SheeplzSubscribe
What Nvidia could be worth
By Dylan Siegel
This is my understanding and opinion, for education. Not personal financial advice.
New to this? Read first: Nvidia, taken apart · Where AI takes the economy
Nvidia's stock price comes down to three numbers: how much it sells, how much of that it keeps as profit, and how many years of that profit people will pay for. Here's the math to $1,000 a share.
My take, in one line: if AI becomes machine labor at the scale I think it can, $1,000 by around 2030 doesn't need a miracle, just Nvidia beating already-big expectations. But the stock can still fall hard along the way.
The formula
Share price = revenue × profit margin × P/E ÷ shares
- Revenue: total sales in a year.
- Profit margin: the share of sales left as profit after all costs.
- P/E: how many years of today's profit investors pay for one share. A high P/E means they expect growth.
- Shares: about 24.2 billion.
On October 19, 2026 the stock was about $218, which makes the company worth roughly $5.3 trillion.
Late 2028: three outcomes
| Scenario | Yearly revenue | Profit margin | P/E | Share price |
|---|---|---|---|---|
| Bear | $440B | 40% | 18× | $131 |
| Base | $650B | 48% | 25× | $323 |
| Bull | $800B | 52% | 30× | $517 |
The base case isn't far-fetched. Nvidia did $96 billion in a single quarter this summer and guided to $108 billion the next.
The bear case shows the trap: you can be right about AI and still lose. Even at $650 billion of sales and a 48% margin, if investors only pay 15× profit the stock is about $194, below today's price.
Around 2030: the AI-acceleration cases
| Scenario | What the world looks like | Revenue | Margin | P/E | Share price |
|---|---|---|---|---|---|
| AI boom stalls | Spending cools, chips get cheap | $700B | 35% | 16× | $162 |
| Strong AI growth | The buildout keeps going | $1.2T | 50% | 25× | $620 |
| Agent boom | AI workers everywhere; compute demand explodes | $1.6T | 52% | 30× | $1,031 |
| Intelligence explosion | AI speeds up AI | $2.2T | 55% | 35× | $1,750 |
These are my scenarios, not probabilities and not price targets. The top one would make Nvidia a $42 trillion company, which has never existed.
What $1,000 actually requires
| Working backwards from $1,000 | |
|---|---|
| Company value | $24.2T |
| Yearly profit needed at 30× | $807B |
| Yearly revenue needed at a 50% margin | $1.61T |
That's a huge company, but it's the agent-boom world from the last post, not science fiction. My working view: around $650 is the central case for 2030, and $1,000 is the bull case I think the market underrates.
Why it might not go in a straight line
- Financing. If the money behind the buildout tightens, orders can pause even while AI keeps improving (see Who pays for the AI buildout).
- The multiple. The P/E can shrink fast in a scare. That alone can cut the price in half without profits falling.
- Competition. Google, Amazon, AMD and Chinese chips take share, or squeeze Nvidia's margins.
- Taiwan. Almost everything is made there.
How I'd act on it is my own business. The point of this post is the math: the upside is real, the drawdowns are real, and the difference between them is a few numbers you can watch every quarter.
What I track
| Number | Good sign | Warning sign |
|---|---|---|
| Revenue vs. guidance | Meets or beats | Repeated cuts |
| Gross margin | Holds near 75% | Slides for several quarters |
| Cash collected vs. profit reported | Grow together | A widening gap |
| Big Tech capex plans | Rising, with busy data centers | Delays and cancellations |
| Neocloud credit (CoreWeave bonds) | Yields falling | More distress |
| AI revenue at customers | Growing fast and profitably | Usage up, profit not |
