Broadcom on Thursday evening reported another strong quarter and better-than-expected guidance for the current quarter. Nonetheless, the Club stock gave up its initial pop and traded sharply lower as the Q & A session of the post-earnings conference call kicked off. Investors were apparently not satisfied with CEO Hock Tan’s answer to an important question. Revenue in the fiscal 2025 fourth quarter, which ended Nov. 2, increased 28% year over year to $18.02 billion, ahead of the $17.49 billion consensus forecast, according to the consensus of analyst estimates compiled by LSEG. Adjusted earnings per share increased 37% to $1.95, also outpacing expectations of $1.86, LSEG data showed. Adjusted EBITDA , or earnings before interest, taxes, depreciation, and amortization, grew 34% to $12.22 billion in the quarter, beating the FactSet consensus of $11.61 billion. Why we own it Broadcom is a high-quality semiconductor and software company run by the incredible CEO Hock Tan. The company is a big AI beneficiary through its networking and custom chip businesses. Competitors : Marvell Technology, Advanced Micro Devices , and Nvidia Last buy : Nov. 21, 2024 Initiation date : Aug. 24, 2023 Bottom line The reported results were solid as revenue outpaced expectations, thanks to strength in both of Broadcom’s operating segments: Semiconductor Solutions and Infrastructure Software. Profit margin performance was also strong as the company’s overall adjusted operating income margin expanded nearly 350 basis points, or 3.5 percentage points, leading to strong year-over-year earnings growth, beyond what the Street was looking for. Alongside the strong results, revenue and EBITDA margin guidance for the current fiscal 2026 first quarter were both ahead of expectations as well. Before addressing the part of the call that knocked the stock, we want to stress that, overall, Tan’s remarks got us really excited for 2026. For starters, the CEO confirmed the rumors that the fourth customer we heard about last call, which placed a $10 billion order, is indeed Anthropic, and that they’re buying the Ironwood XPUs, the generation seven TPUs on which Google’s Gemini 3 was trained and run. XPU is the term Broadcom uses to describe custom chips, which are also referred to as application-specific integrated circuits (ASICs). Tan also noted that these TPUs are being used by others, including Club name Apple , Cohere, and SSI, adding that the “scale at which we see this happening could be significant.” TPUs, or tensor processing units, are what Google calls the chips that it co-designed with Broadcom. In a “what have you done for me lately” business, Tan also noted that in the reported quarter, privately held Anthropic doubled down, placing an additional $11 billion order for delivery in late 2026. If that’s not enough, Tan said Broadcom secured a $1 billion initial order from a fifth, yet-to-be-named XPU customer, also for delivery in 2026. It was noted on the call, however, that in the back half of fiscal 2026, there could be some margin pressure. CFO Kirsten Spears said, “[In] the second half of the year, when we do start shipping more systems, the situation is straightforward. We’ll be passing through more components that are not ours. … Those costs will be passing through more costs within the rack. And so those gross margins will be lower.” So, that brings us back to the question: Why did a stock, which initially jumped over 3% on the release, proceed to give up the gains and reverse lower by 4.5% in the after-hours session? It’s about concerns regarding the long-term partnership between Broadcom and Google-parent Alphabet , and maybe that back-half margin talk. The Q & A part of the call kicked off with a question about XPU customers possibly looking to bring more development in-house and what that might mean for Broadcom in the coming years. Tan responded by discussing the benefits of custom semiconductors, noting that what can be built into purpose-designed hardware would only be possible to code via software with other solutions. He then went on to opine, saying, “Now, will that mean that over time they all want to go do it themselves? Not necessarily. And in fact, because the technology in silicon keeps updating, keeps evolving. And if you are an LLM [large language model] player, where do you put your resources in order to compete in this space, especially when you have to compete at the end of the day against merchant GPUs, which are not slowing down in the rate of evolution. So, I see this concept of customer tooling as an overblown hypothesis, which frankly, I don’t think will happen.” Customer tooling refers to the idea that companies look to develop their own, in-house designed, custom hardware accelerators for AI training and inference without the help of Broadcom. Tan’s reference to GPUs, or graphics processing units, was meant to highlight the competitive landscape that customer chips face from these gold-standard all-purpose chips, dominated by Club name Nvidia . Sellers of stock may have taken Tan’s remarks to be a bit dismissive and not quite the concrete “it’s not happening” answer they had been hoping for. That said, we appreciate Tan because he provides a no-nonsense view of things, regardless of what he thinks Wall Street wants to hear. At the moment, this hypothesis is indeed nothing more than speculation, and Tan was, in our opinion, clear in his view that he doesn’t see this scenario playing out. In the end, Thursday’s after-hours selloff was more about investor concern with a potential bearish scenario in the future, in which key customers move development in-house, rather than anything clear-cut that would impact Broadcom’s business outlook. It’s an understandable concern, after all, we have seen those with the financial ability to do so, look to move more chip development in-house. However, it is nothing more than speculation at the moment and, in our opinion, not nearly enough to get out of our position, given the clearly strong demand that Broadcom is now seeing and expects to see increase as we work our way through 2026. If the margin commentary was why the stock was down, it’s an opportunity because at the end of the day more business, even at a lower gross margin, means more earnings growth. And that is what we value the stock based on. AVGO YTD mountain Broadcom YTD That said, even just the possibility of hiccups down the road was enough to drive a move lower in the stock when investors are sitting on huge gains, especially in the middle of December, and looking to book profits before year-end. Broadcom shares, as of Thursday’s close, were up 75% year-to-date, and trading right around all-time highs coming into the print. This decline doesn’t strike us as anything more than that. Out of respect for this year’s rally, we’re reiterating our 2 rating hold on Broadcom stock and will look for a better opportunity to upgrade it to our buy-equivalent 1 rating should this selloff persist in the coming sessions. We are, however, raising our price target to $425 per share from $415, as Wednesday’s record-high close of nearly $413 was bumping up on our previous PT. Segment commentary Broadcom’s fiscal fourth-quarter revenue in Semiconductor Solutions, the much larger of the two operating segments, increased 34.5% year over year to $11.07 billion, exceeding expectations of $10.77 billion, according to FactSet. Within that result, AI semiconductor revenue surged 74% year over year to $6.5 billion, ahead of the $6.22 billion the team guided to months ago after its fiscal Q3 release. AI networking was again strong, with Tan noting that customers continue to build out data center infrastructure before they deploy AI accelerators. As a result, the backlog for AI switches now exceeds $10 billion, with the CEO adding that the Tomahawk 6, which he considers unmatched in its capabilities, is seeing bookings come in at record rates. Adding in the other components necessary to build out an AI data center, including XPUs, and Broadcom is looking at an AI-related backlog of more than $73 billion — about $53 billion of which is XPUs. Tan expects the team to convert that into realized revenue over the next 18 months, with $8.2 billion expected to be realized in the current fiscal 2026 first quarter. Regarding the legacy semiconductor sub-unit, fiscal Q4 revenue of $4.6 billion represented a 2% year-over-year increase and 16% sequential increase, “based on favorable wireless seasonality,” Tan said. That seasonality he’s referring to is the launch of the iPhone 17, which has been met with solid demand. Tan added that broadband revenue continues to recover, wireless was flat versus the year-ago period, and enterprise remains under pressure as “spending continued to show limited signs of recovery.” In Broadcom’s other operating segment, Infrastructure Software , revenue grew about 19% year over year to $6.9 billion, ahead of the $6.72 billion consensus estimate, according to FactSet. On the call, Tan said, “Bookings continued to be strong as total contract value booked in Q4 exceeded $10.4 billion, versus $8.2 billion a year ago.” As a result, the software infrastructure backlog ended the quarter at $73 billion, a major increase from the year-ago $49 billion. Guidance For its fiscal 2026 first quarter, which will end on Feb. 1, Broadcom forecasted total revenue to be about $19.1 billion. That target is ahead of the $18.27 billion LSEG consensus. Importantly, AI revenue is expected to keep growing in the coming quarter, with Tan stating in the release, “We see the momentum continuing in Q1 and expect AI semiconductor revenue to double year-over-year to $8.2 billion, driven by custom AI accelerators and Ethernet AI switches.” Add in the legacy semiconductor business forecast of approximately $4.1 billion, and we get a Semiconductor Solutions segment guide of about $12.3 billion, well ahead of the $11.53 billion consensus forecast, according to FactSet. The $6.8 billion Infrastructure Software revenue guide for fiscal Q1, however, came in short of the $7.136 billion estimates from FactSet. The company expects fiscal Q1 adjusted EBITDA to be approximately 67% of projected revenue, or $12.78 billion, ahead of the 66% profit margin and $12.06 billion consensus estimate, according to FactSet. (Jim Cramer’s Charitable Trust is long AVGO, AAPL, NVDA. 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.