Principle 02 of 06

Follow the capex

Capital expenditure is a company putting its money where its mouth is.

The idea

Executives can say anything on an earnings call. What they spend is harder to fake. Capital expenditure - capex - is the money companies commit to physical things: buildings, servers, chips, power connections. It’s a forward-looking signal of where they believe the future is.

Since 2023, the world’s largest technology companies have lifted their capex to levels never seen before, almost all of it directed at AI. One company’s spending is another company’s revenue - so following that money shows you who gets paid next.

Why it matters

Capex announcements are public, quantified and often guided a year or more ahead. They gave me an early, reliable map of where demand would flow - and let me position before the revenue showed up in suppliers’ results.

How I applied it

Across the four-year journey

  1. 2023

    Tracked hyperscaler capex guidance each quarter as it began to climb, confirming the compute thesis.

  2. 2024

    Saw spending shift from chips to buildings, power and cooling, and rotated part of my chip gains into data-centre suppliers.

  3. 2025

    Noticed power connections becoming the binding constraint in capex plans - the trigger for my energy sleeve.

  4. 2026

    Watching capex move toward inference and edge deployments, and adjusting exposure accordingly.

In practice

My working rules

  • ✓Read the capex guidance in big-tech earnings every quarter - it’s the best free research there is.
  • ✓Ask: who receives this money? Then research those companies.
  • ✓Watch the direction of change, not just the level. Acceleration matters most.
  • ✓Be alert when capex growth starts to slow - that’s when supplier stocks are most vulnerable.

Watch out

The trap: capex cycles turn

History is full of over-investment booms - railways, telecoms fibre, shale. When spending is cut, suppliers suffer first. Following the capex means following it down as well as up.

Key takeaway

“Don’t listen to what companies say about AI. Watch what they spend on it.”