Q2 Semiconductor Pulse: AI Demand Is Rewriting the Rules for Chipmakers
Are AI compute constraints capping semiconductor demand?
The AI semiconductor ecosystem, led by Nvidia, should sell everything it can make in 2026 and 2027. Data centers remain compute-constrained, and hyperscaler capex shows no sign of slowing. Are stocks keeping up?
US semiconductor stocks have been unusually volatile this year, surging 55% in the second quarter before holding steady in the third. They now make up roughly 16% of the US market, a historically high share. Despite the run, valuations aren’t stretched: The median chipmaker was significantly undervalued as of Sept. 1.
Investors, analysts and business leaders tracking the AI buildout need to know where genuine capacity constraints end and valuation noise begins. Get the full picture on AI compute constraints, hyperscaler capex and memory pricing cycles.
