AI moat remains solid, is Broadcom Inc. (AVGO.US) on an upward trend worry-free?

In the wake of the recent financial report release, analysts still recommend holding Broadcom Inc. (AVGO.US) as its AI-driven moat remains intact, but in a cautious market, there is almost no room for error in its valuation. AI revenue is accelerating structurally, driven by collaboration with hyperscale enterprises, but customer concentration (especially Alphabet Inc. Class C) poses significant risks. Additionally, Broadcom Inc.'s high profit margins in infrastructure software are threatened by customer churn and migration trends. There are downside risks on the technical front, with $277 being a key support level; patience and caution should be maintained before increasing or initiating positions. Moat truly exists Broadcom Inc. reported total revenue of $19.3 billion in the first quarter, with the semiconductor solutions business accounting for 65%. This is important because in the previous fiscal year, the ratio was 58%/42%, with the primary driver of this change being artificial intelligence. AI revenue is continuing to accelerate, achieving growth for eleven consecutive quarters. Revenue is projected to reach $3.8 billion in fiscal year 2023, $12.2 billion in fiscal year 2024, $19.9 billion in fiscal year 2025, and already reached $8.4 billion in the first quarter of fiscal year 2026. This has structural significance as it involves a shift in the concept of computational infrastructure in hyperscale data centers. This shift involves the use of XPUs designed specifically for these companies' workloads, enabling more predictable memory access patterns, a fixed network topology, and ultimately models capable of executing millions of times while lowering total ownership costs by 40% to 65% at gigawatt scale. Broadcom Inc. collaborates with its customers to design these XPUs, and with Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR (TSM.US) to manufacture the architecture into chips without assuming manufacturing risks. Hyperscale data center operators are projected to invest $600-700 billion in capital expenditures, with 75% earmarked for AI infrastructure. Additionally, Broadcom Inc. anticipates that the AI chip market will exceed $100 billion by 2027. This is enough to demonstrate that any negative events would likely result from a slowdown in growth among hyperscale data center operators themselves, rather than due to a competitive disadvantage on the part of Broadcom Inc. Another critical aspect in this sector is the composition of AI revenue. Looking first at Alphabet Inc. Class C's TPU project, which is a core engine, particularly with the significant evolution of Ironwood, each successive generation of TPUs deployed by Alphabet Inc. Class C represents more orders, underscoring the expanding dependency relationship. Additionally, Anthropic recently confirmed plans to purchase as many as one million Ironwood chips for its Claude project. This suggests that Alphabet Inc. Class C is also starting to directly sell chip capacity to third parties. Meta Platforms (META.US) has confirmed that hundreds of thousands of MTIA chips developed in collaboration with Broadcom Inc. have been deployed in production for Facebook and Instagram. Over the next two years, the company plans to introduce four new versions, indicating another industry giant closely partnering with Broadcom Inc. in architecture design, a collaboration unlikely to be easily replaced. However, it is imperative to mention the less attractive aspect, as not all hyperscale data centers rely on the company. For example, Amazon.com, Inc.'s (AMZN.US) cloud service AWS collaborates with Marvell Technology, Inc. (MRVL.US) and AlChip to design the Trainium chip, while Microsoft Corporation (MSFT.US) collaborates with Maia. Therefore, Broadcom Inc. finds the most active customers for its XPUs in only Alphabet Inc. Class C, Meta, Anthropic, and OpenAI, indicating a concentration of AI revenue with Alphabet Inc. Class C leading the way. This is significant as it poses a related risk: if 40% or 50% of revenue in this sector were to shift to other suppliers or internal designs, this would impact financial models, rendering the current valuation completely disproportionate. Software business high profit margins, but earnings base under pressure Broadcom Inc.'s transformation post-VMware era is progressing smoothly within management's expectations. The software business's gross margin has reached 93%, operating profit margin exceeds 77%, and the sector generated $27 billion in revenue in the previous fiscal year. In the latest report, the business segment generated $6.8 billion in revenue, a 1.4% increase year-on-year, a stark contrast to the 26% growth rate witnessed in the previous fiscal year. If the operating profit margin remains constant, with remaining performance obligations (RPO) exceeding $45 billion and quarterly orders surpassing $9.2 billion, what is the reason behind this growth? This situation arises because previously, customers could purchase a component and use it indefinitely; however, with VMware's new business model, they must now pay annual subscription fees for the complete software package, which sometimes includes functionalities customers do not need, ultimately affecting billing calculations as they are no longer based on server counts. In other words, in a scenario where a company's infrastructure resembles the current situation with a server having 128 to 192 cores, the costs borne by Broadcom Inc.'s clients are based on these cores. As a result, the company's revenue multiplies without the need to add additional functionalities. Therefore, this 'forced' cycle has essentially ended, and the growth seen in this business segment in the previous fiscal year was largely a consequence of accounting treatment. Thus, this fiscal year, we are likely to witness the true level of growth in this business given the current rate of growth. Additionally, considering this rate, erosion of the revenue base must be taken into account. For instance, Nutanix reported migrations of 2700 VMware customers in the previous fiscal year, and Red Hat accumulated approximately $500 million in OpenShift virtualization contracts over two years. Furthermore, Gartner predicts that by 2028, 70% of enterprise customers will migrate 50% of their workloads. Hence, although the fees paid by retained customers are higher (evidenced by a 19% increase in annual recurring revenue), the earnings base is gradually shrinking. In response to this situation, management argues that VCF will become an indispensable component for AI; meaning that as enterprises expand AI workloads, they will require more VMware resources to offset customer churn. While this viewpoint is understandable, there is not enough evidence beyond VCF's technical integration with NVIDIA Corporation (NVDA.US) and AMD (AMD.US) to suggest that any hyperscale data center or leading AI company publicly commented on using VCF as a key infrastructure for their training clusters. It can be inferred that the real market should lie in the private cloud for enterprise inferences and, to a lesser extent, in edge clusters. Conclusion Although the infrastructure software business accounts for only 35% of the total revenue, its profit margins make it a critical lever for sustaining the company's profitability. Hence, if the rate of customer churn accelerates further, exceeding the levels described earlier, Broadcom Inc. will lose a significant growth driver. Has the market already priced this situation into the stock price? Alternatively, is this merely a temporary adjustment in growth trends? Technical analysis reflects risks For Broadcom Inc., the technical structure of its price indicates that the market has already digested some risks as the high and low-point sequence trends downward since mid-December of last year. Looking at the weekly chart, the 200-day moving average is beginning to flatten, which may already be indicating the likely direction of the primary potential trend. The primary concern lies in the short-term daily chart as the 30-day moving average (MA30) is forming dynamic resistance, with its slope continuously pressing prices downward. Furthermore, in the $320 to $350 range, volume has substantially decreased, marking the most significant drop in almost 15 months. This may lead to a significant number of sellers being trapped, who, to obtain liquidity, would suppress any rebounds. Therefore, at the current price level, unless there is a change in trend, $277 appears to be the most likely direction. This is typical in a bullish overall market. The issue lies in the risk that if prices fall below this level in a situation of volume loss, the next relevant trading range might be around $250. At present, caution is the best strategy. Conclusion Broadcom Inc. receives a short-term rating of "hold" as its moat remains intact, and AI revenue continues to grow steadily. One 'but' is the concentration of Alphabet Inc. Class C-related business, as this may pose a risk in an industry slowdown; however, there is currently no evidence supporting this or indications that Alphabet Inc. Class C will reduce its reliance on Broadcom Inc. Thus, the issues to be addressed are more related to industry cyclicality rather than this specific relationship. Another 'but' is the very high profit margins in infrastructure software, implying that while it accounts for up to 35% of revenue composition, the impact of accelerating customer churn, as described in the article, would significantly exceed the level shown by sales percentage. These factors are reflected in the current price trend and are the reasons behind the bearish market structure, further exacerbated by the current market environment, potentially leading investors to sell stocks after a rise in prices to hedge against potential larger market downturns and increase liquidity risks.