Sector Outlook — 2026 H2 (Semis Overstretched)
Trigger prompt: "What industry/sectors would perform best in the next half year? Semiconductors are too overstretched in the past year."
Semis are indeed stretched (TSM +46% 6M, AMD +143%, MU +310%, ASML/AMAT/LRCX all +69–141%) —
mean-reversion risk is high. Below are the higher-alpha directions for the next 6 months, ranked by conviction.
🥇 Tier 1 — AI Power + AI Physical Infrastructure
Thesis: AI DC buildout is locked in → power / cooling / construction / interconnects.
Semis are "making the chips"; these names are "powering, cooling, and racking them." Cycle lags semis by
~6 months and valuations haven't yet absorbed the hyperscaler-capex premium.
| Ticker | Role | 6M |
|---|
| VRT | Liquid cooling — the one already tracking semis | +83% |
| GEV | Grid + generation (GE Vernova) | +58% |
| ETN | DC power management | +25% |
| PWR | DC construction leader | +58% |
| APH | High-speed connectors (GB200 NVLink) | +19% |
| ANET | AI switching (NVDA Spectrum-X standard) | +20% |
Among these, VRT is the only one that has already caught up to the semi rally; the rest have material catch-up room.
🥈 Tier 2 — Software AI Applications (Beaten Down)
Thesis: While semis rallied, software SOLD OFF 20–40% (MSFT −23%, ORCL −24%,
PLTR −42%, NOW −36%, CRM −40%). This reflects the AI-capex vs. AI-monetization
transmission gap. In H2 the story will pivot toward "AI revenue" and hyperscaler capex payoff —
software is where that narrative catch-up runs.
| Ticker | Rationale |
|---|
| MSFT | Highest conviction — Azure AI monetization; $2.77T + FPE 19 + 66% of 52WH |
| CRM | Cheapest AI-agent SaaS by valuation (FPE 10) |
| PANW | AI cybersecurity — long secular, breaking out |
| CRWD | AI cybersecurity — same secular, moat |
🥉 Tier 3 — Financials + Utility (Rate-Cut / Defensive Rotation)
Thesis: If the Fed accelerates rate cuts in H2 2026 (current expectation),
REITs + Utility + Regional Banks are the biggest beneficiaries. Also the classic destination for
"de-crowd from semis" flows.
| ETF / Ticker | Role |
|---|
| XLU | Utility ETF |
| XLF | Financials |
| KRE | Regional banks |
| NEE / DUK / SO | Utilities directly benefiting from AI-DC power demand |
Note: NEE / VST / CEG are already crowded into this story. Look for laggards within the group before adding.
⚠️ Clear Avoids
- Semis broadly — reduce all core semi positions except NVDA / AVGO / TSM / AMD; take MU profits.
- Small-cap consumer / retail — cycle hasn't turned.
- Solar / EV / clean-energy narrative names — TSLA / FSLR both in downtrend, no AI tailwind.
- Commodities — unless DXY clearly weakens.
Tactical 6-Month Allocation
30% Semi Core (NVDA + AVGO + TSM + AMD only — half-size vs current; take MU)
15% AI Phys Infra (VRT + ANET + GEV + APH)
20% SW Core (MSFT + CRM + PANW + CRWD)
15% Util / REIT / Financials (rate-cut plays)
10% Cash (opportunistic reload if semis correct 15%+)
10% Wildcards (PLTR + NBIS + SNOW — AI-narrative rebound bets)
Key Judgment
The 300% semi run was a single-pole NVDA narrative. The next leg of alpha is in:
- The "next-door" ecosystem to NVDA — power / cooling / networking
- The AI-monetization layer — software agents catching up to the capex story
Do not chase semis higher, but do not turn anti-AI just because semis are crowded —
rotate to a different layer of the same trade.
Caveats
- All 6M / valuation numbers are 2026-06-29 close snapshots. Refresh before making sizing decisions.
- "Reduce semis" assumes the current bull cycle continues without a regime break. A hard correction
(SPY −8%+) would invalidate the rotation thesis — cash weight should rise first, then re-enter
semis after 15%+ drawdown.
- Utility / REIT rotation depends on Fed rate-cut delivery. If cuts stall, Tier 3 loses its primary
catalyst — trim exposure.
- Cross-reference with ai_companies_layered_2026_06_29
for the per-ticker AI tier rankings that back these picks.