud
Managing Member, UpRound
