newsfilter.io
Lecture, Other

ITPM Flash Ep113 Standing on the Edge

  • S&P 500 valuation metrics are at historical extremes, with price-to-sales at 3.66, price-to-book nearly six times the historical average, and forward P/E under 24, though AI-driven infrastructure spending suggests these figures may not fully reflect overvaluation.
  • Capital expenditure by five major companies has risen from $160 billion to $700 billion, with a current run rate of $737 billion, projected to reach approximately $1 trillion next year and remain elevated for at least 18 months.
  • Profit margins for data center builders are estimated at 40%, facilitating a profit transfer of over $300 billion currently to 50–80 supply chain participants, increasing to $400 billion next year.
  • MAG7 earnings growth for Q1 2026 is projected to surge 60% year-over-year; excluding MAG7, S&P 500 earnings growth is estimated between 10% and 13%, while 70% of S&P returns since ChatGPT's launch originated from just 42 companies.
  • The market currently exhibits high certainty regarding concentrated structures and earnings trajectories, with hyperscalers likely to maintain construction plans even if bond yields rise by 25 to 50 basis points.
  • Risk events may include a small correction if AI integration improvements fall short, or a significant crash when unit economics become a primary concern regarding token costs and returns.
  • Specific sell signals are identified when hyperscalers match capital expenditure to demand rather than exceeding it, or when they downgrade planned spending two years out, driven by mean reversion rather than asset price levels alone.
  • Investors are advised to maintain an overweight position in the capital expenditure trade for the next 18 months provided compute demand materially exceeds supply, utilizing active management and holding $3,000 to $5,000 in downside protection for every $10,000 of long exposure.
  • Market reaction to higher interest rates is anticipated only upon the emergence of job losses and defaults among lower-income consumers, which are not currently observed.