Sometimes the stakes are so high, the degree of difficulty so immense, that it simply may be too hard to game. When that’s the case, no amount of formal research will help you fathom the stock implications. Yet, you have inherited the issues and they must be dealt with — or you are too at sea to judge them. We have not one, but two situations — and potentially three — that concern me especially because the price-to-earnings multiples are very high. The two stocks in question? Broadcom and Costco . Broadcom, the nervous system for many of the hyperscalers, is trying to encroach upon fellow Club name Nvidia , the leading AI chipmaker whose fast processors are at the heart of so many artificial intelligence data centers. Let’s take Broadcom first. For its custom AI chip business, Broadcom’s list of clients include Alphabet -owned Google, Meta Platforms , TikTok parent ByteDance, and OpenAI . Additionally, AI startup Anthropic also was recently revealed as a $10 billion customer . Meanwhile, Broadcom is rumored to be talking with Microsoft about shifting its business away from its director competitor in the custom chip design space, Marvell Technology . And there were also concerns that Marvell was losing some business from Amazon. Importantly, Marvell CEO Matt Murphy, whom I trust implicitly, came on “Mad Money” and said he hadn’t lost any business. I believe him. At the same time, Bloomberg News on Friday reported that Oracle pushed back the opening date for some of the data centers it’s building for OpenAI, the giant startup run by Sam Altman. OpenAI happens to be committed to spending $300 billion over five years for computing power from Oracle. That figure is thought to be rock solid because it is in Oracle’s RPO, or remaining performance obligations. It represents more than half of Oracle’s $523 billion RPO. Anything that indicates that OpenAI is not money good could cause a tremendous negative ripple for this entire ecosystem — not just OpenAI, although OpenAI is at the center of the debate. According to Bloomberg, the timeframe for the pushout is from 2027 to 2028, with labor and material shortages cited as the reason. Importantly, Oracle said in a statement there have been “no delays to any sites required to meet our contractual commitments, and all milestones remain on track.” Oracle is to be trusted because it is Larry Ellison’s company and Ellison doesn’t make false claims. But is Sam Altman to be trusted? We don’t know enough about him and his company is private. Bloomberg could be wrong in its story. But maybe it isn’t. Many took the story as gospel despite Oracle’s response in that statement. It is possible, however, for everyone to be right. We know from Coreweave’s quarterly report that these sites can have problems being built . They are very complicated and companies are all fighting for the same components. Oracle holds itself out as an expert in building them. What happens, however, if Oracle has problems building the data center sites for OpenAI and that is the source of the pushout? What happens to the pace of chip orders from Nvidia, which is almost always a part of every data center? These are the fundamental questions that must be answered. We thought we would get some clarity on the broader state of the AI buildout when Broadcom reported quarterly results Thursday night. But the answer was obscured by an issue identified by CFO Kristen Spears on the Broadcom conference call. At the beginning of the call, Broadcom said it had some $73 billion in AI backlog, including orders for its AI server systems that contain its custom chips and other components. That number excited Wall Street and initially drove the stock up about $15 a share in after-hours. But later on the call, Spears said the AI system business was less profitable than other chip-only orders because of some pass-through costs with lower margins. When Spears revealed that, Broadcom’s stock did the dreaded pirouette and it fell to about $380, giving up a frightening $35 from its overheated after-hours level. When that happens it’s a nightmare, which is why the stock fell even more during Friday’s regular session and ended the day at $359.93. Some of that additional decline came from the first issue I mentioned, the possible delay related to Oracle’s work for OpenAI. The rest was from the pass-through issue. AVGO YTD mountain Broadcom’s year-to-date stock performance. Now let’s go back to the first issue. I never like to be in a battleground because the possible results are too murky. These issues created their own battleground. They can’t really be resolved because OpenAI is private. When we hear about potential delays involving OpenAI, even if other reasons are cited in the article, we can’t help but wonder whether it will have the money to meet all its obligations in the coming years. How do we know if Broadcom’s business is not as robust as we thought? We do know OpenAI has access to $40 billion in capital , or at least that it says it has that access. We do know that it just landed a billion dollars from Disney for a stake in the company. It was all very odd. Why didn’t OpenAI have to pay Disney and not vice versa? Was it really about making sure OpenAI was able to get the characters for its AI video generation tool Sora and while blocking Google? Still, I found the deal murky and very similar to the kinds of crazy deals I heard about in 2000, deals that everyone told me were smart and I thought were preposterous. All of this is very theoretical. I don’t like theoretical. Who wants to be caught in this web of intrigue? Not me. Not anyone else. Hence the collapse in Broadcom’s stock. I can go round and round about how OpenAI is worth more than we thought because of this business-to-business deal. Enterprise business is worth more than business-to-consumer deals, the current focus model of OpenAI. That’s more like the aforementioned Anthropic, whose heavyweight investors include Amazon, Microsoft and Google. Anthropic is loved by the Street. OpenAI is not as trusted because of the craziness we have seen from the firm, including CFO Sarah Friar’s odd comment that the government could always “backstop” the company . That’s been denied later on by Friar, but it’s kind of a genie-out-of-the bottle comment. Again, it’s all too hard. Which means that Broadcom’s stock is worth less than we thought, at least around this one issue. Once again, we have to play a game of “who do you trust?” I trust Hock Tan, the longtime CEO of Broadcom, which means you shouldn’t be bailing from the stock. Others clearly have less faith, or else the stock wouldn’t have come down a horrendous $46, or 11.4%, on Friday. This is not the first time Hock has been doubted by the market. It is also not the first time that the market has been wrong. I am with Hock. We are, of course, standing by Broadcom, which even with Friday’s pullback remains up 55% year to date. However, because its price-to-earnings ratio is so high, at almost 42 before earnings, there’s not much room for error. That’s just how it goes with high-multiple stocks. So, it’s fraught and we don’t like fraught – the battleground. We do think Hock will be right, just as we did a few years ago when the stock broke down after another quarter and it turned out to be a false worry accompanied by a huge amount of insider buying . Could that happen again? I think so. We just don’t know yet. To sum up, my judgment is that Broadcom is fine, but the position is a lot harder to defend at this moment. We will defend it by owning it, not buying more. Now, let’s cover Costco. The first, and most salient, issue is the P/E multiple, and yes it almost always comes back to the multiple. At 43 times next year’s earnings, it is high versus the S & P 500, which trades at roughly 22 times forward earnings. But Costco’s valuation being well above the market is not unusual historically speaking. In fact, at this time a year ago, Costco’s P/E ratio was north of 50 while the S & P 500’s was still around 22, according to FactSet data. Costco’s multiple coming down would be fine if the stock weren’t near its lows. What we’re seeing now indicates that the fear is the stock must go lower because investors are not as willing to pay up as much for future earnings — that is how you get multiple compression. To be sure, Costco’s quarter was solid, in line with the estimates. But it wasn’t better than the estimates. The earnings have to be better than the estimates to maintain its high multiple: witness Walmart with a 40 multiple, as close to Costco as I can recall. That, in and of itself, is telling. Why is that? There’s a couple of reasons. First, customers aren’t renewing their membership as they used to. We have seen this impact for several quarters, and it is quite unusual. The company has an excuse: These are mostly younger people who get their membership online versus warehouse signups. But I don’t care about the excuse. It is a red flag. Further, there was “lumpiness” to the quarter, something I don’t like but I think got better as the quarter ended. Third, Costco CFO Gary Millerchip – a rather new hire from Kroger brought in to replace the irreplaceable long-time finance chief Richard Galanti – once again used the word “choiceful” about the consumer. Choiceful, I think, is a code word for “too expensive.” I don’t associate that word with Costco. Suboptimal. COST WMT YTD mountain Costco’s year-to-date stock performance versus Walmart. So, what to do? As I said during a Morning Meeting last week , I am very concerned about this and about how the analysts seem to be focused more on Costco’s technology initiatives – although two analysts on the conference call did ask on about the renewal rate, and the answer was that thanks to some targeted initiatives, the renewal rate for online members will be a little better than it was. That wasn’t reassuring. Unlike Broadcom, the high multiple here can’t stay high when the comparable sales aren’t much better than expected. This distresses me. I have been kind of possessed by it. Is Walmart catching up? Is Walmart passing it? Is Walmart better than Costco? Comparisons, as I know from my mother, are odious. But it’s a real worry. I was thinking Costco would go up when it reported that quarter because there was progress with online and there was additional talk about bolstering its advertising initiatives like Walmart has. But they are so far behind, that, again it is worrisome. I play with an open hand. I weigh all of this against a long history of being special. I don’t think it helps that Costco is tussling with the administration over tariffs and, before that, diversity efforts . Whether you like or agree with Costco, you have to accept that some people might be turned off by these stances. As a shareholder, I am not happy about this because I am trying to figure out the real reason why there is a lower renewal rate, especially among young people. Why be as concerned as I am? Because of Target , that’s why. Many stuck with Target long after things went awry there. It’s retail. Retail is one of the hardest businesses. You can slip. You can fall. That’s why you should not be surprised if we take action on the position. I hate to ever sell this stock. The company is so amazing. My trips to the stores remain exciting. But we can’t afford a Target. We just can’t. That said, you have to expect some action. I can’t lose sleep over this one. So, sigh. It’s not what we want. But it almost has to happen. (Jim Cramer’s Charitable Trust is long COST, AVGO, NVDA, AMZN, MSFT and META. 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.