Congratulations, you survived the bursting of the AI bubble

About US$5 trillion of market value has been erased as the Magnificent Seven and broader tech names trade double-digits off 52-week highs; the S&P 500 multiple has fallen from >24 to roughly 20. U.S. software stocks are down ~30%, Constellation Software saw a max drawdown of 56%, Microsoft and Meta are down ~32% and ~27% respectively, and Nvidia’s valuation is roughly half of its post-ChatGPT peak. Top five hyperscalers are committing ~US$700 billion in AI capex this year, but rising inflation and Strait of Hormuz-related energy risks could push rates higher and further pressure tech valuations.

Analysis

The market is no longer simply marking down glamour names — it is re-allocating a multi-year capex stream into discrete winners and losers along the hardware + services supply chain. Providers of commoditized, low-margin software and ad-dependent platforms are most exposed to multiple compression because rising rates shorten valuation duration and because enterprises can substitute cheaper, in‑house models for routine workflows; conversely, firms that sell critical physical infrastructure (power conversion, racks, thermal management, contract manufacturing) should see more predictable orderbooks in the next 6–18 months even if software multiples fall further.

A key second-order dynamic: front-loaded hyperscaler spending will create a pronounced inventory and capacity cycle across semicap, contract manufacturers and data‑centre components. That implies a 12–24 month window where OEMs and systems integrators (with flexible manufacturing and low fixed cost) capture outsized margins, followed by a >24 month risk of oversupply once hyperscalers shift from build to utilization/efficiency. Meanwhile, companies with proprietary vertical data or gross-margin-rich recurring revenue will gain pricing leverage as customers prioritize predictable OPEX over speculative vendor relationships.

Risks and reversals are clear and fast. A prolonged geopolitical inflation shock or a meaningful slowdown in enterprise AI projects (integration costs, privacy/regulation, model inference expense) can trigger another leg down in 3–6 months; conversely, a visible monetization inflection (meaningful incremental ARR recognized by software vendors or stronger-than-expected ad/resilience at large platforms) or a rapid oil/interest-rate normalization could snap valuations higher. Given the asymmetry, express views with defined-risk option structures and pairs to capture convexity while limiting idiosyncratic headline risk.

Key Decisions for Investors

• Long CSU.TO (12‑month horizon): buy shares on any further weakness or initiate via a long‑stock + protective 12‑month 25–30% OTM put. R/R: target +35–45% if EV/EBITDA re-rates toward historical mid-cycle; downside ~‑30% if niche end-market erosion accelerates. Add on a second 10% pullback; trim into any earnings/multiple recovery.
• Long CLS (3–6 month tactical): buy shares or 3‑month ATM calls after next market open; thesis is earnings-driven mean reversion at the supplier margins. R/R: target +20–30%; stop‑loss at ‑12% on mounting inventory signs. Consider pairing with a small short in a large hyperscaler ETF to neutralize beta risk.
• Convex long on semis: buy a 9‑month NVDA call spread (buy ATM, sell ~1.4× strike) sized to <2% book exposure and funded by selling 9‑12 month calls on META (or buying a 12‑month META put spread). R/R: captures upside if AI demand persists while funding cost reduces net premium; downside limited to premium paid if AI monetization disappoints.
• Short/hedge META (6–12 months): initiate a put spread (buy 30–40% OTM put, sell deeper OTM put) sized to hedge tech exposure around earnings and macro events. R/R: profit if ad elasticity and margins prove worse-than-modeled; main risk is an operational cost cut/monetization beat that restores confidence.