Situational Awareness: AI as Strategic Infrastructure
How frontier AI may reshape compute, power, chips, cloud, defence and public-market capital flows.
Last updated 26 Jun 2026
AI capability growth may turn compute, power and national-security infrastructure into strategic assets. The winners may be the companies that control cloud, chips, power, data centres and defence infrastructure — but the most crowded AI names may face near-term de-rating risk.
This framework is not simply 'buy AI stocks.' The positioning is a barbell: LONG the physical enablers (data-centre power, cooling, specialized GPU cloud) and SHORT the crowded, richly-valued AI-hardware names (semis, SMH, NVDA, AVGO, AMD, ORCL) via puts. A name can be central to the theme and still be a poor entry — central to AI ≠ a good position today.
- ›Hyperscaler capex is the key confirmation signal — and it is still rising.
- ›Power availability is becoming a binding constraint on the buildout.
- ›Sovereign AI is moving from policy statements to funded programmes.
- ›Crowded AI leaders (notably NVDA) carry growing near-term de-rating risk even as the long-term theme holds.
The barbell pays off. AI compute demand is real, but the value accrues to power, cooling, siting and specialized cloud — the Long Book — while the crowded, richly-valued chip names in the Short Book de-rate as capex growth slows and compute gets more efficient. The ~$8.46bn of puts also hedges the longs if the whole AI complex sells off together.
The chip shorts are the pain trade. If hyperscaler capex keeps compounding, NVDA / AVGO / AMD and the semis keep re-rating and the put book bleeds premium. Meanwhile several longs are unprofitable, power-hungry, crypto-adjacent infrastructure names that can fall hardest in a risk-off — so both legs can lose at once.
Want the full positioning — long vs bearish vs private vs proxy vs ETF? Open the .
Curated research / editorial — a framework decoder, not a recommendation or a live market signal. Exposure types (beneficiary, bearish, private, proxy, ETF) are research classifications, not advice. Not investment advice.