Qualcomm (QCOM.US) 2025 fiscal year Q4 earnings call: Details and performance indicators of AI chips will be announced early next year.

Recently, Qualcomm (QCOM.US) held a conference call to discuss the results for the fourth quarter of fiscal year 2025. Qualcomm stated that they are very excited to enter the data center field, and believe it is a "new chapter" in the company's diversification expansion. The core competitive advantage of the company lies in energy efficiency, generating the most tokens with the least power consumption. The company is focusing on two main aspects: having a highly competitive high-efficiency CPU that can be used for AI cluster head nodes and general computing, and building a brand new architecture designed specifically for AI inference. The development of the AI 200 and AI 250 chips and acceleration card solutions align with this strategy, and progress is satisfactory. More details and performance metrics will be announced early next year. Regarding customer collaboration, the company has indeed had discussions with a "very large scale" customer and are very satisfied with the results. More details of this collaboration will be updated as part of the company's future data center roadmap release. The market clearly demands competition, and Qualcomm will continue to prove itself with strong product performance. Q&A: Q: First, could you talk about Qualcomm's core competitive advantage in the data center field? Secondly, apart from the information already disclosed in the press release, could you provide more details about the specifications of the AI 200 and AI 250 chips? Lastly, you had mentioned collaboration with a "very large scale" customer last quarter. Is this different from the collaboration with Humain announced this time? And if so, do you have any updates on the timeline of the collaboration with that large-scale customer? A: We are very excited about entering the data center field, and believe it is a "new chapter" in the company's diversification expansion. Our core competitive advantage lies in energy efficiency, generating the most tokens with the least power consumption. We are focusing on two main aspects: having a highly competitive high-efficiency CPU that can be used for AI cluster head nodes and general computing, and building a brand new architecture designed specifically for AI inference. Our AI 200 and AI 250 chips align with this strategy, and progress is satisfactory. More details and performance metrics will be announced early next year. Concerning customer collaboration, we did have discussions with a "very large scale" customer and are very satisfied with the results. More details of this collaboration will be updated as part of our future data center roadmap release. The market clearly demands competition, and we will continue to prove ourselves with strong product performance. Q: Recently, there have been rumors in the market suggesting that your biggest Android customer (referring to Samsung) may plan to use more of their own baseband chips. Do you have any visibility on the business with that customer? And how do you expect your share with that customer to change in the next year or so? A: First, the high-end Android phone market is continuously expanding, bringing healthy growth to our Snapdragon platform. Even in the overall stable phone market, our Android business can continue to grow by increasing content and average selling price (ASP), mainly due to the expansion of the high-end market. Regarding our relationship with Samsung, we have stated that the new collaboration benchmark is 75% share, which is also our financial forecast assumption for the future. While sometimes due to outstanding execution, we may gain a higher share (such as 100% on Galaxy S25), our planning and assumptions for future models (like Galaxy S26) will always be based on that 75% share. Q: We have heard that the installation cost of mainstream AI training clusters is approximately $300 to $400 per watt. In comparison to this number, what cost or cost-effectiveness target do you expect when deploying AI 200 or AI 250 chips in inference scenarios? In other words, can you help us understand where Qualcomm's cost advantage lies in deployments of similar scale? Additionally, concerning collaboration with Humain to deploy 200 megawatts of computational power, what income contribution do you expect that to bring? A: I will provide as much information as possible without revealing too many details that will only be announced early next year. First, about revenue expectations, we originally anticipated significant revenue growth for the data center business starting from the 2028 fiscal year. But thanks to the collaboration with Humain and our progress on AI accelerators, we are now bringing that timeline forward by one year to the 2027 fiscal year, where the data center business revenue will start to become considerable. Second, we are gaining broad market attention. In the context of current data centers facing constraints on power consumption and computing density, many companies requiring large-scale deployment of AI inference power have shown strong interest in our solutions. It is certain that if our solutions were not competitive, there would not be so many in-depth conversations. However, specific platform KPIs will be detailed further when we update our roadmap early next year. Q: The market still seems very concerned about the performance in the March quarter next year, mainly due to changes in share with that major Android customer. Historically, Qualcomm's smartphone business has seen high single-digit declines (e.g., 7-9%) from the December quarter to the March quarter. Considering the decrease in share, does this historical quarter-over-quarter decline trend still apply? Or should we expect a different decline? A: We are not providing specific guidance for the first quarter yet. However, as you have seen, the strong business momentum we exhibited at the end of the 2025 fiscal year is already reflected in our recently released performance and guidance for the December quarter, which is expected to continue throughout the fiscal year. The only additional note is that we expect to complete the acquisition of Alphawave in the first quarter of the 2026 fiscal year. Besides that, as you mentioned before, our business momentum remains strong. Q: You mentioned that the performance in the September quarter was better than expected, mainly driven by the high-end Android business, but it actually seems to be more so from your largest customer (Samsung). You had previously hinted that due to share decline, the revenue from that customer would decrease by around $5 billion, but the financials show a slight year-over-year decline. Can you explain this difference? Also, as part of this question, can you disclose the baseline assumption for the amount of business with that customer in the December quarter? A: Our initial expectation was that we would have three out of four models in the new phone released, and the reality is indeed that. The actual share of course depends on market sales. Regarding the September quarter, the guidance we provided already accounted for demand from that customer (Samsung). Therefore, the unexpected growth in performance in that quarter was not from Apple Inc., but driven by other Android customers, particularly those adopting our new Snapdragon chips. Looking forward to the December quarter, we expect the revenue from QCT's smartphone business to achieve approximately 11%-13% (low teens) quarter-over-quarter growth. This growth is also mainly driven by the Android business. While Apple Inc.'s business also contributes, the core DRIVE of the quarter-over-quarter growth is indeed the shipment volume of high-end Android devices. Q: Can you provide any updates on the patent licensing negotiations with Huawei? It seems like these negotiations have been going on for some time now. Can you discuss the current status of the situation? A: We currently do not have any new progress to disclose. Discussions are still ongoing, but we have no more substantial information to share at this time. Q: Excluding the Apple Inc. business, QCT revenue grew by 18%. Even if we exclude the automotive and IoT businesses, it is clear that the Android smartphone business achieved very strong double-digit year-over-year growth. How much of this revenue growth is driven by the increase in the average selling price of Snapdragon chips themselves? It is evident that as you adopt more advanced processes, wafer costs are also rising. Could you discuss how the increase in chip ASP affects your current and future growth in the smartphone business? Additionally, how is the entire smartphone industry digesting and absorbing this ASP increase? A: We have observed a long-standing trend that we discuss almost every year. There is a strong demand in the market for flagship chips that are more powerful and capable. This demand stems from two factors: intense competition among smartphone manufacturers and consumers' desire for more functionality on their phones. We have clear plans for our future generations of chips and have entered deep discussions with customers, so we are very confident that this trend of increasing value per device will continue in the coming years. Additionally, it is important to remember the second critical factor I emphasized earlier: the shift in devices consumers are buying towards higher-end models. This does not refer to the increase in value within the flagship segment but rather the fact that more consumers are choosing to buy higher-grade smartphones. This is another long-term trend we have observed for years and expect to continue. Q: The strong performance from Snapdragon Android business in the September and December quarters seems to be mainly coming from the Chinese market. Is this simply due to the timing of new device releases, or are there instances of customers pulling purchases forward? For this strong momentum, should we consider any potential risks or special factors? A: No, there are no instances of customers pulling purchases forward. We have seen that almost all major Chinese customers have already released new models, and the initial market response to these devices has been very positive. Additionally, later this quarter and into the beginning of next year, we will also see many global customers releasing new devices. Therefore, the strong momentum you see reflects the normal purchasing patterns after new device releases and the enthusiastic initial response from consumers to these new products. Q: According to the information in the press release, your product architecture, such as using DDR memory and PCI Express interface, appears to be different from the solutions adopted by other market participants. How should we interpret this? Is this just the "first step" for Qualcomm entering this market, with more product offerings to follow? Or does this represent a completely different philosophy in attacking the marketlike you mentioned, focusing more on energy efficiency? In summary, is this a differentiation strategy that is completely different from the current market mainstream? A: Yes. Our approach to this issue is thinking about what the "future AI architecture should look like." This approach applies not only to the data center but also to the edge computing field where we have already established a presence. When we focus on building dedicated inference clusters, our goal is very clear: to achieve the highest computing density at the lowest cost and energy consumption, to efficiently generate tokens. Under this goal, we believe there is an architecture superior to the traditional combination of GPU and HBM memory. This is what we are diligently developing towards. We must successfully execute this strategy, as it is currently the core focus of the company. Q: How much of the growth you see in the smartphone business is driven by the increase in the average selling price of Snapdragon chips themselves? It is clear that as you adopt more advanced processes, wafer costs are also rising. Could you discuss how the increase in chip ASP influences the current and future growth of your smartphone business? Furthermore, how is the entire smartphone industry absorbing this increase in ASP? A: There is a long-standing trend we discuss almost every year. The market shows a strong demand for flagship chips that are more powerful and capable. This demand stems from two factors: fierce competition among smartphone manufacturers and the desire from consumers to have more functionality on their phones. We have clear plans for our future generations of chips and have entered deep discussions with customers, so we are very confident that this trend of increasing value per high-end device will continue in the coming years. Q: QCT revenue grew by 14% this quarter, and based on your guidance for the next quarter, it seems that growth expectations for the QCT sector are forecast to outperform the market by another 600 basis points. Could you please analyze the components of this "outperformance" for next quarter in more detail? Can you speak separately about the expected performance of the smartphone, IoT, and automotive business segments? Compared to the initial expectations for the last quarter, which parts do you think will perform better? Can you provide an overview of the growth drivers for the various business segments of QCT next quarter? A: Automotive Business: We achieved a revenue record of approximately $1.1 billion in the September quarter. For the next quarter, we expect revenue to remain flat or slightly increase. With more new vehicles featuring our technology entering the market, we are in a very favorable position and expect continued growth throughout the fiscal year. IoT Business: The situation is similar. The performance in the September quarter has significantly exceeded our previous guidance, and we expect revenue from the IoT business to continue growing from the first quarter onwards in the coming fiscal year. Smartphone Business: The growth expectations you see in the December quarter are mainly due to the successful release of our next-generation flagship chips. All major smartphone manufacturers (OEMs) have launched devices with this chip and have received strong market responses, which is reflected in our financial forecast. As I mentioned earlier, we expect QCT's smartphone business revenue to achieve "low double-digit" quarter-over-quarter growth. Q: Historically, Qualcomm's first and second fiscal quarters tend to be the strongest quarters of the year, given new product launches in the Chinese market and with the consumer purchasing season of New Year and Chinese New Year. From a seasonal perspective, what trend can be expected for the following quarters? A: We expect the first and second fiscal quarters of the company to be stronger quarters of the year. Usually, the third fiscal quarter in June is considered a relative low point. Therefore, regarding the smartphone business, the seasonal trend should remain consistent with what you have observed in previous years. Q: QCT revenue increased by 5% year-over-year this quarter, but the profit margin decreased by over 100 basis points. I would like to understand the reasons behind this decline: is it mainly due to changes in the product portfolio, or is it because of rising manufacturing costs? Or is it simply a result of increased R&D investment by the company to drive future revenue growth? A: When looking at the year-over-year trend of profit margins, it is important to understand that we are making significant investments in the data center field. Over the past few years, we have been transitioning operational expenses from mature businesses to high-growth areas. Now, the data center business is an additional investment focus in our overall investment blueprint, which is affecting our current profit margin performance. Q: Further focusing on your non-Android smartphone business, do you have any new guidance or information to share on how we should predict the performance of this business in the 2026 calendar year? A: Our view on the share we hold in the Apple Inc. business remains unchanged and is fully consistent with what we have previously stated. Q: I remember that the XR business was set at $20 billion, it was the smallest in scope among the opportunities listed at that time. If you plan to hold an event specifically for the data center business and provide updates, does this mean that the potential scale of the data center business will be significantly larger than what was shown at the last Analyst Day? Can we expect it to become another significant opportunity of several billion dollars? Could you clarify this? A: We see that the market, especially the AI smart glasses market, is taking off very rapidly. Therefore, we do see that the current development is significantly ahead of the previous guidance provided, and we see enormous potential for growth. More broadly, if we consider personal AI as a broader opportunitywhether through glasses, watches, or ear-worn devices in various formsthis could potentially evolve into a very large market. If the market develops as expected, it will create substantial upward business potential. It can be confirmed that the potential scale of the data center business will be higher than that figure. We believe that if successful in this field, it has the potential to bring us a several billion-dollar level of incremental revenue opportunity over the next few years. This is our current view on this matter.