Broadcom Q1 FY2026 — Record Revenue, AI Acceleration
Reported Mar 4, 2026 · After Market Close
Total Revenue
$19.31B
Est: $19.14B
+29% YoY · BEAT +$170M
Non-GAAP EPS
$2.05
Est: $2.02
BEAT · +$0.03
AI Semi Revenue
$8.4B
Guidance was $8.2B
+106% YoY · ABOVE guide
Free Cash Flow
$8.01B
41.6% FCF margin
$10.9B returned to holders
Key quote (Hock Tan): "AI revenue growth is accelerating — we expect AI semiconductor revenue to be $10.7 billion in Q2, with total Q2 revenue of $22 billion (+47% YoY). Our custom accelerator business grew 140% YoY in Q1." Management confirmed visibility of >$100B in AI chip revenue for 2027, backed by multi-year supply agreements through 2028. Infrastructure software (VMware) at +1% YoY remains the drag; however, Q2 guidance of +9% YoY suggests recovery. $10B share repurchase authorized. Dividend raised to $0.65/qtr (+10%).
Catalysts vs Risks · Forward 12M
🚀
Catalysts
Google-Broadcom TPU agreement extended through 2031 (Apr 2026). Multi-year custom AI silicon partnership locked in for 5+ additional years. Exactly the structural anchor catalyst that anchors a 5-10y compounding thesis: contracted revenue visibility into 2031, beyond which renewal becomes the real moat test.
Q2 FY26 earnings (Jun 4). AI revenue beat + Q3 guidance above consensus = leadership reaffirmed. Hock Tan has visibility on $10.7B Q2 AI guide. Any number above $11B re-prices the trade aggressively.
Anthropic + OpenAI as new XPU customers. Both have been added as hyperscaler-tier custom accelerator clients. Disclosure of contract sizes / volumes during the Jun 4 call would diversify the customer base structurally.
$100B AI chip revenue 2027 target. Management's stated ambition, backed by $73B AI backlog over 18 months. Represents 3× current annualized run rate. Credible trajectory given +106% / +140% sequential acceleration.
FCF machine: $32B annualized. Capital return of $10.9B per quarter ($7.8B buybacks + $3.1B dividends). At this rate, 10%+ of market cap returned annually. Buyback floor in any drawdown.
VMware re-rating opportunity. Q2 guide +9% YoY vs. Q1 +1%. If subscription conversion accelerates, infrastructure software flips from drag to multi-billion incremental.
⚠️
Risks
Valuation P/E 82x · zero margin of safety. +44% from Mar 31 ($299) to Apr 23 ATH ($429). Morningstar fair value $413 = at current price. Multiple compression to 50x = 35–40% drawdown scenario on any miss.
Customer concentration: 2-3 hyperscalers = ~60%+ AI revenue. Google is by far the largest. Mitigated by 2031 deal extension and Anthropic / OpenAI customer additions, but structural risk persists. Morningstar explicitly flags this.
Marvell-Google talks (Apr 19, 2026). Reported negotiations — not signed — for two AI chips: Memory Processing Unit + inference-focused TPU. Marvell would be 3rd design partner alongside Broadcom (anchored through 2031) and MediaTek. Signals multi-vendor world for hyperscalers, but additive to Broadcom not displacing. Monitor: design-win conversion rate over next 12-18 months.
AI capex normalization. OpenAI user/revenue miss in late April shook the entire AI trade — first signal that the hyperscaler capex chain is not absorbing every narrative without scrutiny. AVGO is the highest beta in the cluster (β 2.09).
TSMC capacity constraint. AVGO previously warned advanced packaging is "maxed out." Constrains the ability to monetize the +140% AI demand trajectory, capping near-term revenue upside.
VMware integration risk. Still absorbing $40B+ acquisition debt (D/E 0.83x). Enterprise IT spending slowdown could pressure VMware ramp to the promised $8.5B run-rate.
Consensus Counterpoint · Bull vs Bear Frame
📈
Bull Thesis · Consensus
Google-Broadcom TPU partnership extended through 2031. Apr 2026 deal lock-in for 5+ additional years confirms the structural moat. Custom XPU dominance vs NVDA standardized GPUs — purpose-built per customer (Alphabet, Meta, Anthropic, OpenAI) with cost/efficiency edge.
$20B AI rev 2025 → $100B+ visibility for 2027. Hock Tan target with multi-year supply agreements through 2028. ~5x AI revenue trajectory transforms the entire financial profile. Anthropic + OpenAI as new hyperscaler XPU customers add diversification.
FCF machine returning ~10% of mkt cap/yr. $32B annualized FCF + $10.9B/qtr capital return. PEG ~0.85 against 41% est EPS CAGR is reasonable on growth, even at P/E 82x.
What this thesis underweights: P/E 82x at ATH leaves zero margin if 41% CAGR slips even modestly; TSMC packaging as physical ceiling on growth conversion; high beta (2.09) amplifies drawdowns in tech selloffs.
📉
Bear Thesis · Consensus
P/E 82x at all-time-high · zero margin of safety. Morningstar fair value $413 = at price. Multiple compression to 50x = 35-40% drawdown scenario on any earnings disappointment or capex normalization signal.
Multi-vendor world emerging. Google in talks with Marvell (Apr 19, not signed) for inference-specific chips, alongside MediaTek as 3rd partner. Even with the 2031 Broadcom extension, hyperscalers structurally diversify suppliers — long-term moat erosion is plausible if not happening today.
AI capex normalization risk. OpenAI revenue/user miss in late April was first canary signal. AVGO has the highest beta in the AI cluster (β 2.09) — drawdowns amplify on any narrative crack.
What this thesis underweights: $73B contracted backlog provides 18-month revenue floor; 2031 Broadcom-Google extension was the most important April news (positive); demand-exceeds-capacity environment means multi-vendor world is additive to total custom XPU TAM, not zero-sum.
Horizon Calibration · Tactical vs Anchor
Score Lens by Investment Horizon
Tactical Score · 12–24m Lens
67/ 100 · Hold / Selective Trim
Headline gauge — drives tactical decisions. Penalizes today's stretch (P/E 82x, EV/EBITDA 48x breaching band, 96% of 52W range) but offset by Apr 2026 Broadcom-Google deal extension through 2031. Action: hold core 55sh; selective trim 5–10 sh above $430 if rally extends; add-zone < $345.
Anchor Score · 5–10y Hold Lens
74/ 100 · Compounding Machine
+7 vs tactical. The 2031 Broadcom-Google deal extension is exactly the type of structural anchor catalyst — it locks in contracted revenue visibility through 2031 and beyond. Today's P/E stretch dilutes over the compounding window where AI rev moves 5x by 2027. Capital return at 10%/yr structurally returns ~half current mkt cap over the decade.
Thesis Breakers · Conditions That Force a 10y Exit
1. Google does NOT renew/expand Broadcom partnership beyond 2031 — non-renewal is the real moat collapse signal, not addition of 3rd-party design partners. Watch closely 2029–2030. 2. Hyperscaler in-housing custom silicon at scale — Apple-style vertical integration adopted by 2+ majors (Meta MTIA scaling to internal sufficiency, OpenAI silicon team ramping past dependence). 3. AI capex sustained decline >24 months without offset from new application categories (post-LLM workload paradigm not absorbing custom silicon). 4. EBITDA margin permanently below 60% — pricing power loss signals commoditization of custom XPU category. 5. Hock Tan succession + strategic drift — execution discipline has been singular and exceptional; succession risk is real if M&A pivot or capital allocation diverges from the FCF-machine model.
10y
Compounding Read
What you actually own at 5–10y horizon: the dominant custom AI silicon franchise, with the Google TPU partnership now contractually anchored through 2031, $73B contracted backlog providing 18-month revenue floor, $32B annualized FCF returning 10%+ of market cap each year, and proven margin discipline (68% EBITDA record). The Marvell-Google talks (not signed) and MediaTek as 3rd partner reflect a multi-vendor world in a compute-constrained environment — Google diversifying suppliers, not abandoning Broadcom. At 10-year horizon, the structural moat test is the 2031 renewal cycle, not the addition of inference-specific challengers today. Today's entry stretch matters less than whether AVGO remains the #1 custom XPU partner — which today is structurally true and contractually anchored through 2031.
Bottom Line · Investment Verdict
Analyst Rating Spectrum
STRONG SELL
REDUCE
HOLD
SELECTIVE BUY ◆
BUY
STRONG BUY
Thesis in one line
Best-in-class franchise with 2031 partnership lock-in, but priced for perfection. The Apr 2026 Broadcom-Google deal extension through 2031 is the most important structural news of the period — anchors visibility and frames Marvell talks as multi-vendor diversification, not moat erosion. The debate is purely valuation at P/E 82x.
Valuation read at $420
EV/EBITDA at ~48x is breaching the 35–45x band. Trailing P/E 82x is the highest level since the stock has been listed. Morningstar fair value $413 = at current price. Zero margin of safety — any miss compresses the multiple aggressively.
Entry discipline
Selective trim 5–10 sh above $430 if rally extends, not aggressive trim. Add-zone < $345. The June 4 earnings is the next binary event — do not chase ahead of the print but the 2031 deal removes the urgency to defend.
AVGO scores 67/100 — SELECTIVE BUY / HOLD. Two structural updates since the prior scoring shift the balance: (1) the Apr 2026 Broadcom-Google deal extension through 2031 locks in custom XPU partnership for 5+ additional years — exactly the type of multi-year visibility that supports both tactical and anchor cases; (2) the Apr 19 Marvell-Google news is talks for inference-specific chips, not a signed deal, and Marvell would be 3rd partner alongside Broadcom + MediaTek in a compute-constrained multi-vendor world. Net: the structural moat is reinforced, not eroded. The only remaining risk is valuation — P/E 82x at ATH gives zero margin of safety. Optimal action: hold core 55sh position, selective trim 5–10 sh above $430 if rally extends, monitor the June 4 print, set add-trigger only below $345.