Shares of Nvidia were under modest pressure Thursday, despite the company reporting an overall strong quarter with guidance that came in ahead of expectations. Clearly, some investors are finding issues within the report. While we think those concerns are misguided, here are three nitpicks generating the most noise on Wall Street. Data Center revenue missing expectations Inventory increasing by roughly 33% sequentially Customer concentration 1. Data center revenue ‘miss’ While data center results did technically miss the consensus FactSet estimate, we were not that concerned. In our Nvidia earnings analysis Wednesday evening, we talked about seeing clear signs of sustained strong demand. We also got the sense from the post-earnings conference call that Nvidia simply wasn’t able to ship as much as they would have liked to due to capacity constraints. Nonetheless, we decided to do a little more digging into the numbers that made up the consensus estimate. Out of 24 firms covering Nvidia, we found that one of them, with a sell rating and $100 price target, provided what we think amounts to an outlier on the upside. The firm in question, Seaport Research, provided FactSet with a data center estimate of $47.29 billion, $4 billion higher than the second-highest estimate. This one estimate added $258.6 million to the overall consensus. In other words, if we stick with the other 23 estimates — ranging from $39.7 billion to $43.08 billion — the consensus estimate drops to $41.084 billion, only slightly below the result $41.096 billion that Nvidia delivered. Rounding that, it’s really a match. In its earnings reaction note, Seaport remains worried about data center revenue and reiterates its sell rating and price target. Some might argue we should also remove the lowest estimate if we’re removing the highest. However, we don’t think that is necessary, given that the lowest estimate does not amount to an outlier. The bottom four estimates range from $39.7 billion to $40.09 billion — so it was not uniquely low, whereas the highest estimate was, indeed, uniquely high. 2. Inventory build-up The inventory concern was really odd. The bearish argument: The company is sitting on old inventory that it simply can’t offload. If you think that’s the case here, then we wonder which conference call you were listening to. The bullish argument, which we think is the only one that makes sense given the commentary on the call, is that inventory is being built up to meet what the company believes will be high demand — build inventory so that you don’t lose out on sales due to a lack of product. Indeed, Nvidia noted on the earnings release, “Inventory was $15 billion, up from $11.3 billion sequentially, to support the ramp of Blackwell Ultra.” The idea that some may be looking at this as a negative is perplexing, to say the least. The positive price action in the semiconductor cohort more broadly also supports the argument that the build is in anticipation of demand, and should therefore be viewed bullishly. Otherwise, chip-focused exchange-traded funds such as the VanEck Semiconductor ETF would be lower on the day, especially since Nvidia carries the most weight in that fund. Investors are not buying these other names because they think customers are rotating to them and away from Nvidia. They are being bought because Nvidia just said that demand is strong and likely to sustain well into next year and through to the end of the decade as the global compute infrastructure is refreshed for the age of artificial intelligence. 3. Customer concentration Is Nvidia too reliant on a handful of deep-pocketed customers buying its chips for data centers? That’s not a new debate, but it is once again surfacing in the just-reported fiscal 2026 second quarter, which ended July 27. In its quarterly securities filing, Nvidia disclosed that three direct customers accounted for a combined 56% of its accounts receivables balance at the end of the July quarter — basically, this is the money that its customers still owe the company. The companies aren’t named, but the percentages given for each were 23%, 19% and 14%, respectively. That is actually a pinch below the prior April quarter, when Nvidia said three customers totaled 57% of the balance combined, based on 27%, 18% and 12%, respectively. To be sure, both quarters represented an increase compared with Nvidia’s filing at the end of last fiscal year, when the company only had two customers that exceeded 10% of revenue — a threshold that requires disclosure under accounting rules. At that time, Nvidia said two customers represented 17% and 16% of its accounts receivable balance, equal to 33%. . In general, investors look at customer concentration as being risky — consider a hypothetical smartphone component supplier that derives 75% of its revenue from one manufacturer. If that manufacturer takes its business elsewhere, the supplier could be in trouble, at least in the near term, until it figures out a new plan. We view Nvidia’s situation differently for a few reasons. For starters, the cloud-computing providers like fellow Club name Microsoft are among the biggest spenders on its technology — even if we don’t know for certain which companies these three big Nvidia customers are. While these cloud providers are spending billions of their own dollars on the chips and networking gear, they’re turning around and renting a lot of that computing power to tons of other customers. That’s more comforting to us than it would be if Microsoft were buying all these chips for itself. Now, layer in the fact that the cloud providers frequently say that they have more demand for their AI services than capacity, and it makes sense why they’re trying to get their hands on all the computing power they can. Additionally, as Jim Cramer put it Thursday morning: “There are probably seven or eight customers that would like to be” accounting for a larger share of Nvidia’s orders. “Sovereign AI could be really big, but you don’t just say I’m going to shaft Azure and give it to Iceland.” Even with that dynamic being at play, Nvidia’s business catering to so-called sovereign AI is growing, adding a new kind of customer to its revenue stream at the same time these unnamed three tech giants are gobbling up chips. Sovereign AI, a term Nvidia has popularized, refers to nations building and controlling their own computing infrastructure so they can develop AI applications in their native languages, utilizing their own data. “We are on track to achieve over $20 billion in sovereign AI revenue this year, more than double that of last year,” CFO Colette Kress said on Wednesday night’s conference call. Among the reasons to like this business for Nvidia is that countries are their own kind of deep-pocketed customer with less immediate pressure to show a financial return. As Nvidia’s manufacturing capacity at Taiwan Semiconductor Manufacturing Co. grows, perhaps the company is able to divert more chips to these Sovereign AI customers without sacrificing supply for the likes of Azure. Taiwan Semi is Nvidia’s main manufacturing partner. Nvidia designs its chips and hardware, but does not have factories, or foundries as they are called in the chip industry. (Jim Cramer’s Charitable Trust is long NVDA, MSFT. See here for a full list of the stocks.) As a subscriber to the CNBC Investing Club with Jim Cramer, you will receive a trade alert before Jim makes a trade. Jim waits 45 minutes after sending a trade alert before buying or selling a stock in his charitable trust’s portfolio. If Jim has talked about a stock on CNBC TV, he waits 72 hours after issuing the trade alert before executing the trade. THE ABOVE INVESTING CLUB INFORMATION IS SUBJECT TO OUR TERMS AND CONDITIONS AND PRIVACY POLICY , TOGETHER WITH OUR DISCLAIMER . NO FIDUCIARY OBLIGATION OR DUTY EXISTS, OR IS CREATED, BY VIRTUE OF YOUR RECEIPT OF ANY INFORMATION PROVIDED IN CONNECTION WITH THE INVESTING CLUB. NO SPECIFIC OUTCOME OR PROFIT IS GUARANTEED.
