Memory chips emerge as AI's bottleneck, with 15-to-1 demand-supply gap

The semiconductor industry has become central to AI infrastructure deployment, but not all chips face equal scarcity. While GPU (graphics processing unit) capacity has dominated headlines, analyst Dan Ives highlights a deeper constraint: the memory chips required to support AI systems are in far shorter supply relative to demand. The 15-to-1 demand-to-supply ratio indicates that memory chip production cannot match the pace at which AI data centers are being built out. This imbalance is not a temporary constraint; Ives expects it to persist for years, meaning memory chip makers will remain a structural bottleneck in the AI supply chain. For investors and business leaders, this signals where capital allocation is likely to concentrate—toward memory chip manufacturers and the infrastructure companies that depend on securing steady supply. The stocks Ives is tracking reflect this shift in where the money is moving within semiconductors.