Is NVIDIA a good business model?

Dan Bin said: In the U.S. stock market, we see the real business models. If they are relatively good and have strong competitiveness, whether traditional or tech, they perform extremely well. So we say that in the U.S., when it comes to investing, you just need to pick these good ones, hold them long-term, and not worry too much about systemic changes.

So today, let’s analyze what a business model is and what makes a good one.
A business model is essentially: who pays, why they pay, how they pay, how money is made, how it’s spent, and whether the model can be sustained long-term and grow bigger.
A good business model has: 1) High gross margins and strong cash flow; e.g., Microsoft’s Office subscriptions—stable revenue, extremely low marginal costs, fat margins, cash flow like a money printer. 2) Scalability and replicability, getting easier as it grows; adding one more Office user costs Microsoft almost nothing in server costs, keeping margins high. Once proven, it can expand globally. 3) A moat: hard to copy, and even if copied, competitors can’t win; e.g., Microsoft Office/Windows and Adobe—entire workflows are tied to these tools, making switching hard. 4) Healthy revenue structure: Ideally, recurring revenue like subscriptions—predictable and plannable. Microsoft: Office 365, Azure subscriptions. 5) Pricing power: Can raise prices without losing customers. Microsoft Office and Adobe Creative Cloud hike prices regularly; companies complain but still renew.

Now, look at Nvidia’s business model: 1) Obviously high margins and strong cash flow. 2) Scalable and replicable, getting easier over time. Jensen Huang’s so-called "AI factories" aren’t pure software (like Microsoft); clients must invest heavily in data centers/racks, so the model is "light design + heavy client capex," with growth tied to overall AI infrastructure investment. 3) A moat: Hard to copy, and even if copied, rivals can’t compete. The integrated hardware-software ecosystem locks in users, creating an extremely high moat. 4) Healthy revenue structure: Currently, most revenue comes from "selling hardware + systems," meaning it’s more cyclical than subscription giants like Microsoft. Subscriptions are still small but promising. Huang recently said all Nvidia’s investments so far aim to expand CUDA’s reach and grow the ecosystem. 5) Pricing power: Can raise prices without losing customers. With AI demand exploding, Nvidia has absolute pricing power now—but watch AI infrastructure demand and competitive threats.

Overall, Nvidia’s business model is undeniably excellent. But due to semiconductor cyclicality, sudden shifts in competition (who knows if tech will suddenly leap ahead with something better than Nvidia?), and policy risks, it’s also why Buffett avoids such stocks.

So, the key when investing in Nvidia is to focus on three things:
1) The pace and intensity of AI infrastructure capex and the long-term demand it drives.
2) The competitive landscape in data center business and where gross margins will stabilize.
3) Whether Nvidia’s revenue growth, EPS, and free cash flow (FCF) growth can still justify current and future valuations.
Don’t go all-in recklessly; place patient bids in undervalued ranges. When a great price appears and fundamentals are solid, be bold in adding. Don’t let market frenzy or crashes overturn your long-term logic. Adjust positions and portfolio structure dynamically during extreme valuations or sentiment to deflate bubbles. Regularly ask: "Am I overestimating my understanding of Nvidia and the AI race?" "Are there critical counterarguments I’m ignoring?"
Only then can you navigate the bubbles and noise of this AI revolution and share in its ultimate rewards.