Concerns & Risks — 6/10

Best-in-class catalyst slate: multi-GW AI commitments across 6 customers, $30B quarterly bookings, visibility to 2028. But China at 17% of revenue exceeds the 10% threshold, and FY27 EV/EBITDA of ~20x sits at peer average rather than below it. Export-control and customer-concentration overhangs are real. Weight: 15%
Fwd EV/EBITDA (FY27)
~20x
vs ~21x peer avg — at peer
China Exposure
17%
Above 10% threshold; declining from 20%
AI Bookings
$30B
In-quarter FQ2'26; visibility to 2028
Customer Concentration
~40%
Top-5 rev share; 6 core AI customers
Valuation
Metric FY2027E Multiple Peer Avg
EV / EBITDA $93.3B EBITDA ~20.0x ~21x
P / E $19.43 EPS ~19.7x ~32x (NVDA)
EV / Sales $173.3B rev ~10.8x
~20x is roughly in line with NVDA but at large discount to MRVL (~54x). Only "cheap" if EBITDA roughly doubles FY26 to FY27.

China Exposure
Detail Value
FY2025 China/HK % of Net Revenue 17%
FY2024 China/HK % of Net Revenue 20%
Direction Declining
Threshold 10%

Shipments booked to China/HK; end customers are often located elsewhere. Export controls on advanced compute and networking are a persistent tail risk. Direction is declining but still above the 10% comfort threshold.


Catalysts
Catalyst Detail
FQ3 FY26 AI Semi $16.0B guided, +200%+ YoY
FY2026 AI Semi Total ~$56B, +180% YoY
FY2027 AI Semi Guide >$100B reiterated
AI Bookings $30B in-quarter; visibility to 2028
OpenAI Production late 2026; 1.3GW FY27, 10GW by 2029
Anthropic 1GW in 2026, +5GW from 2027
Meta MTIA 1GW initial, 3GW through 2028
Apollo / Blackstone AI First $35B tranche; 20GW+ through ~2030
Networking Tomahawk 200T tape-out; 1.6T CPO

Regulatory & Key Risks
Risk Severity Detail
Export Controls HIGH Restrictions on high-end compute and networking chips to China. Persistent and tightening. 17% of revenue directly exposed.
Customer Concentration HIGH Top-5 customers ~40% of revenue. Single hyperscaler insourcing risk is real; 6 XPU customers mitigates but does not eliminate.
Retaliatory Tariffs MEDIUM Broad tariff escalation could disrupt supply chains and increase costs for networking and infrastructure hardware.
Gross Margin Mix Shift MEDIUM XPU mix stepping up as a share of AI semi; rack-level and HBM pass-through dilute gross margin. Operating margin target ~67% holds for now.
CFO Transition LOW Orderly handoff from Spears to Teiner. No operational disruption expected but new CFO untested at this scale.

Bull vs Bear
Bull Case
  • "Insatiable" demand with $30B quarterly bookings and visibility to 2028.
  • AI semi compounds from $56B to $100B+.
  • 6 XPU customers broadening the base beyond any single hyperscaler.
  • VMware at 93% gross margin is a high-quality annuity stream.
  • ~20x FY27 EV/EBITDA in line with NVDA; operating margin holds ~67%.
Bear Case
  • 17% China exposure + export escalation risk.
  • Top-5 = 40% of revenue; insourcing threat is structural.
  • FY27 multiple only works if EBITDA roughly doubles.
  • Gross margin stepping down on XPU mix shift.
  • CFO transition adds marginal uncertainty.

Score Rationale

Score of 6/10 reflects a catalyst slate that is among the best in semiconductors but material risk factors that prevent a higher mark. The $30B quarterly AI bookings and multi-GW commitments from 6 customers provide exceptional forward visibility. However, China at 17% of revenue exceeds the 10% comfort threshold, FY27 EV/EBITDA of ~20x sits at rather than below peer average, and customer concentration (top-5 at ~40%) creates structural insourcing risk. Export controls are persistent and tightening. The valuation only looks "cheap" if EBITDA roughly doubles from FY26 to FY27 — a high bar even with the current demand trajectory. Gross margin mix shift from XPU ramp is a second-order concern. The CFO transition appears orderly but adds marginal uncertainty.


Data sourced from Daloopa (company_id 22), FMP, AVGO 10-K, and FQ2 FY2026 earnings transcript.