Chipmaker Nvidia flouts AI scepticism with record revenue

Nvidia reported record fourth-quarter revenue of $68.1bn, up 73% year-over-year and beating analyst expectations, driving after-hours gains; full-year revenue was $215.9bn and management provided a rosy near-term outlook (notably excluding China chip revenue guidance). CEO Jensen Huang pointed to accelerating AI compute demand as Nvidia expands into new product areas (an open-source autonomous-driving model ‘Alpamayo’ and a planned robotaxi service) and completed a roughly $20bn acquisition of Groq to strengthen inference capabilities. The stock-market reaction and massive revenue print reinforce Nvidia's central role in AI infrastructure, though investors remain wary of deal-related scrutiny over 'circular financing' and geopolitical/export-control uncertainty around China.

Analysis

Market structure: Nvidia (NVDA) is consolidating pricing power in AI training (leadership in high-end GPUs) and is moving to capture inference and end-products via Groq acquisition and robotaxi/platform bets — expect NVDA to account for a growing share of hyperscaler GPU spend (target 30–40%+ of AI capex in next 12–24 months). Winners include GPU ecosystem suppliers (TSMC, ASML, memory makers) and cloud providers; losers are pure-play inference chip vendors and legacy CPU vendors (INTC). Cross-asset: a sustained NVDA rally usually drives equity risk-on, pushes real yields up ~10–30bp, strengthens USD intraday, and raises implied vols in semis while pressuring long-duration bond proxies.

Risk assessment: Key tail risks are expedited US/China export tightening (cuts >20% China revenue), antitrust/competition action, and integration failure of Groq or failed robotaxi monetization; probability medium but impact high. Immediate (days) risk: post-earnings multiple expansion snapback; short-term (weeks–months): order cadence and China sales clarity; long-term (quarters–years): secular AI capex sustainability and margin mix shift from training to inference. Hidden dependencies include Nvidia’s investments/partner financing (“circular financing”) masking true end-customer demand and hyperscaler budget cycles.

Trade implications: Direct asymmetric exposure via NVDA long and selective LEAP call spreads captures secular upside while capping premium; hedge with short positions in AMD or INTC to play margin compression in inference. Consider relative-value pair trades long NVDA vs short AMD (or INTC) and long ASML/TSMC suppliers. Options: buy 12–18 month NVDA 20% OTM call spreads and sell nearer-term calls into rallies to monetize IV; use put spreads to protect concentrated positions.

Contrarian angles: Consensus assumes perpetual hyper-growth; missing is saturation/tapering risk in hyperscaler incremental GPU demand and the potential for competitors/sovereign fabs to undercut margins in inference. The post-earnings rally may be overdone short-term — a 10–15% pullback would be healthy entry; historical parallel: cyclical hardware leaders that concentrated on one node saw 30–50% subsequent drawdowns when market breadth rotated. Unintended consequence: Nvidia moving downstream (robotaxi/platform) could attract regulatory scrutiny and capital allocation trade-offs that compress core GPU capex ROI.

Key Decisions for Investors

• Establish a 2–3% long position in NVDA via a 12–18 month 15–25% OTM LEAP call spread (buy calls ~20% OTM, sell a higher strike 40–60% OTM) to capture secular AI training upside while limiting premium risk; scale in if NVDA retraces 10–15% from current levels.
• Initiate a pair trade: long NVDA 1–2% of portfolio vs short AMD or INTC 1% (use equities or equal-dollar CFDs) to exploit NVDA’s training moat and expected inference margin pressure on incumbents; tighten the short if NVDA/AMD outperformance exceeds 30% absolute.
• Purchase a 3–6 month protection strategy if NVDA position >2%: buy 5–10% notional of NVDA downside via 3–6 month 10% OTM put spreads to limit tail risk from China export shocks or regulatory actions; reduce hedge if China sales confirmation arrives (clearance or first shipments within 30–60 days).
• Overweight suppliers: add 1–2% positions in TSM (TSM) and ASML (ASML) exposure via equities or ETFs given multi-year wafer and lithography demand; take profits when these positions appreciate 25–35% or if order flow from hyperscalers meaningfully decelerates over two consecutive quarters.