Why I don't think AGI is imminent

The author contends that despite large investments and rapid advances in LLMs, there is little evidence AGI has meaningfully arrived for the broader economy—most tangible gains are concentrated in software productivity while job disruption, misinformation and eroded trust are mounting. Investors are visibly allocating capital to compute and data centres, but the piece warns of hype-driven risks (labour displacement, regulatory and ethical challenges, energy/power constraints and potential systemic fragility) and argues that missing orchestration, verification and governance remain the primary barriers to material real-world deployment.

Analysis

Market structure: Winners are large cloud/data‑center owners (GOOGL/GOOG, Equinix) and power suppliers that can monetize incremental data‑center demand; losers are small AI app vendors and legacy content/media that lose trust and pricing power. Big players gain asymmetric pricing power because scale lowers marginal AI cost (model, infra, talent) and raises switching costs; expect data‑center rents and power demand to outpace supply for 12–36 months, pushing upward pressure on power and copper input prices.

Risk assessment: Tail risks include aggressive US/EU regulation (privacy, liability, breakups) or an AI‑driven reputational shock that forces capex pullback—each could trigger >30% re‑rating in small names within 3 months. Immediate (days) risks: headline/model releases and earnings; short term (1–6 months): regulatory hearings and capex cadence; long term (12–36 months): structural labor displacement, energy constraints, or concentration risk around a few winners (systemic counterparty exposure).

Trade implications: Prefer core long exposure to GOOG (AI monetization) and data‑center/power names while shifting small‑cap growth into cybersecurity (CRWD/PANW) and infra (NEE, EQIX). Use defined‑risk option structures (6–12m call spreads on GOOG, long 12m calls on CRWD) and maintain a 1% portfolio tail hedge (deep OTM SPX puts or VIX call spreads) to protect against regulatory/market shocks.

Contrarian angles: Consensus underprices infrastructure monetization and overprices app‑layer hype; historical parallel: post‑1999 internet consolidation where majors captured most value. Mispricing threshold: if GOOG underperforms by >10% on macro noise, add to 3–5% weight; conversely, trim small AI names if revenue guidance misses by >5% or if power costs rise >15% in 90 days.

Key Decisions for Investors

• Establish a 2–3% long position in GOOG (class C) via a 6–9 month call spread (buy 1x ATM+5–10% call, sell a higher strike to fund) targeting 20–40% upside; cut if next‑quarter ad/cloud/AI guidance misses by >3%.
• Rotate 3–5% from small‑cap growth into infrastructure: buy 2–3% NEE (NextEra) for power exposure and 1–2% EQIX (Equinix) for data‑center rents; increase allocation if regional power prices rise >10% over 90 days.
• Allocate 1–2% to cybersecurity (long CRWD or PANW) using 12‑month calls 25–35% OTM or small equity positions; take profits on 30–50% moves and add on pullbacks >15%.
• Establish a 1–2% net short basket of overvalued small‑cap AI application names (select 8–12 names, market cap <$1bn, revenue multiples >10x) via put spreads or short basket to capture rerating risk; unwind if major cloud capex guidance contracts <‑20%.
• Buy a 1% portfolio tail hedge: 6–12 month deep OTM SPX puts or VIX call spread to protect against regulatory or systemic shocks; reassess after 90 days or post any major AI regulation announcement.