Broadcom's AI custom-silicon and networking franchise is executing exceptionally, but a 12% rally since the prior report restores a premium that already prices in the ~50% forward EPS growth consensus expects.
Researched 44 days ago
Broadcom's AI custom-silicon and networking franchise is executing exceptionally, but a 12% rally since the prior report restores a premium that already prices in the ~50% forward EPS growth consensus expects.
Quality is only Watch — hold, do not add until the thesis firms up.
Recommendation
Conviction
67/100
watch
Upside
33/100
bull 80% · ~30% odds · +20% expected
Risk-adjusted upside
10/100
+10% after downside pressure
Thesis quality
8.2/10
Opportunity
5.4/10
Risk pressure
5.3/10
Valuation
FairAI fair value
$420.00
Fundamentals check
$309.86
12-24mo fair-value range
$340.00 / $420.00 / $500.00
width 38%
Estimates disagree — look closer.
Buy below
$304.06
Trim above
$600.00
Implied expectations
implausible53.1% implied revenue CAGR
Today's P/S implies ~53.1% revenue CAGR for five years versus 22.0% realized growth.
Broadcom sits directly in the custom-ASIC and AI-networking bottleneck: hyperscalers seeking alternatives to merchant GPUs route through Broadcom's XPU co-design and Ethernet switching franchises. The 46% YoY revenue growth with expanding 69.5% gross margin in the latest quarter is direct evidence the bottleneck position is converting to pricing power. Defensibility is high on 18-36 month horizon given design-win lead times, though hyperscaler in-housing and MRVL competition cap the score below maximum.
Moat
Wide moat from hyperscaler custom-ASIC co-design lock-in, Ethernet switching IP, and VMware switching costs.
Bottleneck fit
Direct fit in the custom-ASIC/AI-networking bottleneck, evidenced by 46% YoY growth with expanding margins.
Valuation
~71x trailing P/E and ~60x EV/EBITDA already price in delivered consensus growth; no margin of safety at $427.76.
Catalyst
Q3 FY2026 print (early Sep) with a specific readout: AI revenue trajectory and initial FY2027 framing.
Why not higher
Valuation fails: without a variant view against the ~50% forward EPS growth the price embeds, buying here echoes consensus rather than exploiting an edge — the honest call is WATCH, not CONSIDER.
Description
Broadcom designs custom AI accelerators (XPUs) and merchant networking/switching silicon for hyperscalers, alongside a large infrastructure-software business (VMware) that adds recurring high-margin cash flow.
Value Chain
Component and infrastructure layer: custom ASICs, Ethernet switching, and interconnect silicon feeding hyperscaler AI datacenter buildouts, plus platform-level infrastructure software.
Moat
wideMulti-year custom-ASIC co-design relationships with hyperscalers, dominant Ethernet switching IP, and VMware switching costs create high barriers; 68.4% TTM gross margin evidences durable pricing power.
Pricing Power
highQuarterly gross margin expanded to 69.5% (2026-05-03) from 64.4% a year prior while revenue grew ~46% YoY — simultaneous growth and margin expansion indicates strong pricing power.
Customer Concentration
Hyperscaler-concentrated by design: custom accelerator revenue depends on a small number of large cloud customers; specific percentages not disclosed in packet.
| Metric | Value |
|---|---|
| Revenue growth YoY | 46.40% |
| Gross margin | 69.52% ↑ |
| Operating margin | 48.76% |
| FCF margin | 46.25% |
| Cash position | 19720000000 |
| Net debt / EBITDA | 1.30 |
| Share count change YoY | — |
| ROIC | 20.13% |
Forward P/E
22.42
Trailing P/E
71.21
PEG
0.43
EV/EBITDA
59.77
P/S
26.13
| Peer | Metric | Value |
|---|---|---|
| NVDA | Peer set | Packet identifies NVDA as a semiconductor peer but provides no comparable valuation metrics. |
| AMD | Peer set | Packet identifies AMD as a semiconductor peer but provides no comparable valuation metrics. |
| MRVL | Peer set | Packet identifies MRVL (closest custom-ASIC comparable) but provides no valuation metrics. |
| TXN | Peer set | Packet identifies TXN as a semiconductor peer but provides no comparable valuation metrics. |
Bull fair value
$650.00
Probability
30%
Horizon
Long term
YTD
—
1Y
40.82%
Vs sector
—
From 52w high
-13.58%
| Risk | Severity | Explanation |
|---|---|---|
| Valuation compression on AI capex digestion | high | At ~60x TTM EV/EBITDA and ~71x trailing P/E, any hyperscaler capex pause or slower XPU ramp would compress the multiple sharply; the stock fell 23% from its June high on precisely this sensitivity before rebounding. |
| Customer concentration in a few hyperscalers | high | Custom-ASIC revenue depends on a small number of cloud customers who can in-source, dual-source (e.g. MRVL), or reschedule programs; a single program cancellation would be material. |
| Geopolitical / export-control exposure | medium | Semiconductor supply chains and China end-demand carry export-control and tariff tail risk, applied here symmetrically as for other AI-silicon names. |
| Leverage from VMware acquisition | low | Net debt/EBITDA has deleveraged to roughly 1.3x on surging EBITDA (packet-derived), so balance-sheet risk is receding but rate-sensitive debt remains. |
0%–0%
WATCH: elite fundamentals but the current price sits at or above a conservative fair value with no variant view; monitor for a pullback toward the mid-$300s or evidence at the Q3 print that growth duration exceeds what the multiple embeds.
AVGO trades at ~71x trailing P/E, ~60x TTM EV/EBITDA, and 26x TTM sales — multiples that require the AI capex boom to persist for years without a digestion phase. The packet itself shows how fragile that is: the stock fell from $495 (2026-06-03) to well off that high and has bounced +9.1% in five days on no new fundamental data. Revenue is hyperscaler-concentrated custom-ASIC demand from a handful of customers who can in-source or dual-source to MRVL; one program reschedule flows straight through a ~$2.04T market cap priced for perfection. Meanwhile insiders (Brazeal -50k sh, Samueli, Delly, Page) sold into June-July with zero buys, and receivables turnover has deteriorated from 10.87 (2024-11) to 6.91 TTM — receivables growing much faster than sales, consistent with lumpy concentrated billings a bear would probe.
| Load-bearing assumption | Why it might be wrong | Severity |
|---|---|---|
| Consensus forward EPS of ~$19.1 (backed out from forwardPE 22.42 at $427.76) is a reliable fair-value anchor. | $19.1 is ~3.2x TTM EPS of $6.01. Even with epsGrowthTTMYoy of 125.8% (decelerating from a low base that included a -$0.40 quarter in Aug-2024), tripling EPS in one forward year is implausible; the forwardPE datum is likely stale or mismatched, making the entire $340-$500 fair-value band unreliable. | high |
| 46.4% YoY quarterly revenue growth, derived from salesPerShare (4.5502 vs 3.109). | Per-share series conflate revenue with share-count changes; share_count_change_yoy_pct is 'unavailable' in the report and FMP statements 403'd, so true revenue growth is unverified from the packet. | medium |
| Gross margin expansion to 69.5% evidences durable pricing power. | Margin expansion coincides with VMware software mix shift post-acquisition, not necessarily semiconductor pricing power; the packet cannot separate segment mix from pricing, so the 'high pricing power' score is inferred, not evidenced. | medium |
| Net debt/EBITDA ~1.3x means balance-sheet risk is receding. | The figure is doubly derived (netDebtToTotalEquity x book value / annualized EBITDA) with no statement-level confirmation (FMP 403). LT debt is still 35% of assets; at 4.69% 10Y yields, refinancing cost on a large fixed-debt stack is a live sensitivity the report downgrades to 'low'. | low |
| Q3 FY2026 print is a catalyst that can confirm growth duration. | A single print cannot validate the multi-year growth duration a 60x EV/EBITDA multiple embeds; the catalyst is symmetric — a modest guide-down likely produces the 23%-type drawdown already observed from the June high. | medium |
The $420 fair value applies 22x to a consensus forward EPS (~$19.1) that is itself backed out of the packet's forwardPE — a circular anchor: it reprices consensus at consensus, so 'fair value ≈ price' is nearly tautological. Worse, $19.1 forward vs $6.01 TTM EPS implies >200% one-year EPS growth, which contradicts the packet's own decelerating growth series and suggests the forwardPE datum is unreliable. The $340 low case (18x) and $500 high (26x) inherit the same suspect denominator, so the entire range — including the mid-$300s 'pullback buy' guidance — rests on an unverified earnings number.
Independent red-team pass · claude-fable-5 · 2026-08-10
| Claim | Source | URL | Retrieved |
|---|---|---|---|
| Current price $427.76, market cap ~$2.035T | finnhub:profile / packet currentPrice | — | retrieved 2026-08-09T23:56:09.345Z |
| Forward P/E 22.42, forward PEG 0.427, EV/EBITDA TTM 59.77, EPS TTM 6.0071, gross margin TTM 68.35% | finnhub:basic-financials | — | retrieved 2026-08-09T23:56:09.357Z |
| Quarterly (2026-05-03) gross margin 69.52%, operating margin 48.76%, FCF margin 46.25%, ROIC TTM 20.13%, P/S TTM 26.13 | finnhub:basic-financials (quarterly series) | — | as of 2026-05-03 · retrieved 2026-08-09T23:56:09.357Z |
| Revenue growth ~46.4% YoY latest quarter and ~22.0% FY2025, derived from salesPerShare series (4.5502 vs 3.109 quarterly; 13.1644 vs 10.7941 annual) | finnhub:basic-financials (salesPerShare series) | — | as of 2026-05-03 · retrieved 2026-08-09T23:56:09.357Z |
| Net debt/EBITDA ~1.3x derived from netDebtToTotalEquity 0.5163 x book value $87,691M over TTM EBITDA $34,804M | finnhub:basic-financials (derived) | — | as of 2026-05-03 · retrieved 2026-08-09T23:56:09.357Z |
| Cash position ~$19.7B derived from cashPerShare 4.1456 x 4,757.58M shares | finnhub:basic-financials (derived) | — | retrieved 2026-08-09T23:56:09.357Z |
| 52-week high $495 (2026-06-03); 1-year return 40.82%; 5-day return 9.06% | finnhub:basic-financials | — | retrieved 2026-08-09T23:56:09.357Z |
| Analyst recommendations Aug 2026: 17 strong buy / 34 buy / 5 hold / 0 sell — stable vs May, read as neutral momentum | finnhub:recommendations | — | as of 2026-08-01 · retrieved 2026-08-09T23:56:09.362Z |
| Insider selling June-July 2026: Brazeal -50,000 sh across two sales, Delly -1,890 sh, Page -1,602 sh, Samueli -6,695 sh open-market (plus zero-price gifts) | finnhub:insider-transactions | — | as of 2026-07-14 · retrieved 2026-08-09T23:56:09.368Z |
| Peer set: NVDA, MU, AMD, INTC, TXN, MRVL, ADI | finnhub:peers | — | retrieved 2026-08-09T23:56:09.367Z |
| Most recent 10-Q filed 2026-06-09; three 8-Ks filed June-July 2026 | sec:recent-filings | Link | as of 2026-07-06 · retrieved 2026-08-09T23:56:09.459Z |
| FMP income-statement, balance-sheet, cash-flow, and ratios endpoints returned 403 | fmp:income-statement | — | retrieved 2026-08-09T23:56:09.563Z |
| Macro: 10Y Treasury 4.69%, Fed funds 3.63%, HY OAS 2.71 (2026-08-06) | fred:macro-snapshot | — | as of 2026-08-06 · retrieved 2026-08-09T23:56:09.478Z |