If we look back to the start of NVDA’s fabulous run of delivering extremely strong results and positive earnings surprises, it was clear from the outset that two factors were driving the gains:

–strong sales growth of the company’s cutting-edge AI chips, and

—operating leverage, the fact that the expenses of running the company’s research and sales efforts are rising at a much slower rate than sales. Over the past twelve months sales are up by 94% and expenses by 44%. The consequence of this is that profit growth has far exceeded that of sales.

It’s been evident to anyone who looks at the NVDA financials, however, that while sales are still expanding faster than costs, the rate at which they’re doing so has decelerated. In addition, the absolute gross margin, at 75%, is at eye-popping levels. Noticing this (how could anyone not?), TSMC has upped the price it charges to manufacture NVDA offerings.

This adds up to the likelihood that profit growth has lost the turbo boost operating leverage has been giving it and will depend solely on sales growth, at least for the near future, to make earnings gains.

All this was crystal clear on August 28th when NVDA reported 2Q25 earnings. The stock dropped by 20% in the ten days or so after the results were published. Arguably, though, the decline was triggered by the announcement that NVDA’s latest AI chip was having teething problems. In any event, the stock then roared ahead by 40% to a new all-time high just before the 3Q earnings announcement. It’s down slightly today as I’m writing this.

Where to from here?

I think that for the first time in a while, valuation will be based more heavily on current earnings rather than factoring in the possibility of large future positive earnings surprises. My sense is that, again for the first time in a while, the stock is fairly valued. I’m holding most of what I bought years ago, but I’m selling enough so that, for now, it will no longer be my largest position.