⚠ Forward P/E source dispersion: GuruFocus 34.96× and StockAnalysis 36.10× (both 8 Jun) appear to lag post-Q2 EPS revisions; recomputed NTM at $396.60 is closer to ~30×. Page uses ~32× as the working figure — verify against broker feed before execution.
AVGO
Broadcom Inc.
AI Custom Silicon & Networking
📌 Current holding · ~1.7% NLV
$396.60
Mkt Cap ~$1.93T | ATH $437.68 (6 May 2026)
52W: $206.52 – $437.68
Next earnings: Sep 2026 (Q2 FY26 reported 3 Jun)
▲ Strong Add
Entry zone: $370–400 | Add more below $370
Q2 FY26 was a clean beat-and-raise, yet AVGO sold off ~15% because Hock Tan reiterated — rather than raised — the "in excess of $100B" FY27 AI-chip target.4 The market is misreading three things. First, the $100B is silicon-only: AI networking just hit the guided 40% of AI revenue in Q2,2 so on a networking-inclusive basis the FY27 path is closer to ~$140B. Second, Q2 AI bookings exceeded $30B against $10.8B shipped — backlog is now growing >3× faster than revenue.3 Third, the Q3 AI-semi guide of $16.0B (+200% YoY) annualises to ~$64B,2 making the FY27 $100B look conservative, not stretched. The selloff handed a 15% discount on the only company with both the custom-silicon expertise and hyperscaler trust to run locked multi-year XPU programs across six of the world's largest AI spenders at once — anchored in contract, not narrative.
Fwd P/E (NTM)
~32×5
PEG 0.53 (g≈60% NTM) · vs 3Y avg ~33× · regime Low–Median
Rev YoY (latest Q)
+48%1
QoQ +14.9% ↑ · $22.19B
Gross Margin
77.1%1
adj · 69.5% GAAP · stable QoQ → (−230 bps YoY)
FCF Margin (TTM)
~44%1
$32.8B TTM · 46% in Q2 ↑
EPS Revision (90d)
↑ up6
post-Q2 raise · Strong Buy, 48 analysts ⚠ ↑/↓ split tbc
Next Earnings
Sep 2026
Q3 FY26 · Est
AI Rev % of total
49%1
$10.8B Q2 · networking 40% of AI
AI Semi Rev YoY
+143%2
Q2 · Q3 guide $16.0B (+200%)
Thesis Durability
A
Variant perception: AVGO's custom-silicon moat is execution-based, not patent-based — hyperscalers can't replicate it without 5+ years of co-development and yield engineering Broadcom has already run. The AI backlog is largely locked: Google TPU contract to 20319, Anthropic 1GW 2026 → 3GW+ 20279, Meta MTIA >1GW confirmed Apr 20269, OpenAI 10GW 2026–20299 (Phase 1 financing pending; ~10% of 2027 TAM).
Q2 confirmed the missed leg. AI networking hit the guided 40% of AI revenue in Q2 (up from ~33% in Q1)2 — the networking-inclusive read is now a printed fact, not a projection. Tan's "in excess of $100B" FY27 target is silicon-only; on a sustained 35–40% networking mix the FY27 envelope is ~$140B. This is the line the sell-side still truncates.
Bookings are the real signal. Q2 AI semiconductor bookings exceeded $30B against $10.8B shipped — backlog now growing >3× quarterly revenue.3 Q3 AI-semi guide $16.0B (+200% YoY) annualises to ~$64B; FY26 AI revenue is guided to ~$56B (+~180%).2 Tomahawk 6 (100T) is shipping and the next-gen 200T switch taped out this quarter.3
TAM: Hyperscaler AI capex $700–725B in 20269, up 60–77% YoY; Broadcom holds 55–70% custom-ASIC share9.
Bear case steelman: NVDA's Spectrum-X Ethernet (~$10B run-rate9) and own-silicon ambitions could soften custom-XPU demand if NVDA dominates inference; Marvell is a credible #2 (Trainium, Maia); MediaTek is a Google inference-TPU second-source.9 Survivable: AVGO keeps the core TPU architecture and the 2031 training contract, and the networking leg benefits from Ethernet adoption regardless of GPU vendor.
Q2 confirmed the missed leg. AI networking hit the guided 40% of AI revenue in Q2 (up from ~33% in Q1)2 — the networking-inclusive read is now a printed fact, not a projection. Tan's "in excess of $100B" FY27 target is silicon-only; on a sustained 35–40% networking mix the FY27 envelope is ~$140B. This is the line the sell-side still truncates.
Bookings are the real signal. Q2 AI semiconductor bookings exceeded $30B against $10.8B shipped — backlog now growing >3× quarterly revenue.3 Q3 AI-semi guide $16.0B (+200% YoY) annualises to ~$64B; FY26 AI revenue is guided to ~$56B (+~180%).2 Tomahawk 6 (100T) is shipping and the next-gen 200T switch taped out this quarter.3
TAM: Hyperscaler AI capex $700–725B in 20269, up 60–77% YoY; Broadcom holds 55–70% custom-ASIC share9.
Bear case steelman: NVDA's Spectrum-X Ethernet (~$10B run-rate9) and own-silicon ambitions could soften custom-XPU demand if NVDA dominates inference; Marvell is a credible #2 (Trainium, Maia); MediaTek is a Google inference-TPU second-source.9 Survivable: AVGO keeps the core TPU architecture and the 2031 training contract, and the networking leg benefits from Ethernet adoption regardless of GPU vendor.
High-conviction structural thesis — strengthened by Q2. Cyclicality check: not triggered (demand-driven secular ramp, not commodity-price cycle). Six confirmed XPU customers, 2031 runway, >$30B single-quarter bookings, and the 40% networking print all harden the falsifiable, contractual anchors.
Business Quality
A
FCF inflection: Q2 FCF of $10.26B at a record 46.2% margin (up from 41.5% in Q1), on just $231M capex — an exceptionally asset-light cash engine. TTM FCF ~$32.8B; OCF $10.49B (+60% YoY).1
Software re-accelerated. Infrastructure software grew +9% YoY to $7.18B in Q2 — up from the +1% stall that had it pegged as a margin anchor. VMware VCF conversion is now translating into growth, not just margin.1
Margin stabilised: non-GAAP GM 77.1% (Q1 77.0% → Q2 77.1%, essentially flat QoQ; −230 bps YoY on AI-semi mix). The feared mix-driven compression has paused. Non-GAAP operating margin 67.3%; Adj EBITDA 69% of revenue.1
Deleveraging: net debt fell to ~$45.3B ($64.9B debt − $19.6B cash) from ~$51.9B — roughly 0.9× TTM EBITDA, covered by a single quarter of OCF. $0.65 quarterly dividend maintained.1 Solvency screen: not triggered (net debt <1× EBITDA, FCF runway not at risk).
Software re-accelerated. Infrastructure software grew +9% YoY to $7.18B in Q2 — up from the +1% stall that had it pegged as a margin anchor. VMware VCF conversion is now translating into growth, not just margin.1
Margin stabilised: non-GAAP GM 77.1% (Q1 77.0% → Q2 77.1%, essentially flat QoQ; −230 bps YoY on AI-semi mix). The feared mix-driven compression has paused. Non-GAAP operating margin 67.3%; Adj EBITDA 69% of revenue.1
Deleveraging: net debt fell to ~$45.3B ($64.9B debt − $19.6B cash) from ~$51.9B — roughly 0.9× TTM EBITDA, covered by a single quarter of OCF. $0.65 quarterly dividend maintained.1 Solvency screen: not triggered (net debt <1× EBITDA, FCF runway not at risk).
Insider Form 4 (trailing 90 days): no material open-market sales. The 8 Apr gift (22,000 shares) was non-cash; the last open-market sale (6 Jan, ~$24.3M) has now rolled out of the 90-day window.8 The relentless-CEO-selling yellow flag from last quarter has paused — though the longer pattern remains zero insider buys. Not a cluster; no conditional metric card triggered. Short interest 1.12% of float and declining.7
Pristine, now improving on every axis. Record FCF margin, software back to growth, margin compression stalled, net leverage under 1×, and the insider-selling overhang paused. Upgraded A− → A.
Entry Price Discipline
A−
The pullback arrived. At $396.60 (8 Jun), AVGO is ~9% below the $437.68 ATH and sits inside the $370–400 "better entry" band the prior verdict was explicitly waiting for. The 15% post-print derate was driven by sentiment (Tan reiterated rather than raised FY27), not by any fundamental miss.4
Multiple regime: Fwd P/E ~32× (NTM) sits Low–Median within AVGO's own post-AI 17–57× range and roughly in line with its ~33× 3Y average; the derate already compressed the multiple, so median-reversion risk is materially lower than at $428. PEG 0.53 — growth more than covers the multiple.5 EV/EBITDA NTM ~22×.
Distance from entry zone: current price is within the $370–400 entry zone (≈7% above the $370 add level). This is the first time since the AI re-rate that fundamentals beat and the price is in-zone simultaneously.
Q2 settled the debate it was waiting on: revenue $22.19B and AI semis $10.8B both beat; Q3 guide $29.4B (+84%) cleared the ~$28.5B Street bar.1,4 The "needs to confirm the guide" caveat from last quarter is resolved to the upside.
Institutional flow: short interest 1.12% of float and declining7 — minimal squeeze fuel, but also no balance-sheet bears. Sell-side consensus Strong Buy, average PT ~$490–518 (48 analysts).6
Multiple regime: Fwd P/E ~32× (NTM) sits Low–Median within AVGO's own post-AI 17–57× range and roughly in line with its ~33× 3Y average; the derate already compressed the multiple, so median-reversion risk is materially lower than at $428. PEG 0.53 — growth more than covers the multiple.5 EV/EBITDA NTM ~22×.
Distance from entry zone: current price is within the $370–400 entry zone (≈7% above the $370 add level). This is the first time since the AI re-rate that fundamentals beat and the price is in-zone simultaneously.
Q2 settled the debate it was waiting on: revenue $22.19B and AI semis $10.8B both beat; Q3 guide $29.4B (+84%) cleared the ~$28.5B Street bar.1,4 The "needs to confirm the guide" caveat from last quarter is resolved to the upside.
Institutional flow: short interest 1.12% of float and declining7 — minimal squeeze fuel, but also no balance-sheet bears. Sell-side consensus Strong Buy, average PT ~$490–518 (48 analysts).6
In-zone for the first time in the cycle. A 15% sentiment-driven derate on a beat-and-raise put the price inside the entry band; risk/reward is favourable here, not stretched. Upgraded C+ → A−.
Verdict
The forcing function fired and it fired clean: Q2 beat on revenue, AI semis, EPS and FCF; Q3 guided $29.4B (+84%) above the Street; networking hit 40% of AI revenue; and Q2 AI bookings topped $30B against $10.8B shipped. The only "miss" was sentiment — Tan reiterated the FY27 $100B chips-only target instead of raising it, and the stock gave back ~15% to $396.60. That is exactly the margin of safety the prior verdict was waiting for. With the price now inside the $370–400 entry zone, Business Quality upgraded to A (record 46% FCF margin, software back to +9% growth, net leverage under 1×) and Entry Discipline to A−, this moves from Add (pullback) to Strong Add. Currently a ~1.7% starter holding — add into this dip toward a 6–8% target; reserve further adds for any move below $370.
Current holding ~1.7% NLV → add to 6–8% on this pullback (price in the $370–400 zone) → reserve incremental adds for <$370. Hard cap 20% single stock. Conviction: High (Dim 1 A, Dim 2 A).
Opportunity-cost check
▾
vs CSPX
CSPX delivers ~10–12% long-run CAGR. AVGO at ~32× forward P/E with +143% AI-semi growth and a PEG of 0.53 offers asymmetric upside — you are paying roughly half the growth rate, a quality you cannot buy inside an index. The custom-silicon + AI-networking TAM is not accessible via CSPX, and the 15% post-print derate widened the edge.
vs current holdings
AVGO sits in a different layer to MU (memory) and GOOGL/META (software/ad) — custom silicon + AI networking, a leg otherwise absent from the book. It does NOT stack with any existing factor. Funding the add to 6–8% comes from the undeployed active-picks dry powder, not from displacing a held name; it diversifies AI exposure rather than concentrating it.
Thesis-breakers
▾
- 01 Q3 FY26 AI semiconductor revenue below $14.5B (vs $16.0B guide) when Q3 reports in Sep 2026 — a guide miss this early in the ramp would break the +200% YoY acceleration narrative. Status: Q2 beat at $10.8B; on track.
- 02 By the Q4 FY26 report (~Dec 2026): management lowers the FY27 AI-chip target below $90B, OR AI bookings-to-billings falls back under 1.5× (from >3× in Q2) — either confirms the backlog acceleration was a one-quarter pull-forward, not structural. Status: reiterated >$100B; bookings >$30B; intact.
- 03 Non-GAAP gross margin falls below 75% in any quarter through FY27 as AI-semi mix climbs past 55% of revenue — would confirm the mix-driven margin compression resumed after stabilising at 77.1% in Q2. Status: flat QoQ at 77.1%; off track.
Key risks
▾
- 01 Hyperscaler concentration: Google alone is estimated at ~$25.8B / 51% of AI accelerator revenue in FY26 (Citi est. via Beth Kindig). If Google insources TPU design or shifts fab strategy — even partially — this is a multi-billion-dollar revenue cliff. Mitigation: contract runs to 2031.
- 02 Top-2 customer concentration: Google + Anthropic estimated ~92% of AI accelerator revenue in FY26 (Citi). Any disruption to either relationship — financing, geopolitics, technology pivot — creates severe downside that no other customer can absorb in the near term.
- 03 NVDA Spectrum-X competition: Nvidia's Ethernet AI networking business is now at a ~$10B annualised run-rate and growing. It directly competes with Broadcom's Tomahawk/Jericho AI fabric — representing ~33–40% of AVGO's AI revenue. If NVDA captures 20% of the Ethernet switching market by 2028, that is ~$3B headwind on AI networking revenue at current run-rates.
- 04 FY27 deceleration debate: the market's whole reason for the 15% selloff is that Tan didn't raise FY27 above $100B despite a Q3 AI guide annualising to ~$64B. If FY27 AI revenue lands near $100B (not the ~$140B networking-inclusive bull math), the stock is roughly fairly valued here — the re-rate would stall and the ~$555 bull DCF would not be reached. This is the single biggest valuation risk.
- 05 Gross margin compression: non-GAAP GM stabilised at 77.1% in Q2 but is still −230 bps YoY. If AI semis grow past 55–60% of revenue, blended GM could dip toward 73–74% — compressing EBITDA margin and, at the same forward multiple, shrinking the absolute earnings base. Thesis-breaker #3 tracks this directly.
- 06 VMware/software durability: infrastructure software re-accelerated to +9% YoY in Q2 (from +1%), easing the stall risk — but SMB/mid-market churn on the VCF subscription conversion is still real. If software growth fades back toward low-single-digits through FY27, the high-margin software contribution can't offset AI-semi mix dilution.
- 07 MediaTek inference second-sourcing: Google is diversifying TPU inference workloads to MediaTek, gaining pricing leverage and supply-chain optionality.9 AVGO retains the core TPU architecture and the training wallet through 2031, but a multi-billion-dollar inference stream is contestable over FY27–28 if MediaTek scales. Bounded threat — monitor at each print.
DCF scenarios
▾
Discount rate: 11.0% — high-growth profitable tech with customer concentration risk (top-2 hyperscalers a large majority of AI revenue) but improved balance sheet: net debt ~$45.3B at ~0.9× TTM EBITDA. FCF base year FY26: ~$44B (H1 actual $18.27B + a stronger AI-weighted H2). Terminal growth 4.5%. Shares ~4.90B; net debt ~$45.3B.
5Y FCF CAGR
27%
Terminal growth
4.5%
Fair value/share
~$555
vs current
+40%
Bull: FY26 FCF ~$44B grows at 27% CAGR to ~$145B by FY31 as AI-semi revenue clears $100B in FY27 (networking-inclusive ~$140B), networking holds ~40% of AI revenue, and FCF margin stays above 43%. Requires the >3× bookings cover to convert, Meta MTIA at multi-GW scale, and no non-GAAP GM break below 75%.
5Y FCF CAGR
13%
Terminal growth
4.5%
Fair value/share
~$370
vs current
−7%
Bear: AI capex digestion in 2027 as hyperscalers absorb the build-out; FY27 AI revenue lands near $100B with no networking upside; Google in-sources a slice of TPU work; FCF growth decelerates to 13% CAGR as AI-semi mix pushes non-GAAP GM toward 73%. FCF reaches ~$81B by FY31.
Position: at $396.60 the stock sits just above the bear fair value (~$370) and ~40% below the bull (~$555) — a markedly better skew than the $428 pre-print level, where bear downside was −23%. Multiple regime context: Fwd P/E ~32× is Low–Median within AVGO's post-AI 17–57× range; the 15% derate already removed most of the multiple-premium risk, so a mean-reversion toward the ~33× 3Y average is roughly neutral from here. The bull case still needs the networking-inclusive FY27 to deliver — the stock has to grow through, not re-rate through, the valuation.
Catalyst timeline
▾
-
3 Jun 2026 ✓Q2 FY26 — hit. Beat on revenue ($22.19B), AI semis ($10.8B, +143%), EPS ($2.44) and FCF (46% margin); Q3 guided $29.4B vs ~$28.5B Street; networking hit 40% of AI revenue; AI bookings >$30B. Stock fell ~15% on FY27 target being reiterated, not raised.
-
Sep 2026Q3 FY26 earnings. Confirm/deny the $16.0B AI-semi guide (+200% YoY) and watch the bookings-to-billings ratio (was >3× in Q2) — the key tell on whether backlog acceleration is structural. Maps directly to thesis-breaker #1.
-
2026 Q3–Q4OpenAI Project Nexus Phase 1 financing / offtake resolution. If confirmed: positive on FY27 SAM. If it slips to 2027: contained (the bulk of FY27 demand is non-OpenAI), but headline risk.
-
~Dec 2026Q4 FY26 report + FY27 AI guide. The first binding test of "in excess of $100B" FY27 — a raise here is the catalyst the market wanted on 3 Jun and didn't get. Single biggest risk/reward event in the thesis; maps to thesis-breaker #2.
Named analyst commentary
▾
"Custom silicon's cost advantages and ability to drive lower inference serving costs at scale creates a strong value proposition for Big Tech. As more and larger clusters are deployed to serve exploding inference demand, there will be additional long-term tailwinds for the Ethernet networking giant."
— Beth Kindig, 11 Dec 2025
"Broadcom's $AVGO custom AI accelerator revenue is projected to be as high as $50.5 billion in fiscal 2026, up ~3.5X from 2025, with Google $GOOG accounting for roughly half of that total at $25.8 billion, followed by Anthropic at $20.9 billion, per Citi."
— Beth Kindig — X (@Beth_Kindig), 2026 · sourced via Citi estimate
"The easy-to-miss detail on the guide is that the $100 billion is only for silicon and does not include networking. … If we assume this mix continues on the low end for about 30% mix in AI networking of total AI revenue, then it's reasonable to assume Broadcom's AI revenue will be $143 billion with networking of $43 billion. This represents a QoQ growth of 30% for 7–8 quarters — which is an excellent baseline to set."
— Beth Kindig, IO Fund commentary, May 2026
⚠ Dylan Patel / SemiAnalysis: No direct quote obtained from open-access sources. SemiAnalysis free-tier and X (@SemiAnalysis_) did not surface an AVGO-specific piece during research (May 2026). Paid tier recommended for next refresh.
⚠ Serenity / @aleabitoreddit: No coverage found from open-access sources (Reddit, X, aibottlenecks.app) on AVGO specifically as of May 2026.
Sources
1 Q2 FY26 financials (revenue $22.19B, segments, non-GAAP GM, FCF $10.26B/46%, EPS $1.91 GAAP / $2.44 adj, balance sheet, $0.65 dividend) — SEC 8-K EX-99.1, filed 3 Jun 2026: https://www.sec.gov/Archives/edgar/data/0001730168/000173016826000051/avgo-05032026x8kxex99.htm
2 Networking 40% of AI revenue, AI semi +143%, Q3 AI guide $16.0B (+200% YoY), FY26 AI ~$56B, FY27 >$100B reiterated — Hock Tan, Q2 FY26 earnings call (Motley Fool transcript, 3 Jun 2026)
3 AI bookings >$30B vs $10.8B shipped, Tomahawk 6 shipping, 200T switch tapeout — SDxCentral, 3 Jun 2026
4 Stock −15% to ~$396.60 on FY27 target reiterated-not-raised; Q3 guide $29.4B vs ~$28.5B Street — CNBC "Broadcom earnings report Q2 2026" (3 Jun 2026); price as of 8 Jun 2026
5 Fwd P/E (GuruFocus 34.96×, StockAnalysis 36.10×, both 8 Jun) and PEG 0.53 — gurufocus.com/term/forward-pe-ratio/AVGO; stockanalysis.com/stocks/avgo/statistics/ ⚠ provider dispersion; appears to lag post-Q2 EPS revisions
6 Consensus Strong Buy, avg PT ~$490–518 (48 analysts) — S&P Global via public.com/stocks/avgo/forecast; marketbeat.com (8 Jun 2026)
7 Short interest 1.12% of float, declining — marketbeat.com/stocks/NASDAQ/AVGO/short-interest; stockanalysis.com (8 Jun 2026)
8 Insider Form 4 — no material open-market sale in trailing 90d; last sale 6 Jan 2026 (~$24.3M); 8 Apr gift non-cash — secform4.com/insider-trading/1211588.htm; benzinga.com insider trades (as of 3 Jun 2026)
9 Carryover thesis facts — Google TPU contract to 2031 (hashrateindex.com); Anthropic 1GW→3GW+ & OpenAI 10GW (Q1 FY26 call, TIKR); Meta MTIA >1GW (about.fb.com, Apr 2026); hyperscaler capex $700–725B (CNBC, Feb 2026); AVGO ASIC share 55–70% (tradingkey.com); NVDA Spectrum-X ~$10B run-rate & MediaTek inference second-source (Beth Kindig / IO Fund commentary, May 2026)
10 Beth Kindig / IO Fund — networking-inclusive FY27 framing (~$140B); now corroborated by the printed 40% Q2 networking mix — io-fund.com / beth-kindig.medium.com (Dec 2025 – May 2026)
All Q2 FY26 financials sourced from primary SEC 8-K (filed 2026-06-03). Customer-concentration figures cited in Key Risks are Citi analyst estimates re-circulated by Beth Kindig — treat as projection, not company guidance. Forward P/E varies materially across open-access providers; page uses ~32× (NTM) as the working figure until verified against a broker feed.
⚠ = figure not verifiable from primary SEC filing or has significant provider dispersion.
1 Q2 FY26 financials (revenue $22.19B, segments, non-GAAP GM, FCF $10.26B/46%, EPS $1.91 GAAP / $2.44 adj, balance sheet, $0.65 dividend) — SEC 8-K EX-99.1, filed 3 Jun 2026: https://www.sec.gov/Archives/edgar/data/0001730168/000173016826000051/avgo-05032026x8kxex99.htm
2 Networking 40% of AI revenue, AI semi +143%, Q3 AI guide $16.0B (+200% YoY), FY26 AI ~$56B, FY27 >$100B reiterated — Hock Tan, Q2 FY26 earnings call (Motley Fool transcript, 3 Jun 2026)
3 AI bookings >$30B vs $10.8B shipped, Tomahawk 6 shipping, 200T switch tapeout — SDxCentral, 3 Jun 2026
4 Stock −15% to ~$396.60 on FY27 target reiterated-not-raised; Q3 guide $29.4B vs ~$28.5B Street — CNBC "Broadcom earnings report Q2 2026" (3 Jun 2026); price as of 8 Jun 2026
5 Fwd P/E (GuruFocus 34.96×, StockAnalysis 36.10×, both 8 Jun) and PEG 0.53 — gurufocus.com/term/forward-pe-ratio/AVGO; stockanalysis.com/stocks/avgo/statistics/ ⚠ provider dispersion; appears to lag post-Q2 EPS revisions
6 Consensus Strong Buy, avg PT ~$490–518 (48 analysts) — S&P Global via public.com/stocks/avgo/forecast; marketbeat.com (8 Jun 2026)
7 Short interest 1.12% of float, declining — marketbeat.com/stocks/NASDAQ/AVGO/short-interest; stockanalysis.com (8 Jun 2026)
8 Insider Form 4 — no material open-market sale in trailing 90d; last sale 6 Jan 2026 (~$24.3M); 8 Apr gift non-cash — secform4.com/insider-trading/1211588.htm; benzinga.com insider trades (as of 3 Jun 2026)
9 Carryover thesis facts — Google TPU contract to 2031 (hashrateindex.com); Anthropic 1GW→3GW+ & OpenAI 10GW (Q1 FY26 call, TIKR); Meta MTIA >1GW (about.fb.com, Apr 2026); hyperscaler capex $700–725B (CNBC, Feb 2026); AVGO ASIC share 55–70% (tradingkey.com); NVDA Spectrum-X ~$10B run-rate & MediaTek inference second-source (Beth Kindig / IO Fund commentary, May 2026)
10 Beth Kindig / IO Fund — networking-inclusive FY27 framing (~$140B); now corroborated by the printed 40% Q2 networking mix — io-fund.com / beth-kindig.medium.com (Dec 2025 – May 2026)
All Q2 FY26 financials sourced from primary SEC 8-K (filed 2026-06-03). Customer-concentration figures cited in Key Risks are Citi analyst estimates re-circulated by Beth Kindig — treat as projection, not company guidance. Forward P/E varies materially across open-access providers; page uses ~32× (NTM) as the working figure until verified against a broker feed.
⚠ = figure not verifiable from primary SEC filing or has significant provider dispersion.