Broadcom’s latest quarterly results indicate rising demand for custom AI chips and data center networks, with 18 months of lag visible, even as investors question margin pressure and economies of scale in silicon.
Image: https://www.abnewswire.com/upload/2025/12/fce5456dc838b0fe7ecbf8601c3f42bb.jpg
As markets eye the next phase of the AI investment cycle in December, Carvina Capital Pte. Ltd. points to Broadcom’s fourth-quarter numbers as a signal that spending is hardening into contract-backed demand, with group revenue up 28% compared to the same quarter a year earlier.
Broadcom reports revenue of $18 billion for the quarter, higher than the consensus of $17.5 billion for the same period, while adjusted earnings come in at $2 per share, versus $1.9 expected. The stock is up 2.3% in after-hours trading immediately after the release and is up about 122% over the trailing twelve month period, a reminder that the market is already pricing in a significant amount of optimism.
The AI semiconductor activities remain the engine of growth. AI-related chip revenue this quarter is $6.5 billion, up 74% from the corresponding quarter a year earlier, while the company announces a $73 billion AI backlog scheduled to ship over the next 18 months. At the end of the quarter, the AI backlog accounts for nearly half of the total $162 billion backlog, underscoring the extent to which data center buildouts and network constraints are driving demand.
A research note from Carvina Capital emphasizes visibility rather than exuberance, with Private Equity Director Peter Jacobs arguing that “a backlog of this size matters because it turns the AI story into a contracted pipeline, giving investors a timetable rather than a slogan”. For corporate finance teams, the distinction is important: capital expenditures more easily survive higher financing costs when orders are anchored in multi-year implementation schedules.
Future expectations now depend on whether these orders are smoothly converted into delivered revenue. Analysts forecast revenue of about $96 billion for the next fiscal year, roughly 50% higher than revenue generated in the previous 12 months, a pace that eclipses Broadcom’s average annual growth of about 20% over the previous five fiscal years. Management guidance indicates that AI semiconductor sales will be approximately $8.2 billion in the upcoming fiscal first quarter, and Jacobs describes the strategic bet as “a scale competition that takes custom silicon from a niche product to industrial production, without relinquishing control over cost, delivery and quality.”
This shift in production brings its own tensions. Broadcom signals that consolidated gross margin in the coming quarter will likely decline by about 100 basis points from the immediately preceding quarter as lower-margin AI products account for a larger share of sales, a reminder that customers buying at hyperscale will have to negotiate hard. Investors are responding by scrutinizing the quality of earnings as closely as they are examining overall growth, and Jacobs characterizes the test of the market as “whether volume growth occurs faster than margin dilution, because revenue only deserves a premium if profitability accompanies it.”
Broadcom’s customer strategy reinforces this point. The company relies on long-standing design relationships where cloud operators and model developers co-create accelerators, interconnects, and circuits for their own data centers, tying performance to specific software stacks while locking in purchases for multiple years. Reports of consecutive orders worth $11.2 billion and $12.3 billion over the two most recent quarters, with delivery schedules extending through the end of next year, illustrate how quickly spending scales once a platform standardizes on a chip architecture.
Software provides the other counterbalance. With VMware integrated, the software unit generated $6.6 billion in revenue in the second fiscal quarter and represented about 36% of quarterly revenue, with recurring revenue rising above 85% of software revenue in the last reporting period as subscription licenses increase. In a cash-strapped world, predictable cash flows are an advantage, although customer pushback on pricing and flexibility remains a current issue as contracts are renewed.
As central banks keep policy restrictive and boards apply tougher tests to investment projects, the question is not whether AI will transform corporate computing, but whether the economics of that transition remain attractive to suppliers and their shareholders. Carvina Capital notes that Broadcom’s combination of backlog, customer concentration and recurring software revenue will require close monitoring in the coming quarters, and Jacobs summarizes the calculation as: “The market pays for visibility and discipline, and punishes any indication of AI becoming a volume business without pricing power.”
About Carvina Capital
Carvina Capital Pte. Ltd. (UEN: 201220825D) is a Singapore-based company founded in 2012. It focuses on research-driven, long-only public equity strategies for institutional and professional markets and evaluates products that could be accessible to retail investors. The investment approach combines fundamental research with disciplined risk management, designed to compound capital over full market cycles.
Further information is available at https://carvina.com [https://carvina.com/].
Media contact
Company Name: Carvina Capital Pte. Ltd.
Contact person: Huacheng Yu
Email: Send email [https://www.abnewswire.com/email_contact_us.php?pr=carvina-capital-broadcom-sees-strong-aidriven-growth]
Country: Singapore
Website: https://carvina.com
Legal Disclaimer: The information on this page is provided by an independent third party content provider. ABNewswire makes no guarantees, responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality or reliability of the information contained in this article. If you have any complaints or copyright issues related to this article and wish to have this article removed, please contact retract@swscontact.com
This release was published on openPR.