Can the Stock Market Keep Climbing in 2026? Ignoring This Trend Could Cost Investors Dearly.

The S&P 500 has delivered outsized gains—about 24% in 2023, 23% in 2024 and more than 14% YTD in 2025—leaving it roughly 75% above the start of 2023 and nearly doubled since the October 2022 low. Historical analysis shows that of 37 prior instances where the index rose three years in a row, 65% continued for a fourth year, suggesting further gains are more likely than not, even as valuations look historically rich. J.P. Morgan estimates AI-related capex added roughly 1.1 percentage points to U.S. GDP in H1, underscoring AI’s role in the rally but also raising bubble risks whose timing is unknowable; the piece concludes that long-term trends favor equities while advising caution around timing.

Analysis

Market structure: AI-driven capex is concentrating demand into a narrow set of winners — Nvidia (NVDA), hyperscalers (MSFT, AMZN), semiconductor equipment (ASML/AMAT) and data‑center REITs — giving these names near-term pricing power for GPUs, wafers and colo capacity. Losers are broad cyclicals, small-cap discretionary and legacy industrials whose earnings are insensitive to AI; expect higher concentration risk in index returns (top-10 names >30% of S&P performance). Cross-asset: equity risk premium compression supports higher bond yields sensitivity; equities upcycle will raise equity implied vols for NVDA and friends, push EM FX weaker on USD risk‑on corrections, and increase industrial metals and power demand over quarters.

Risk assessment: Tail risks include a rapid AI funding pullback (funding velocity fall >30% QoQ), export/regulatory shocks to advanced-node chips, or Fed surprise tightening that widens credit spreads 50–100bp. Immediate (days) risks: earnings/guide from NVDA or hyperscalers; short-term (weeks) risks: GPU supply shocks and IV spikes; long-term (quarters/years): capex cycles, software monetization failing to materialize. Hidden dependencies: AI ROI depends on hyperscaler budgets and software monetization models — hardware demand can outpace sustainable revenue for many vendors.

Trade implications: Express convex upside to AI while limiting concentration: use short-dated defined-risk options on NVDA and overweight AI infrastructure stocks versus trimmed S&P exposure (VOO) over the next 3–12 months. Relative value: favor semicap equipment (ASML/AMAT) and cloud infra over small-cap growth; hedge market beta with low-cost SPX tail protection sized to expected drawdown. Time entries around NVDA earnings and next Fed prints; take profits on +20–30% moves or if IV rises >40% from baseline.

Contrarian angles: Consensus underestimates concentration and execution risk — valuations price near-perfect AI outcomes and underprice regulatory/export tail risk. Reaction may be underdone in small-cap value, which historically outperforms after extreme concentration (mean reversion over 6–18 months). Historical parallel: late-1990s tech extended long before a sharp reset; unintended consequence is crowding into a few market-makers and option desks, amplifying crashes when liquidity withdraws.

Key Decisions for Investors

• Establish a 2–3% portfolio exposure to NVDA via a 3‑month call spread (buy 5–10% OTM calls, sell 15–20% OTM calls) to capture AI upside while capping premium; trim/close if NVDA rallies +25% or implied vol rises >30 pts.
• Within 2 weeks, reallocate 2% from broad S&P ETF (VOO) into AI infrastructure: 1% long ASML (or AMAT) for equipment-led secular demand and 1% long MSFT for software monetization exposure; review after next quarterly earnings (90 days).
• Purchase SPX 3‑month puts 7–10% OTM sized to 0.5–1% of portfolio as a convex tail hedge against a >10% market correction; reassess cost if put IV >30% and roll/extend only if IV falls below prior cost basis.
• Establish a 1.5–2% short position in high‑beta/sentiment ETF (e.g., ARKK or an equivalent small‑cap growth basket) versus a 1.5–2% long NVDA/ASML pair trade to express relative value; cover shorts on mean reversion within 3–6 months or if short ETF underperforms by >20%.