Broadcom Stock Is Back Where It Ended 2025, but Its Earnings Are About 43% Higher. Is It a Buy?

Broadcom raised its AI-chip revenue outlook to $58 billion for fiscal 2026, approximately $115 billion for fiscal 2027, and $230 billion for fiscal 2028, supported by secured supply commitments. Fiscal Q3 AI-chip revenue rose 221% year over year to $16.7 billion, while adjusted EPS increased 96%; management expects more than $30 of EPS in fiscal 2028. Shares near $350 remain 29% below their 52-week high and trade at roughly 36x trailing adjusted EPS, but customer concentration is increasing, with the top five end customers representing about 55% of Q3 revenue and financing exposure tied to certain AI labs.

Analysis

AVGO's valuation discount is better understood as a financing-and-counterparty discount than a simple customer-concentration discount. Revenue tied to customers requiring lease backstops or residual-value guarantees carries materially different economics: reported semiconductor revenue can grow before cash conversion is proven, while a failed deployment could create both a demand shortfall and credit-like losses. The key diligence item is not aggregate AI backlog, but the share supported by customer self-funding versus Broadcom/Apollo/Blackstone-enabled structures.

Near-term, the raised outlook should stabilize the earnings multiple if quarterly AI revenue conversion remains on plan, but upside is unlikely to be linear. The market will demand evidence over the next 1-3 quarters that customer capex commitments translate into installed clusters and recurring order releases; any deferral by a single major lab could reset FY27 estimates sharply because concentration amplifies operating leverage. NVDA is a relative beneficiary if buyers favor merchant GPUs to avoid custom-ASIC project and financing complexity, although AVGO's custom-silicon economics remain superior once deployments reach volume.

The non-obvious spillover is to alternative asset managers: APO and BX gain fee-bearing deployment exposure, but the residual-risk structure can turn a headline financing win into marked-to-market and realized-loss risk if AI hardware obsolescence outpaces lease assumptions. This is a 6-18 month issue, likely invisible in initially reported management-fee growth. Consensus appears too focused on the headline forward P/E and too little on whether that earnings denominator embeds risk transfer back to AVGO and its financing partners.

Contrarian view: AVGO may be inexpensive only if management's long-dated revenue targets are treated as firm demand rather than capacity reservations. The appropriate catalyst is disclosure of contractual protections, cash deposits, cancellation terms, and guarantee exposure—not another upward revenue target.

Key Decisions for Investors

• Initiate a scaled 3-6 month long AVGO position only after confirming post-earnings order conversion and disclosure that guarantee/backstop exposure is capped; target a return to its recent 40x adjusted-EPS valuation range, with downside protected by a stop/review if FY27 AI guidance is cut or a major customer defers capacity.
• Express the financing-risk thesis via a modest AVGO/NVDA relative-value trade: long NVDA, short AVGO in equal beta-adjusted dollars for 1-3 months if AVGO's next filing shows expanding guarantees or receivables. Cover if AVGO demonstrates predominantly self-funded customer deployments or NVDA's data-center growth decelerates materially.
• Avoid adding APO or BX solely on AI infrastructure financing headlines. Monitor quarterly disclosures for principal capital at risk, residual-value assumptions, and realized-credit marks; a widening in any disclosed loss reserves would be a catalyst to underweight the more exposed manager.
• For existing AVGO longs, use 6-9 month downside puts or put spreads around the next earnings cycle rather than selling outright; the primary left-tail event is customer funding slippage, which can produce a discrete multiple reset despite continued reported AI revenue growth.