Broadcom: Market Is In Disbelief; Don't Be Mistaken

Broadcom is targeting more than $350 billion in AI semiconductor revenue over the next two years as it deploys multi-gigawatt AI chip capacity with Anthropic and OpenAI. AVGO revenue estimates continue to rise, while operating margins are projected to stay above 68% despite gross-margin pressure. The stock's forward valuation has compressed below 21x earnings, though buying support is cited near $360.

Analysis

The investable question is not whether custom AI silicon demand is large, but whether AVGO can convert announced capacity into durable, high-margin revenue before hyperscaler buyers diversify designs. At sub-21x forward EPS, the multiple now embeds a meaningful execution discount relative to AI infrastructure peers despite a business mix that should produce unusually high incremental operating leverage. If operating margin holds above 68%, each $10 billion of incremental semiconductor revenue could add roughly $4.50-$5.00 billion of operating income, making estimate revisions—not headline capacity—the primary 1-3 month catalyst.

The second-order beneficiary is TSMC (TSM), where advanced-node and advanced-packaging utilization should tighten if custom accelerators ramp on schedule; networking optics and interconnect suppliers such as ANET and MRVL may also gain from scale-out clusters. Conversely, AVGO's success raises the medium-term competitive risk for NVDA: custom ASICs do not need to displace CUDA broadly to pressure the premium assigned to NVIDIA's inference revenue. The likely outcome over 6-18 months is a more segmented AI silicon market, with NVDA retaining training leadership while AVGO captures a larger share of captive, predictable hyperscaler workloads.

Consensus may be overly focused on gross-margin dilution from semiconductor mix. That concern matters only if lower gross margin is accompanied by weaker operating leverage or customer concentration-driven pricing concessions; otherwise, higher-volume ASIC revenue can still be accretive to EPS and reduce the cyclicality historically attached to AVGO's non-AI semiconductor franchise. The thesis is falsified by a material delay in customer deployment schedules, operating-margin guidance below 66%, or evidence that incremental AI revenue requires materially higher working capital and capex than currently implied.

Near term, $360 is a technical risk marker rather than a fundamental valuation floor. A break below it without estimate cuts would improve entry economics; a break accompanied by lower AI revenue guidance would signal that capacity announcements are ahead of end-demand and should not be bought mechanically.

Key Decisions for Investors

• Accumulate AVGO on weakness toward $360 only if next-quarter AI semiconductor revenue guidance and consolidated operating-margin outlook remain intact; target a rerating toward 24-25x forward EPS over 6-12 months, with risk defined by a sustained margin guide below 66% or a customer-ramp delay.
• Express the custom-silicon share-gain thesis as long AVGO / short a small notional NVDA overlay over 6-18 months, rather than an outright NVDA short. The pair benefits if hyperscaler ASIC adoption narrows NVDA's inference multiple; cover if NVDA's data-center guidance accelerates while AVGO's AI estimates fail to rise.
• Monitor TSM quarterly advanced-packaging commentary and utilization as independent confirmation. A tightening packaging backdrop supports adding AVGO exposure; evidence of excess CoWoS/advanced-packaging capacity would challenge the assumed deployment cadence.
• Do not chase a post-earnings gap higher solely on capacity rhetoric. Add only after verifying backlog conversion through revenue guidance, customer concentration disclosure, and working-capital trends; these are the missing data points separating a high-quality ramp from a low-visibility design-win narrative.