All research
Equities & sectorsSeilern Investment Management

Buy Now, Expense Later

United StatesArtificial intelligenceAI infrastructure

Our summary

The S&P 500’s strong earnings growth is less durable and less reliable than it appears because AI capital spending is boosting suppliers’ profits before hyperscalers fully recognise the associated depreciation.

  • S&P 500 earnings are expected to grow 28% in 2026, versus a long-term rate of 10%, while the index trades at about 20 times forward earnings.
  • The five largest hyperscalers are expected to spend around $800 billion on capex in 2026, up 96% from 2025, rising to $1.1 trillion in 2027 and $1.2 trillion in 2028.
  • Hyperscalers are expected to record roughly $201 billion of depreciation and amortisation in 2026, far below current capex, while semiconductor, hardware, industrial and utility companies account for about half of this year’s consensus S&P 500 earnings growth.
  • Goldman Sachs estimates that AI capex will add 11 percentage points to S&P 500 earnings growth in 2027, but depreciation will offset nearly half of that benefit, reducing the net contribution to about 7 points.
  • By 2028, Goldman expects rising depreciation to offset the earnings uplift from continued AI investment entirely, while the revenue opportunity remains uncertain: Anthropic has discussed a potential addressable market above $30 trillion, against an expected $100 billion annualised revenue run-rate by year-end 2026.

Takeaway: Investors should not treat the current 28% earnings-growth rate as a stable basis for a 20-times forward valuation; the outlook depends on whether AI revenues grow fast enough to absorb rising depreciation, and Seilern is assessing this uncertainty through its investments in Alphabet and Microsoft.

Published by Seilern Investment Management. Summary by HF Best Ideas.

More from Seilern Investment Management

Latest research

Share this summary