AI Stocks Are Stuck – and Not Because of Earnings

The AI trade has been stuck for ~4 months—AIQ is near the same level as the first week of May—despite improving AI fundamentals, attributed to excess liquidity turning negative (largest G10 crash since COVID). Bond-market dynamics also complicate a “lower yields” narrative: 30-year yields hit 5.3% (highest since 2007) and gross federal debt is ~123% of GDP, while corporate bond issuance totals $1.68T through July (+27%), yet credit spreads barely move, implying investors aren’t taking more risk. The article points to a “follow demand, not the multiple” playbook and highlights a potential copper squeeze: copper is up 36% YoY, with ~650,000 tons stockpiled and Goldman estimating ~$14,000/ton if an imported refined-copper tax is implemented.

Analysis

This is a multiple-compression regime, not an earnings recession. When excess liquidity is contracting, the market stops paying for distant cash flows, so the highest-duration AI beneficiaries can lag even as operating data keeps improving. That argues for continued dispersion: profitable, contract-backed suppliers and bottleneck assets should hold up better than names whose valuation still depends on a re-rating.

The key risk is that falling yields are not automatically bullish. If the 10-year rolls over because growth or credit is cracking, the first casualties are levered growth baskets and the second-order hit is to enterprise capex, ad-tech, and semis exposed to budget scrutiny. Immediate rallies can last days, but the cleaner 1-3 month read is credit spreads; if they do not tighten, de-rating can persist into year-end.

Contrarian view: consensus may be underpricing how sticky AI spend becomes once infrastructure is already budgeted, which favors copper, power, and select suppliers over the marquee platform names. The better long is not AI beta but bottleneck exposure with balance-sheet flexibility and pricing power. Falsifier: a sustained break in the 10-year below 4.5% alongside tighter credit spreads and renewed AIQ/SMH breadth would signal the liquidity headwind is easing rather than worsening.

Key Decisions for Investors

• Long GS / short TTD as a 1-3 month relative-value expression of lower-duration financials versus high-duration ad-tech. Best entry is on any broad growth rally that fails to make new highs; stop if the 10-year falls below 4.5% and TTD outperforms financials by ~8-10%.
• Buy FCX or SCCO on a 5-7% pullback as a 6-18 month secular hedge on AI power/grid/copper demand. Risk/reward is attractive if supply stays tight; falsify on a hard-landing tape or a 15%+ copper drawdown.
• Use AIQ or SMH call-sale rallies to trim exposure rather than add. The trade only re-accelerates if liquidity improves; otherwise upside is likely capped until breadth and credit spreads confirm.
• If you need tactical protection, buy 3-month put spreads on AIQ into the next CPI/Fed window. This is a hedge, not a standalone bearish bet; the premium is at risk if yields fall for benign reasons and breadth improves.