AVGO · Broadcom Inc.

Current Thesis

The Q2 FY26 binary that gated the name has cleared, and the tape rejected the print. The 2026-06-03 (after-close) report was a fundamental monster AI-semiconductor bookings over $30B against just $10.8B shipped (~2.8x book-to-bill), Semiconductor Solutions revenue $15.009B vs $8.408B YoY (+78.5%), and a fresh $100B+ FY2027 AI-revenue forecast (2026-06-04). The stock fell anyway, tumbling over 12% on 2026-06-04 because the Street wanted a beat-and-RAISE and got a beat-and-reiterate, then extending lower on 2026-06-05 inside a 3%+ Nasdaq-100 down day on rate-hike fears. Fundamentals are still ACCELERATING; the tape is now CORRECTING into a macro risk-off. A monster quarter the market sells, then continues to de-rate alongside the broad index, is a digestion phase the buyable structure is the post-earnings higher-low that holds, not the day-one knife in a falling complex.

Bull Case

• 2026-06-03 AI bookings >$30B vs $10.8B shipped (~2.8x book-to-bill): the single most bullish datapoint in the report; FY27 backlog is now contractual visibility, not a hope.
• 2026-06-03 Semiconductor Solutions $15.009B vs $8.408B YoY (+78.5%): AI mix is dragging the whole semi segment vertical, contracted growth vs merchant-GPU volatility.
• 2026-06-04 $100B+ FY2027 AI forecast; Deutsche Bank maintained Buy, raised PT to $515: a hard number on the ramp, and a sell-side bull lifting targets into the selloff.
• 2026-06-03 capacity roadmap: CEO Hock Tan plans to ship 10 GW in 2027 with "a lot more" in 2028, says supply secured for 2026/2027 and is working 2028/2029; Anthropic added as a next-gen TPU customer alongside Google, Meta, OpenAI.
• 2026-06-04 Hock Tan: AI is boosting, not hurting, VMware; Q3 software guide ~$8.9B with $6B of purchase orders already booked (2026-06-03) a >70%-margin annuity funding the buyback.
• 2026-06-04 TSMC CEO: capacity constrained "very long time," hints price hikes: scarcity is a moat for whoever already holds allocation, and AVGO says it has 2026/2027 secured.
• 2026-06-05 Dan Niles: the selloff "didn't break the AI trade," semiconductor bulls "should be thrilled"; 2026-06-04 reporting that Broadcom, Apollo and Blackstone are quietly building an AI-infrastructure financing vehicle.

Bear Case

• 2026-06-04 sell-the-news, stock tumbled >12% on a monster print: when a name books $30B and drops double digits, the narrative is priced; a MATURING tape signal, not a dip to buy blindly.
• 2026-06-05 a Broadcom analyst turned bearish (downgrade): sentiment is cracking at the margin even as the fundamentals print clean.
• 2026-06-04 "$100B forecast, Wall Street doesn't believe it": the bar is now beat-and-raise every quarter; a reiterate reads as a disappointment, and the easy multiple-expansion leg is over.
• 2026-06-04 Broadcom is FINANCING Anthropic/OpenAI chip purchases: vendor financing of cash-burning customers is a late-cycle circularity flag and a revenue-quality question if those buyers can't self-fund.
• 2026-06-05 Nasdaq-100 -3%+ on rate-hike jitters, Fear & Greed back to Neutral from Greed (2026-06-04): the whole AI-semi complex took profit on the same headlines; correlated drawdown is live and rates dominate the tape.
• Customer concentration / MRVL-Google TPU threat (carryover): Google + Meta remain the bulk of AI-silicon revenue; any next-gen socket loss to Marvell cracks the FY27 model.
• Valuation ~35x forward already embeds the AI ramp the print just demonstrated the market will not pay up even for an in-line monster.

Setup & Price Structure

No live feed in context; structure read from the tape. AVGO ran from the mid-May ~$430–440 shelf into the 2026-06-03 print, gapped down over 12% on 2026-06-04, then extended lower on 2026-06-05 as the Nasdaq-100 fell 3%+. That sequence is an active correction layered on top of an accelerating fundamental story exactly the regime where chasing the first red candle is the error. The line in the sand is the 2026-06-04/05 reaction low; structural support sits at the May breakout shelf and the rising 20-EMA below. The 2026-06-04 framing ("Tumbles Over 12%: the 3 times it fell harder were all screaming buys") is context, not a signal. A buyable base requires a higher-low that holds above the reaction low and reclaims the 20-EMA on a weekly close; until then the structure is broken short-term and price is hostage to rates. Deutsche Bank's $515 PT (2026-06-04) sits well above the post-drop print, so the reward exists once a base confirms the issue is timing, not direction.

Catalyst Calendar (next 30 days)

• No AVGO-specific binary inside 30 days. Next print is Q3 FY26, est. ~2026-09-04 trading digestion, not an event.
• ~2026-06-05 onward Qualcomm Investor Day data-center/AI push: a custom-silicon demand read-through and peer-sentiment input.
• Mid-June (est.) next FOMC decision / CPI print: with 2026-06-05 selling driven by rate-hike fears, the macro rates path is the dominant near-term driver of AVGO beta, not a company catalyst.
• Ongoing hyperscaler capex commentary (Meta, Google, Anthropic, OpenAI): 2026-06-06 reporting of a Meta investor dumping shares over AI-spend concerns flags that any capex-digestion comment is the live cluster risk.

What Would Change Our Mind

• Bullish reclaim: a weekly close back above the 20-EMA that recovers the pre-earnings ~$430–440 shelf on a higher-low holding the 2026-06-04/05 reaction low confirms the sell-the-news was a shakeout and the digestion is over.
• Thesis-break / invalidation: a weekly close below the 2026-06-04/05 reaction low AND under the 20-EMA turns sell-the-news into a structural de-rate; OR the FY27 AI guide later cut below $100B; OR a confirmed Google/Meta next-gen TPU socket loss to Marvell; OR a hyperscaler capex-digestion print that cracks the whole ai-chip cluster.

Correlation Notes

• ai-chip cluster (NVDA, TSM, AMD, MRVL, MU) moves as one. AVGO's 2026-06-04 print dragged AMD -3.82% and TSM ~-2% the same session.
• On 2026-06-05 Broadcom, Micron, AMD, TSM, Marvell, Intel and Arm all fell together a single macro factor (rates) is subordinating every idiosyncratic story right now, so cluster names gap in unison.
• MRVL is simultaneously a peer and the specific share-loss threat (Google next-gen TPU); a long-AVGO / short-MRVL pair isolates the moat from customer-shift noise.
• Stacking a third correlated name into the same beta compounds drawdown risk in a risk-off tape; size the shared factor, not three independent bets.