Today, I looked at a UBS analysis report on Broadcom $AVGO and was stunned. According to UBS's forecast, by 2028, Broadcom will become a company worth at least 5 trillion US dollars! 🧐
To be honest, this pie is drawn a bit too large. I will break it down for everyone and calculate how much true gold and silver this shipment of 35 million TPUs can bring to $AVGO by 2028! 🧮
UBS provided data on the TPU market in the report. With 35 million TPUs, Broadcom conservatively estimates a 60% market share, which is about 20 million devices, with a selling price of 15,000 US dollars each. This conservatively estimates TPU revenue at 300 billion.
Additionally, with the surge in ASICs, the market demand for AI network infrastructure will explode, leading to increased corresponding demand for Tomahawk 6/7 network switches and PCIe switches. It is expected that the network communication sector will contribute about 40 billion US dollars.
Traditional business + VMware software: Assuming steady annual growth or stability, it will contribute about 35 billion US dollars.
📌 Here, total revenue will astonishingly reach 375 billion US dollars, which is five times that of the current fiscal year 2025.
Based on this year's Q1 net profit margin of 42%, the estimated net profit in 2028 will reach an astonishing 157.5 billion US dollars. What does this mean? For the fiscal year 2026, Nvidia's total net profit was only 120 billion US dollars.
Evaluating Broadcom at a PE ratio of 41, similar to Nvidia, would bring its market value to an astonishing 6.45 trillion US dollars. Even using a conservative PE of 30, it would reach 4.7 trillion US dollars, which is 2.6 times the current market value! 🧐
However, I always feel that something is amiss. This pie is too big and could easily choke you. I'll talk about some major pitfalls that may exist here. 🚨
1️⃣ Supply Chain Bottleneck:
The growth of TPUs from 4.3 million to 35 million represents an eightfold increase in just two years, which will pose a severe test to the supply chain capacity. TSMC's capacity, CoWoS packaging technology, HBM memory supply (Micron, SK Hynix, Samsung), each link must keep up. Right now, Nvidia's GB300 and VR200 are vying for production capacity, and whether AVGO can secure enough wafer allocation is a big question.
2️⃣ Customer Concentration Risk:
AVGO's main customers for ASICs are few:
• Google (TPU)
• Meta (MTIA)
• OpenAI (reportedly under development, cooperating with the Nexus 10GW project)
• Anthropic (3.5GW agreement)
If these major customers cut orders or delay orders, AVGO’s ASIC business would immediately collapse. Moreover, these tech giants are frantically spending money on AI, and if the economy goes down, capital expenditures will be reduced, leading to a significant shrinkage in ASIC orders.
3️⃣ Increased Competition:
ARM AI ASIC chips are also on the way, expected to launch in early 2027, and MediaTek is also working on dual-source supply. While the market pie is expanding, there are also more people sharing it. Whether AVGO can maintain its market share is one of the core uncertainties.
📝 My personal judgment is that UBS’s report, while looking beautiful in data, has too much uncertainty in actual implementation. A shipment of 35 million TPUs feels somewhat unrealistic to me; a more realistic expectation might be around 15 to 20 million.
If we consider 15 million units with a 60% market share, AVGO could capture 9 million units of market share, doubling its market value by 2028. This should be relatively easy.
Key timelines to pay attention to:
• End of 2026 to early 2027: Launch of ARM AI ASICs, observe the competitive landscape.
• Mid-2027: Monitor the progress of OpenAI's ASIC project, which will have a huge impact on AVGO.
• 2028: Check if the actual shipment can meet UBS's expectations.
DYOR 🙏, investing has risks, just take a look at this UBS report and don’t take it too seriously!
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
