BlackRock bets on ‘pick and shovel' trade, singling out clear winners in AI spending spree

BlackRock's APAC chief investment strategist Ben Powell says the AI-driven capex boom is still accelerating, with hyperscalers racing to outspend rivals and increasingly tapping credit markets to fund data‑center, chip and power builds. Powell argues the clearest beneficiaries will be hardware and infrastructure 'picks and shovels' — chipmakers, energy producers and copper/wiring suppliers — as hyperscale procurement and power commitments surge; Nvidia briefly topped a $5 trillion market cap and S&P Global forecasts data‑centre power demand could nearly double by 2030. The dynamic reinforces investor focus on suppliers of compute, power and raw materials even as firms risk overspending in a winner‑takes‑all contest; OpenAI fundraising and a potential $1 trillion IPO add further capital‑markets relevance.

Analysis

Market structure: The immediate winners are GPU/chipmakers (NVDA), hyperscaler cloud vendors (MSFT, AMZN, META) and upstream inputs—energy generators, copper miners and data‑centre REITs—because hyperscalers are locking multi‑year supply and power contracts. Pricing power will concentrate at Nvidia/TSMC and large cloud buyers; expect spot GPU supply tightness and elevated ASPs for 12–24 months while fab and power expansions ramp. Cross‑asset: rising capex implies more corporate credit supply (pressure on IG spreads), higher commodity (copper, natural gas) spot prices and upward pressure on industrial FX like AUD/CAD versus EM currencies tied to consumer tech exports.

Risk assessment: Tail risks include US/ally export controls on advanced GPUs, a sharper macro slowdown cutting hyperscaler capex by >20% YoY, or a grid/construction bottleneck that delays rollouts (6–24 months). Short term (days–weeks) watch for earnings/capex updates and credit issuance; medium (3–12 months) for fab and power contract announcements; long term (2–5 years) for structural shifts if model developers vertically integrate or foundry capacity overshoots demand. Hidden dependencies: power permitting, long lead times for mines/fabs, and concentration risk in Nvidia/TSMC.

Trade implications: Favor underweight duration and overweight equity exposure to semiconductors (NVDA via LEAP call spreads) and materials/energy (FCX/SCCO, NEE) with 6–24 month horizons; protect via buying 3‑month S&P 5% OTM puts (0.5–1% notional). Consider pair trades: long NVDA vs short high‑multiple AI services/smaller AI software names to isolate hardware demand. Options: buy 9–12 month call spreads on NVDA to capture upside while capping premium; sell short‑dated calls into rallies to monetise IV.

Contrarian angles: Consensus underweights structural grid and permitting risk and overweights perpetual GPU demand; history (2000s data‑centre/telecom cycles) shows infrastructure booms can lead to oversupply and consolidation after 2–4 years. The market may be pricing multi‑year growth into NVDA and select miners—a 20–30% repricing would be triggered by a single major export control or a large foundry capacity expansion. Also consider that OpenAI or hyperscalers pursuing internal silicon could blunt third‑party margin expansion starting 2026–2027.

Key Decisions for Investors

• Establish a 2–3% portfolio long in NVDA via a 9–12 month call spread (buy LEAP near‑money call, sell a higher strike ~25–35% OTM) to capture continued GPU demand while capping premium; add on pullbacks of >15% from current levels.
• Allocate 2–3% to materials and energy: buy FCX and SCCO (1–1.5% each) and 0.5–1% into regulated power names (NEE or EXC) to play copper and power tightness; target a 6–24 month horizon and trim after a 20–30% price move.
• Implement a pair trade: long NVDA (1.5% exposure) vs short 0.75% exposure in a high‑multiple AI services/advertising name (e.g., META) to hedge cyclical ad risk; rebalance if NVDA outperforms by >30% or if macro credit costs rise.
• Purchase downside protection: allocate 0.5–1% to buy 3‑month S&P 500 puts 5% OTM (or equivalent VIX call structure) to insure against a market correction driven by a tech capex shock or credit repricing; reassess at expiration.
• Before scaling positions >5%: monitor three catalysts over next 30–90 days — US export control announcements, hyperscaler Q4 capex statements, and TSMC/ASML capacity guidance — and only increase exposure if no adverse regulatory shifts and confirmations of multi‑year procurement contracts.