Trump administration greenlights Nvidia AI chip exports to China

The Trump administration's Commerce Department will ease export restrictions to permit Nvidia to ship H200 AI chips to approved customers in China and other countries in exchange for a 25% fee to the U.S. government, shifting license reviews from a presumption of denial to case-by-case, requiring U.S. third-party testing, capping China receipts at 50%, and banning military use. Reuters reported Chinese firms ordered more than 2 million H200s at about $27,000 each while Nvidia's current inventory is roughly 700,000 — a move that could materially increase Nvidia's addressable market and revenue but is constrained by supply, vetting, and national-security conditions.

Analysis

Market structure: The relaxation is a clear near-term revenue lever for NVDA—Chinese orders reportedly >2.0M H200s vs ~700k inventory and ~$27k list price implies >$54B potential demand backlog; the 50% cap and 25% fee blunt but do not eliminate upside. Winners: NVDA (datacenter ASPs and cloud-rental pricing), TSMC/ASML (capacity), and cloud providers (AMZN, MSFT, GOOGL) that can monetize higher GPU rents; losers: smaller GPU makers and export-restriction trade-arbitrageurs who priced permanent denial. Market pricing power for NVDA rises materially through H1–H2 2026 as production ramps are constrained by wafer capacity and testing bottlenecks.

Risk assessment: Tail risks include abrupt policy reversal or stringent enforcement if chips are used militarily, which could wipe 20–40% of near-term China demand; supply-chain shocks at TSMC or U.S. mandated testing bottlenecks are 5–15% probability but high impact. Immediate (days): volatility around Jan 15 rule text; short-term (1–6 months): production ramp, license approvals; long-term (2–4+ years): Chinese domestic substitution risk and IP transfer that could erode NVDA’s moat. Hidden dependencies: who economically bears the 25% fee (NVDA vs buyer) and U.S. third-party test throughput are choke points that determine realized margin.

Trade implications: Tactical: establish a modest long NVDA equity and options exposure to capture 3–6 month export-driven upside but cap risk with defined-risk structures. Recommended: buy NVDA 3–6 month call spreads (e.g., 3-month buy 10% ITM / sell 30% OTM) sized to 2–4% portfolio; also buy 12–18 month LEAPs (1–2% portfolio) for secular AI exposure. Pair trade: long NVDA / short AMD (equal-dollar) to express NVDA datacenter share gain while hedging broad GPU beta; rotate into semicap suppliers (TSM, ASML) and cloud infra (AMZN, MSFT).

Contrarian angles: Consensus may be overestimating speed and magnitude of China revenue because of the 50% cap, testing delays and the ambiguous 25% fee—market may be underpricing enforcement friction. Historical parallels (Huawei sanction cycles) show rapid policy flip is possible; unintended consequence: enabling Chinese scale could accelerate domestic competitors within 24–48 months, creating a multi-year risk to NVDA’s pricing power. Risk control: set objective sell triggers (policy reversal, quarterly China shipments >25% revenue, or NVDA margin compression >200bps).

Key Decisions for Investors

• Establish a 2–4% long position in NVDA (NVDA) within 1–2 weeks, targeting a 3–6 month hold to capture export-led revenue; hedge with 1–1.5% portfolio equivalent in NVDA 3-month call spread (buy 10% ITM / sell 30% OTM) to limit downside and cost.
• Initiate a 1% portfolio long-dated LEAP position in NVDA (12–18 month expiry) to retain secular AI exposure; size so that total NVDA exposure (equity + options) does not exceed 6% of portfolio to manage policy reversals.
• Implement a pair trade: long NVDA / short AMD (AMD) equal-dollar (size 1–2% net per leg) to express datacenter share capture; unwind if AMD releases a validated H200-class competitor or NVDA announces China shipments >25% of quarterly revenue.
• Overweight TSM (TSM) and ASML (ASML) by 1–2% combined to play foundry and lithography capacity tailwinds; underweight defense contractors benefiting from decoupling (reduce exposure by 1%) where export easing reduces near-term upside.
• Set hard risk triggers: reduce NVDA exposure by 50% on any of the following within 0–6 months—(a) formal U.S. license revocation or tightening, (b) verified Chinese military use, or (c) NVDA quarter where China shipments >25% of revenue; take profits if NVDA shares rise >30% in 60 days absent material confirmations of sustained shipments.