Before you buy NVIDIA ahead of August 26

This series does one thing: it lays out what you should understand about a company before you make a decision, and then stops. There is no recommendation at the end, because we do not know your time horizon, your tax situation, what else you own, or how you behave when something falls 30%.

The business, in one paragraph

NVIDIA designs the processors used to train and run artificial-intelligence models and sells them, increasingly, as complete rack-scale systems rather than as individual chips. Its customers are principally cloud providers and large enterprises building AI infrastructure. In the quarter ended April 26, 2026, $75.2bn of $81.6bn in revenue came from the Data Center segment.

How it makes money

• Hardware sold at very high gross margin. GAAP gross margin was 74.9% last quarter. For context, that means roughly 75 cents of every revenue dollar remains after the direct cost of producing the product.
• An ecosystem that raises switching costs. NVIDIA's software layer is where a great deal of AI development work already lives. Moving to an alternative is not only a hardware decision.
• Scale in a constrained supply chain. Advanced packaging and high-bandwidth memory capacity are limited, and NVIDIA is the largest buyer of both.

Growth, without adjectives

Two of those numbers deserve a caveat. The 85% year-over-year figure compares against a quarter when the AI build-out was already well underway, so it is not a base effect. And the $91.0bn is not a result — it is a target the company set for itself, which brings us to the most important thing in this article.

Why a 'beat' on August 26 will tell you less than you think

Applied Materials demonstrated the practical version of this on Friday: it beat its quarter and raised guidance, and the stock fell anyway. When a good print is sold, the information was never in the results. It was in what the price already assumed.

So if you are watching on August 26, these are the lines that actually carry information:

1. Gross margin versus the 75.0% guide. Margin is where cost pressure, pricing power and product mix show up first. A revenue beat with a margin miss is a worse outcome than the reverse.
2. The guide for the following quarter. Forward guidance moves stocks far more reliably than reported results.
3. Data Center as a share of revenue. Rising concentration is both the growth engine and the risk.
4. The GAAP versus non-GAAP gap. Last quarter GAAP EPS ($2.39) exceeded non-GAAP EPS ($1.87). Whether that repeats tells you how much of reported profit is coming from outside the operating business.
5. Any change in customer concentration language in the filing. Not the call — the 10-Q.

Valuation, briefly

At $225.16 per share the stock trades at roughly 30 annualised non-GAAP quarterly earnings, or 23.6 annualised GAAP quarterly earnings. The S&P 500's forward P/E was 20.0 in early August. Our full workings are in the price tag on NVIDIA.

The single biggest opportunity

If AI infrastructure spending continues at anything like its current pace, NVIDIA is the company that converts that spending into profit at a roughly 75% gross margin, with $48.6bn of free cash flow in a single quarter to reinvest or return. There is no second business in the world currently doing that.

The single biggest risk

The same sentence, read backwards. Ninety-two percent of revenue comes from one segment, sold to a concentrated group of customers whose capital spending is discretionary, cyclical, and decided by a handful of executives. Those customers are also designing their own chips. A single year in which the largest buyers pause is the scenario the price has the least room for.

Questions to answer before you decide

• If this position fell 40% and nothing about the business had changed, would you add, hold, or sell? Answer honestly, in advance, in writing.
• Do you already own this through an index fund? At a $5.45tn market cap, most US equity index funds hold it as their largest position. Buying the stock on top of that concentrates rather than diversifies.
• Is your decision about the next nine days or the next five years? Those require different evidence, and the earnings date is only relevant to the first.