#771: Why AI Demand Won’t Collapse with Mel Mattison

Mel Mattison sees AI demand as durable, not speculative. He argues that the bear case—most prominently made by commentator Ed Zitron—confuses the fate of specific AI companies with the fate of the hardware and compute infrastructure that underlies the entire trend. Even if OpenAI or Anthropic fail as businesses, Mattison says, the demand for memory, chips, and data center capacity will keep growing because AI itself is a transformational technology on the scale of the internet or the railroads. The conversation covers why memory stocks are cheap, why the hyperscalers’ debt is not a risk, and how the fiscal backdrop and new investment vehicles like Trump accounts create a persistent bid under equities.

The Bull Case for Memory and Compute

Mattison’s core argument is that AI requires immense amounts of memory and compute, and that demand is growing far faster than supply. He points to memory manufacturers like SK Hynix and Micron, which trade at four to seven times forward earnings despite capacity growing at 20–30% per year while demand grows at 200–300% per year. He sees this mismatch as unsustainable and believes the stocks are dramatically undervalued. “If you want to say there’s no black swan impossible downside, then you can’t invest in anything,” he says, dismissing the idea that a memory breakthrough could render these companies obsolete overnight.