Prediction: This Is What a $5,000 Investment in Broadcom Will Be Worth by 2028

Broadcom forecasts AI semiconductor revenue of $230 billion by fiscal 2028, up from $16.7 billion currently and $65 billion projected for FY2026, driven by custom-chip demand from hyperscalers including Alphabet, Meta, OpenAI, and Anthropic. The article estimates total FY2028 revenue could reach $280 billion, combining the AI target with roughly $50 billion of non-AI revenue. Assuming a 45% profit margin and 30x earnings multiple, it projects a $3.78 trillion valuation versus Broadcom's current $2.7 trillion market capitalization, implying 118% upside.

Analysis

The key valuation issue is not whether custom silicon grows, but whether AVGO can convert design wins into durable economics rather than low-margin, customer-concentrated pass-through revenue. The published upside math is internally inconsistent: a move from roughly $2.7T to $3.78T market value is about 40%, not 118%. Moreover, assigning a premium software/platform multiple to a hardware mix that is increasingly dependent on a handful of buyers leaves little room for delays in deployment, yield losses, or pricing concessions.

Over the next 1-3 months, the relevant confirmation is hyperscaler capex guidance, disclosed production ramps, and AVGO AI operating-margin progression—not headline revenue targets. Custom ASIC adoption should improve GOOG and META's inference cost per token and may extend their capex cycles by improving ROI; it is therefore not necessarily a net reduction in their spend. Second-order beneficiaries include TSMC (TSM), HBM suppliers and advanced packaging, while the competitive pressure on NVDA is more acute in mature, predictable inference workloads than in training or rapidly changing model architectures.

Consensus appears to treat ASIC share gains as a zero-sum displacement of GPUs. The counterpoint is that customers retain GPUs as the flexible layer while ASICs absorb stable workloads; total accelerator demand can expand even as NVDA's mix shifts. AVGO's risk is that customers internalize more design capability or use multiple ASIC vendors once architectures standardize, compressing returns on its current scarcity value. A material cut to META/GOOG capex, a sequential decline in AI semiconductor gross margin, or evidence that deployment volumes lag customer commitments would falsify the bullish thesis quickly.

Key Decisions for Investors

• Do not chase AVGO on revenue-target headlines; build only after the next earnings release confirms AI segment margin and backlog/production conversion. Use a 8-10% stop from entry or exit on a sequential AI-margin decline, as the multiple is vulnerable to any evidence of lower-quality growth.
• Express the share-shift thesis as a 3-6 month long AVGO / short NVDA pair in equal dollar amounts only if AVGO demonstrates accelerating custom-chip shipments while NVDA's data-center guide remains dependent on pricing rather than unit growth. Target 10-15% relative outperformance; cover if NVDA raises forward demand materially or AVGO fails to validate volume ramps.
• Monitor GOOG and META earnings for capex guidance and inference-cost commentary. A maintained or raised capex outlook alongside comments on lower cost per token is a read-through positive for AVGO, TSM and packaging/HBM suppliers; a synchronized capex reset is the highest-probability catalyst for multiple compression across the AI hardware complex.
• Treat the implied 2028 valuation case as requiring both extraordinary revenue execution and sustained premium margins. Require an updated sensitivity table using 30-40% operating/profit margins and a 20-25x earnings range before underwriting a core long; absent that, AVGO is a tactical momentum exposure rather than a fresh strategic overweight.