AI Breaks the Memory Chip Cycle—and That Changes Everything

Celine Woo, portfolio manager at Lazard Asset Management, spotted something the headlines miss: while OpenAI, Meta, and Google are burning billions on AI infrastructure like there's no tomorrow, the real gold isn't in the companies spending the money—it's in the ones selling the goods.

Translate that: forget hunting for opportunity in the mega-corporation launching the next AI model. Look instead at the supply chain—the outfits manufacturing the parts needed to build it. It's like the Gold Rush: who actually got rich? Not the miners. The people selling pickaxes.

The memory chip market used to follow a rhythm: demand spikes, then crashes, then back up again. Predictable, almost boring. But AI broke that pattern. Demand isn't cyclical anymore—it's a nearly constant, obsessive growth trajectory because AI's appetite for memory is bottomless. No seasonal dips, no annual reset. It's hunger that never stops.

This stability is, paradoxically, excellent news for hardware makers. They used to ride the roller coaster. Now they know demand won't crater (unless the whole AI bubble pops, but who's betting on that?). They can plan, invest, and hire without the dread of sudden collapse.

Woo makes clear that Lazard isn't chasing the famous names you read about. They're hunting the companies supplying the infrastructure: memory chips, servers, connectivity—all the backstage stuff. The invisible enablers of an invisible technological shift.

Bottom line: while everyone's shouting 'AI, AI, AI,' the smart money is watching who sells the screws and nails for building the cathedral.