Why you shouldn’t worry about AI eating the stock market, top analyst says. The U.S. economy is ‘about to take off’

A week-long rout in software stocks sparked by Anthropic’s Claude wiped roughly $1 trillion of market value, but Apollo chief economist Torsten Slok argues the macro outlook is decidedly positive. He points to three growth pillars: already-committed data-center financing (Google, Amazon and Meta disclosed a combined $660 billion of 2026 capex), Bank of America’s forecast of AI capex rising toward $1.2 trillion by 2030, and reshoring of manufacturing and defense production alongside expansionary fiscal policy (CBO projects government spending to add 0.9 percentage points to GDP growth this year). Slok warns stronger real-economy momentum could force the Fed toward tightening later in 2026, creating upside growth with attendant inflation and rate risks.

Analysis

Market structure: The immediate winners are hyperscalers and AI-infrastructure suppliers (GOOGL, AMZN, META, NVDA indirectly) and industrials tied to reshoring (semiconductor fabs, construction, defense). Small/high-duration software names and sentiment-driven SaaS valuations are losers as capex shifts to physical assets; expect gross margin pressure for unprofitable SaaS if funding tightens. Data-center capex already committed for 2026 creates an implied floor for server, GPU, copper and power demand, tightening component supply over 12–36 months and supporting pricing power for suppliers.

Risk assessment: Tail risks include a rapid unwind of AI hype that forces capex cancellations, abrupt Fed hikes if growth/inflation surprise to the upside, and power/permitting bottlenecks delaying projects. Near-term (days–weeks) expect elevated equity volatility; medium-term (3–12 months) watch CPI, payrolls, and hyperscaler capex updates; long-term (1–3 years) reshoring should raise real wages and structural inflation. Hidden dependencies: data-centers rely on grid upgrades, semiconductor supply cycles, and immigration/labor policy — failures amplify cost inflation and project delays.

Trade implications: Favor selective longs on hyperscalers with visible capex (GOOGL, AMZN, META) and materials/industrial plays (FCX, XLI, select defense names) while trimming high-duration SaaS exposure (IGV-like). Rate-sensitive assets: position for higher yields — short 10y Treasuries on CPI or payroll upside; use options to control downside. Catalysts to monitor: monthly CPI >0.3% m/m, 10y yield crossing +40bps, and Q1 earnings guidance from GOOGL/AMZN/META within 30–60 days.

Contrarian angles: The market’s $1tn software markdown likely overstates permanent economic damage — private-sector capex and manufacturing job growth are underpriced. Historical parallels (BlackBerry) are imperfect: here capex is tangible and contracted, making a macro overheating outcome more probable than a systemic equity collapse. The key unintended consequence is Fed tightening that compresses growth multiples even as real economy expands; size positions to survive a 50–100bp rate shock.

Key Decisions for Investors

• Establish a 5% aggregated long allocation split 2.5% GOOGL and 2.5% AMZN via 6–9 month 5/10% OTM call spreads to capture AI/data-center capex upside while limiting premium; take profits if either rallies >30% or if company guidance reduces 2026 capex commitments.
• Initiate a 2% long in FCX (Freeport) or 2% in a copper proxy (CPER) with a 6–18 month horizon to play increased materials demand from data centers and reshoring; add on pullbacks >15% and trim if copper futures fall >20% from current levels.
• Establish a 3% short exposure to high-duration software via IGV using 3‑month put spreads (10–20% OTM) as a hedge against multiple compression; unwind if CPI <0.1% m/m and 10y Treasury yield drops below 3.60%.
• Put on a 3% notional short in 10‑year Treasury futures (or buy equivalent yield‑sensitive options) contingent if CPI prints >0.3% m/m or NFP beats consensus by >150k in next two reports; target a 30–50bp move in yields, stop-loss at a 20bp adverse move.