Are you ready for 2026? Five pitfalls to avoid

J.P. Morgan Private Bank’s 2026 Outlook argues AI-driven demand is reshaping markets — next-gen chips are selling out, North American data‑centre vacancy is just 1.6%, and power demand from AI could exceed the annual output of Texas and California combined — while the median tech IPO now occurs at 14 years and ~$220m revenue. The report warns inflation remains persistent and recommends real assets (global infrastructure with historical 8–12% returns) and allocations to data‑centre REITs, top chipmakers, utilities, private managers, European defence and industrials (Europe aiming for ~3.5% of GDP on defence and expecting ~20% earnings growth for top defence firms), energy/grid modernisation, cybersecurity, and select emerging‑market exposures (India, Taiwan/South Korea semiconductors; Latin America controlling ~40% of global copper and trading near 10x forward P/E).

Analysis

Market structure: The clear winners are “picks-and-shovels” providers — data‑centre REITs (EQIX, DLR), leading foundry/wafer equipment (TSM, ASML, NVDA exposure) and regulated utilities (NEE) because data‑centre vacancy is already 1.6% in North America and AI power demand is forecast to exceed the combined annual output of Texas and California. Losers include long‑duration bonds, low-yield cash and cyclical consumer real estate; pricing power shifts to asset owners of scarce infrastructure and proprietary chip fabs, while IPOs and private funds capture more upside before public markets see it. Cross‑asset: higher real assets demand should push commodities (Cu, Li) and energy prices up, lift USD vs fragile EMs intermittently, and steepen yield curves — expect equity–bond correlations to turn positive during inflation shocks.

Risk assessment: Tail risks include export controls on advanced nodes (6–12 month shock), a material inflation re‑acceleration triggering a 75–100bp Fed move within 3–6 months, or a major grid/cyber outage impacting data centres. Immediate (days) risk is headline volatility around CPI/AI earnings; short term (3–6 months) is capex guidance and supply responses; long term (2–5 years) is overbuild risk in chips/data centres and geopolitically driven supply re‑routing. Hidden dependencies: data centres’ returns hinge on power costs and transmission upgrades; semiconductor concentration in Taiwan/Korea (TSM, Samsung) is a systemic single‑point risk. Catalysts: major model training cycles, foundry capacity announcements, defense budget votes (EU) and monthly CPI prints.

Key Decisions for Investors

• Establish a 2–4% long position in data‑centre REITs EQIX and DLR (split) over 30–60 days; use 6‑month cash‑secured puts to target entry 8–12% below current levels and size puts to 50% of intended position to lower cost basis if vacancy remains <3%.
• Allocate 2–3% to semiconductor exposure via NVDA and TSM (prefer NVDA for AI upside, TSM for foundry leverage); hedge 25–50% of delta risk with NVDA 3‑month call spreads 5–10% OTM to limit premium outlay while capturing likely catalysts (earnings, product launches).
• Rotate 2–3% from long‑duration growth into real assets: buy IGF or BIP for infrastructure and 1–2% TIP (TIPS ETF TIP) + 1% GLD to defend purchasing power; re‑evaluate after each monthly CPI print and increase TIPs by +1% if core CPI >0.3% m/m.
• Initiate a pair trade: long 1–2% LMT vs short 1–2% BA to capture defense budget tailwinds vs commercial cyclicality; close or rebalance if EU defense spending delays exceed 90 days or Boeing commercial order cancellations rise >15% quarter‑over‑quarter.
• Short duration bond exposure: reduce TLT-equivalent duration by 30–50% and buy 3–6 month TLT puts or use an inverse Treasury ETF sized at 1–2% of portfolio if CPI prints two consecutive months >0.3% m/m, as this historically precedes stock–bond drawdowns.