Got $10,000? 2 Artificial Intelligence (AI) Stocks to Buy in December and Hold for the Long Term

Nvidia commands north of a 90% share of the GPU market and has deepened its software and hardware moat via CUDA, the NvLink interconnect and the recent SchedMD acquisition (Slurm), positioning it to capture rising AI infrastructure spend and optimize LLM training. Alphabet’s seventh-generation custom TPUs, optimized for TensorFlow, provide a structural cost and power-efficiency advantage for training and inference of its Gemini models and improve cloud margins, while integration across search, Android and ads amplifies distribution and monetization advantages.

Analysis

Market structure: Nvidia (NVDA) and Alphabet (GOOGL/GOOG) are structural winners — NVDA retains >90% GPU share and benefits from CUDA, NvLink and recent SchedMD acquisition that raise switching costs; Alphabet’s TPU ASICs create a 20–40% estimated price-performance and power-efficiency advantage for its internal training/inference stack, reinforcing its search/ads distribution moat. Other GPU vendors (AMD) and pure-play AI chip startups face margin compression and tougher go-to-market dynamics as hyperscalers either buy NVDA stacks or vertically integrate with TPUs.

Risk assessment: Key tail risks are antitrust/export controls (probability 20–30% over 12–24 months), supply-chain constraints at TSMC impacting NVDA wafer supply, and hyperscaler in-sourcing of chips reducing third-party spend. Immediate risks (days–weeks) include earnings guidance and news on large cloud contracts; medium-term (3–12 months) inventory and pricing normalization; long-term (2–5 years) is software lock-in vs. ASIC commoditization.

Trade implications: Tactical size NVDA and GOOGL as core AI infra exposure but use option hedges: consider 2–3% portfolio long NVDA with 1% in 3-month ATM puts to cap a >15% drawdown, and 2% long GOOGL via 18-month LEAP calls to capture TPU-driven margin upside. Pair trades: long GOOGL / short AMD (smaller size) to express ASIC vs. GPU structural divergence; trim on >40% outperformance or guidance misses >10%.

Contrarian angles: Consensus underestimates the risk that TPUs and other ASICs could materially reduce NVDA TAM in training over 3–5 years — history (Intel→ARM shifts) shows incumbency can be overturned when cost/energy curves move ~30% in favor of disruptors. Also, SchedMD ownership may not prevent customers from orchestrating heterogeneous stacks; look for early signals (benchmarks, consortiums) before fully committing to single-vendor conviction.

Key Decisions for Investors

• Establish a 2–3% portfolio long position in NVDA within the next 2–6 weeks (stock or 9–12 month call spread); allocate 1% portfolio to 3-month ATM protective puts (roll if >15% weekly IV move) to limit downside while capturing near-term AI infra upside; trim position if NVDA reports revenue guidance miss >10% or GPU ASPs fall >15% QoQ.
• Buy 2% portfolio exposure to GOOGL via 18–24 month LEAP calls (or equivalent stock purchase) to capture TPU-driven margin expansion and ad/search integration; increase allocation by another 1% if quarterly TPU/cloud revenue growth >25% YoY or Gemini adoption metrics accelerate; reduce to neutral if regulatory fines or restrictions >$5B materialize.
• Implement a pair trade: long GOOGL (1.5%) / short AMD (1%) to express relative structural advantage of bespoke ASICs vs. second-place GPU vendor; exit or rebalance if AMD publishes third-party benchmarks showing >15% perf/price advantage or if NVDA share falls below 80% in enterprise GPU procurement surveys.
• Use options for income/volatility: sell covered calls on up to 50% of NVDA holdings with strikes ~15–25% OTM and 3–4 month expiries to fund protective LEAPs or puts; if regulatory/export headlines spike (within 30 days), buy 60–90 day puts on NVDA sized to 0.5–1% portfolio as tactical hedges.