Don't buy stocks — buy theses. Each card below is a hypothesis about how the world is changing, the bets that win if it's right, the bets that lose if it's wrong, and what would change your mind. This is how professionals think.
The hypothesis
Microsoft, Google, Meta, and Amazon are in a competitive arms race for AI capability. None can afford to underinvest. Capex won't normalize until either: (a) a clear AI winner emerges and the others retrench, or (b) returns on AI capex visibly disappoint. Neither is happening in the next 18 months.
Why now
Q3 2026 earnings showed hyperscaler capex revised UP, not down. Nvidia data center revenue still accelerating. Meta said quiet part loud: 'risk of under-investing is greater than over-investing.'
Wins if true
Most exposed if wrong
What to watch for
Quarterly hyperscaler capex guidance (revisions UP = thesis intact)
Nvidia data center revenue growth rate (>30% YoY = healthy)
