$AVGO · Broadcom Inc.
Latest call · 2026-09-02
The call
Head-to-head · Claude vs the Street vs reality
| Claude | Street | Actual | |
|---|---|---|---|
| EPS | $3.28 | $3.22 | ⏳ pending |
| Revenue | $29.75B | $29.43B | — |
| Direction (1d) | 🟢 UP |
🟢 UP
34B · 8H · 2S
|
— |
| 1d move | +2.5% | — | — |
| 3d move | +4.0% | — | — |
Thesis
Consensus for a modest revenue beat + in-line-to-slight-beat Q4 guide, but the whole call is about AI custom-silicon narrative injection. AVGO has multiple fresh disclosure levers management can pull (OpenAI Jalapeño ramp date, 1.3GW-2027 FY27 color, any 7th hyperscaler customer). At 70% of average PT and fwd P/E 19.76, positioning is not stretched — cushions any GM headline noise.
What would flip it
Q4 GM guide steps below the reiterated 74% floor — that fires the CSCO GM-compression-amplifier template and flips the reaction to -5 to -8%.
The market's narrative
Custom-silicon story is the whole thesis. Post-OpenAI Jalapeño announcement and Google TPU $46B multi-year, AVGO has named 6 hyperscaler custom-silicon customers (Google, Meta, ByteDance, OpenAI, Anthropic + 1 unnamed) and the Street models AI semi rev $16B this quarter (+200% YoY). Trading at 70% of $526 PT — meaningfully above spot but below the extreme-above-PT threshold that fires the sell-the-news template.
Where the Street may be wrong
- Forward P/E 19.76 is REASONABLE for a mega-cap AI-infra name with +40% total rev growth trajectory — not stretched like PLTR (fwd P/E 240) or CRWD (fwd P/E 95). The multi-compression fear that hit CSCO/NVDA doesn't apply to AVGO which is already at a semi-consumer-staple multiple. Cushions any GM-compression selloff.
- OpenAI Jalapeño Q1'27 ramp + 1.3GW-contracted-2027 disclosure gives management the discrete narrative-injection lever to fire on the call — the CRM Agentforce template (fired +2.63% 1d / +5.13% 3d 8/26) applies directly. Any incremental customer named or GW-contracted number for FY27 is the narrative unlock.
- GM decel to ~74% is already GUIDED (not a surprise) — the CSCO GM-compression-amplifier that fired -8.83% on 8/12 was triggered by NEW GM commentary, not reiterated. AVGO's GM band has been telegraphed for 3 quarters — investors have re-based expectations.
- RBC 8/26 downgrade to Sector Perform PT $400 established a low-bar Street setup — the marginal analyst is already cautious. Fresh downgrade one day before print historically resolves in the direction the fundamentals surprise (positioning cleared).
Peer read: NVDA 8/26 muted print (-0.09% 1d, -4.18% 3d) is a bearish read-through on 'beat + in-line guide = FLAT' template for mega-cap AI-infra. Offsetting: MRVL 8/28 AMC (custom-silicon peer) direction is a more direct read that will inform the tape. If MRVL prints up on AI custom silicon momentum, AVGO gets a positive tailwind; if MRVL fades on GM concerns, AVGO tape softens.
Reasoning
- Fundamentals model: consensus rev $29.43B / EPS $3.22. My estimate $29.75B / $3.28 (rev +1.1% / EPS +1.9% vs Street). Beat magnitude is modest — AVGO typically beats revenue by $300-800M and this cycle's tailwind is Q3 being first full-quarter of OpenAI Jalapeño booking (small revenue contribution but strategic).
- Q4 guide is the swing variable. Street models Q4 rev ~$31.5B. AVGO's beat cadence would put base-case guide $32.0-32.5B (midpoint $32.25B = +2.4% above street). Under the new NVDA rule (mega-cap beat + guide clearing <3% = FLAT), +2.4% is the borderline case. What flips it to UP: any FY27 AI semi rev color (>$70B implied vs street $58B) or new-customer disclosure.
- AI-narrative-injection template (CRM 8/26 fired) applies: AVGO has multiple discrete AI product ARR milestones (Jalapeño Q1'27 ramp date, 1.3GW-2027 contract, any 7th hyperscaler unlock). Any one of these called out on the transcript = re-rating catalyst that overrides the GM-compression concern.
- Templates in play — (a) AI-INFRA-MEGA-CAP-GM-COMPRESSION-AMPLIFIER (CSCO 8/12 fired -8.83%) — CONDITIONAL: fires ONLY IF Q4 GM guide steps DOWN to <73% (below reiterated 74%); if reiterated at 74%, template does NOT fire; (b) AGENTFORCE-AI-NARRATIVE-INJECTION (CRM 8/26 fired +2.63%) — applies to AVGO on OpenAI/Google/incremental customer disclosure; (c) PLTR-trap sell-the-news — DOES NOT fire because AVGO at 70% of PT, well below the >100% PT trigger.
- Positioning read: RBC downgrade to $400 (below-Street PT) day before print = cleared positioning. IV implied ~7-8% move is roughly average (not underpriced like NVDA). Setup is neutral-tilted-up: not extreme, not stretched, with multiple named catalyst levers management can pull. Sizing MEDIUM.
Risks to the call
- Q4 GM guide steps DOWN to <73% (below reiterated 74%) — this would fire the CSCO GM-compression-amplifier template hard, taking stock to -5 to -8% regardless of the rev beat. Named risk: mgmt commentary on TPU-mix-shift or OpenAI-inference-margin cost profile in the transcript.
- Q4 rev guide MISSES street at $31B or below — the 'beat-and-modest-guide-is-FLAT' template (NVDA 8/26 lesson) then inverts to DOWN because base was already FLAT and any downside surprise fires -3 to -5%. Highest-probability trigger: any pull-forward of hyperscaler orders into Q3 leaving Q4 softer than modeled.
- Customer-concentration disclosure — if any hyperscaler named as 'pausing' or 'reevaluating' custom-silicon roadmap (Meta's MTIA program pivots have flagged this before), the entire custom-silicon narrative de-rates.
Prior calls
| Date | Earnings | Claude EPS | Street EPS | Actual | Claude DIR | Actual DIR | Verdict |
|---|---|---|---|---|---|---|---|
| 2026-06-02 | 2026-06-03 AMC | $2.45 | $2.40 | — | 🟢 UP | ⚪ — |