AI Bears Will Watch the Party Through the Window: Ives

Recent company results from Salesforce, MongoDB and others are being interpreted as validation of an accelerating AI-driven software cycle, with adoption still nascent (only ~3% of U.S. companies) and acute chip tightness (cited ~12:1 demand/supply for video chips). The article highlights deepening partnerships (Snowflake with Anthropic), an expected $3–4 trillion of related spending over the next four years, and a U.S.–China semiconductor/tech arms race that will shape market winners and losers, underpinning continued momentum and a view that tech equities could rise roughly another 10% next year.

Analysis

Market structure: The immediate winners are cloud-native AI infrastructure and data-platform vendors (SNOW, MDB, CRM) plus GPU/ASIC suppliers; chip suppliers see >10x stated demand imbalances (article cites ~12:1) supporting pricing power into 2026. Losers are legacy on-prem vendors and firms exposed to China-only supply chains (Huawei competitors); expect software ASPs to rise 5–15% as AI-addons are monetized, compressing margins for resellers. Cross-asset: stronger tech capex pushes real yields up and steepens the curve (pressure on long bonds), increases equity vol in AI names, and supports semiconductor cyclicals and copper/energy demand for data centers.

Risk assessment: Tail risks include an aggressive US-China decoupling or export-control tightening that could trigger a 20–35% drawdown in chip/AI-exposed names within weeks; regulatory AI safety rules or model bans could shave 15–25% off growth expectations over 12–24 months. Time horizons: expect earnings-driven volatility in days/weeks, supply/demand tightness to persist 6–18 months, and adoption ramp (3%→20–30%) over 2–3 years. Hidden dependencies: data-center power limits, cloud pricing moves, and enterprise procurement cycles (often 2–6 quarters) will govern realized revenue.

Trade implications: Tactical: overweight SNOW (material Anthropic integration optionality) and MDB for data gravity, add CRM as defensive enterprise AI exposure; trim CRWD/MSFT relative exposure where growth fails to justify multiples. Use pair trades (long SNOW, short CRWD) to express divergence; implement 3–9 month call spreads on SNOW/SWDB to capture re-rating while selling some vol. Rotate 5–10% weight from cyclical industrials into software and semis now, rebalancing after next two earnings windows.

Contrarian angles: Consensus underestimates friction—only 3% current adoption implies runway, but the market may overpay for “expected” 3–4 trillion spend; if enterprise ROI proves slower, multiples compress. Historical parallel: cloud migrations took multiple waves (2009–2013) before broad monetization; a bifurcated outcome (few winners capture >50% economics) is likely. Unintended consequence: geopolitical bifurcation could force duplicate supply chains, raising capex and benefiting non-China manufacturing beneficiaries (India, Taiwan).

Key Decisions for Investors

• Establish a 2–3% long position in SNOW (Snowflake) equity now, paired with a 6–9 month 15% OTM protective put (or buy a 6–9 month 30–40 delta call if you prefer optionality); target +30–50% upside over 12 months if Anthropic integration accelerates enterprise adoption.
• Add a 1–2% long position in MDB (MongoDB) and a 1% long in CRM (Salesforce) to capture platform and enterprise AI spend; set stop-losses at -12% and review after next two quarterly earnings (Q1/Q2 2026) for repricing on adoption metrics.
• Implement a relative-value pair: long SNOW (1.5%) vs short CRWD (0.75%) to express divergence between platform monetization and security vendor re-rating; maintain for 3–9 months and trim if SNOW trades >25% above entry or CRWD underperforms by >20%.
• Reduce cyclical industrial exposures by 5–10% and rotate into semiconductors/AI-capable hardware suppliers (non-China-heavy), increasing this allocation if chip supply tightness persists beyond 6 months; actively monitor US-China export control headlines—if new restrictions pass in next 30–60 days, increase hedges and cut China-dependent names by an additional 5–10%.