Broadcom's AI Revenue Is Exploding. Here's What Investors Should Watch Next.

Broadcom's AI chip sales rose 65% to $20 billion in fiscal 2025, and the company projects $58 billion in fiscal 2026, $115 billion in 2027 and $230 billion in 2028. Those targets imply rapid growth, but high interest rates and potential restrictions on data-center construction could curb AI spending and force estimate cuts; VMware-based infrastructure software, 30% of latest-quarter revenue, may not offset a sharp chip slowdown. Shares have fallen more than 30% in six months, and the article argues their value depends on continued AI-market expansion.

Analysis

The key risk is not simply weaker AI spending; it is the sensitivity of Broadcom’s growth case to a small set of hyperscaler programs. Custom accelerators can displace GPUs on defined workloads, but customer-specific designs create program-renewal and bargaining risk: a delayed deployment or customer shift toward internal silicon can impair several years of expected growth before the broader AI market visibly contracts. Bundled networking and interconnect may cushion individual chip losses, but remain exposed to the same data-center build cycle.

The cited earnings multiple is not a standalone margin of safety: verify the forward-EPS denominator and how much of it assumes the company’s multi-year AI targets. Near term (days to weeks), rate moves and AI-capex commentary can dominate the stock. Over 1–3 months, hyperscaler spending plans, Broadcom guidance/order conversion, and VMware subscription renewals are the useful tests. Over 6–18 months, permitting, power availability, and customer silicon choices matter more than political rhetoric alone. State-level restrictions could delay projects and concentrate builds in less constrained regions, potentially supporting utilization and pricing for established infrastructure suppliers rather than stopping national demand outright.

Contrarian angle: the market may overread localized permitting proposals as a broad halt, but may underprice customer concentration and execution risk embedded in the out-year forecasts. No valuation conclusion follows from the reported multiple without checking estimates and cash-flow conversion.

Key Decisions for Investors

• Avoid chasing AVGO solely on the stated forward multiple. Before adding, verify the EPS estimate basis, customer/program concentration, order conversion, and whether AI guidance is supported by committed deployments rather than long-dated expectations.
• Use the next 1–3 months as a confirmation window: track hyperscaler capex plans, AVGO AI revenue guidance/backlog conversion, and VMware subscription renewal indicators. A guidance reduction or weakening order conversion falsifies the near-term growth thesis; sustained conversion supports reassessment.
• If AI-capex exposure is being reduced, consider a relative-value hedge—short AVGO against long NVDA—only if the intended thesis is AVGO’s greater custom-program concentration, not a broad AI downturn. Size conservatively and reassess on evidence that GPU demand is also weakening; do not assume the pair is valuation-neutral.
• Treat data-center policy headlines as alerts, not confirmed demand destruction. Monitor enacted permitting restrictions, power availability, and project delays across multiple states; a broad rise in cancellations would strengthen the downside case, while localized pauses alone would not.