Nvidia still waiting on Chinese H200 revenues to materialize

Nvidia projects continued strong demand in datacenter AI despite receiving no Chinese datacenter revenue and uncertainty over H200 approvals by Beijing after the U.S. allowed limited exports (with a reported 25% revenue cut arrangement). Management forecast Q1 FY2027 revenue of $78 billion (+/−2%), following Q4 results with $68.12 billion in revenue and $42.96 billion in profits, and FY2026 net profit of $120 billion on $215.93 billion of revenue; datacenter sales accounted for roughly $62.3 billion (>90%) of Q4 revenue. The company warns of long‑term risks from technological decoupling, while CEO commentary flagged speculative opportunities (and current economic challenges) for GPU deployment in orbital datacenters.

Analysis

Market structure: Nvidia (NVDA) is the clear short-term winner — datacenter accounted for >90% of last quarter's revenue and management guides Q1 to ~$78B without any China datacenter revenue, implying hyperscalers (MSFT, GOOGL, AMZN, META) remain the primary demand source. Limited or delayed China access preserves NVDA pricing power and margin expansion while keeping smaller GPU competitors and Chinese chipmakers starved of high-end nodes, tightening supply/demand for cutting-edge accelerators for at least the next 6–12 months. Cross-asset: tech equity outperformance should continue to compress IG spreads, lift equity vol skew (NVDA), put modest upside pressure on USD vs CNY if trade tensions persist, and increase copper/steel demand modestly via datacenter buildouts over 12–24 months.

Risk assessment: Tail risks include Beijing refusing H200 imports or the U.S. widening export controls — either could remove a potential multi-billion revenue stream and create a 10–25% EPS downside scenario over 12 months depending on substitute sourcing. Near-term (days/weeks) risk = headline-driven IV spikes; short-term (3–6 months) = hyperscaler capex cadence and China decisions; long-term (2–5 years) = competitive threats from custom accelerators (Google TPU, Intel/SambaNova) and Chinese incumbents. Hidden dependencies: NVDA’s trajectory hinges on sustained hyperscaler capex, TSMC capacity, and software ecosystem lock‑in; catalyst watch: China approval window (next 30–90 days), quarterly capex guides from MSFT/GOOGL/AMZN.

Trade implications: For directional exposure, prefer concentrated NVDA rather than broad AI small caps — establish a 2–3% long NVDA equity position via a 9‑month call spread (buy ~30-delta, sell a call ~30% OTM) to cap premium spend and capture upside if hyperscalers keep buying. Pair trade: long NVDA (2%) / short AMD (1–1.5%) to express NVDA’s share shift; hedge event risk with calendar puts (6 months) sized 0.5% of portfolio if China remains closed after 60 days. Rotate 2–4% from cyclical tech into hyperscalers (MSFT, GOOGL, AMZN) ahead of capex updates; short small infra names (CRWV) 0.5–1% via puts where balance sheets are weak.

Contrarian angles: Consensus underestimates two risks — valuation concentration (NVDA already prices multi-year growth) and policy-driven revenue grabs (the U.S. 25% revenue carve‑out or Chinese countermeasures). This creates possible mean reversion: if hyperscaler orders slow or China approval occurs but with onerous revenue sharing, upside may be limited while downside is asymmetric. Historical parallel: platform incumbents (Intel in CPU era) lost share when rivals optimized vertically — NVDA’s moat depends on continuous node advantage and developer lock-in; monitor TSMC allocation and hyperscaler software commitments as early warning indicators.

Key Decisions for Investors

• Establish a 2–3% net long position in NVDA using a 9-month call spread (buy ~30-delta call, sell a call ~30% higher) within the next 10 business days to capture continued datacenter momentum while capping premium; trim if NVDA rallies >25% post-earnings.
• Implement a relative-value pair: long NVDA (2% notional) / short AMD (1–1.5% notional) to express NVDA share gains; size to net 1–1.5% portfolio exposure and close the short if AMD reports >10% sequential margin expansion.
• Buy 6-month protective puts on NVDA sizing 0.5% of portfolio immediately; if Beijing fails to approve meaningful H200 imports within 60 days, add puts to reach 2% protection (costly but hedges a 10–25% EPS shock).
• Allocate 1–2% each to MSFT, GOOGL, AMZN (equal-weight) within 30 days to play hyperscaler capex tailwinds; rebalance after their next quarterly capex commentary (target window 60–90 days).
• Short CRWV (CoreWeave) or buy 3–6 month out-of-the-money puts for a 0.5–1% position size on the thesis small infra providers will face margin pressure and funding risk if capital shifts to NVDA/hyperscalers; cover if CRWV secures strategic hyperscaler contracts within 90 days.