Interview, Other
Will Hyperscalers Justify AI Spend?
- Market sustainability for current spending levels depends on demonstrating Return on Investment (ROI) within the current or next quarter to prevent multiple compression and an unsustainable equilibrium; conversely, the absence of ROI risks a cascading derating of both spenders and receivers.
- Corporate earnings growth for U.S. companies is projected at approximately 24% for the year, with Q2 expectations requiring 23% to 24% EPS growth, creating a high bar where earnings may "travel and arrive" rather than exceed expectations; financials are highlighted as a key sector to monitor.
- Earnings revisions have increased by 3% in the current quarter, though a structural risk exists where U.S. GDP growth becomes overly concentrated in a handful of AI-related stocks and sectors driven by hyperscaler CapEx.
- Primary risks for the back half of the year include a disruption to the AI spending narrative, a mismatch between spending and market expectations on CapEx (particularly with leverage), and market vulnerability stemming from near 50% year-to-date growth in levered ETF assets under management, half of which are in semi-hardware.
- Memory is expected to remain cheap with rising investment rates as the industry cycles up, while other sectors show divergent trends: autos trade at 52-week lows, whereas aerospace and defense are viewed as secular growth orthogonal to AI.
- European markets present specific opportunities and threats: southern European indices (Spain and Italy) have outperformed the U.S. year-to-date, Greek banks offer attractive yields within France, and the region is considered structurally investable; however, many European sectors face pressure from China's aggressive export and overcapacity narratives.
- Potential trades include buying the index while selling single names to capitalize on low index volatility against high implied single-stock volatility, with the thesis that European markets may perform well if the AI trade underperforms.
- Long-term rationalization in AI is anticipated as smart actors find alternative pathways to solve key bottlenecks, alongside ongoing debate regarding the required spend allocation between frontier model levels and other infrastructure areas.