Summary
In this Bloomberg Technology segment, an equity analyst makes the case that AI is fundamentally altering the traditionally boom-and-bust memory chip cycle, turning it into something more structurally sustained. The core argument: as inference workloads and agentic AI drive token demand higher, memory bandwidth and capacity become primary performance bottlenecks — prompting cloud service providers to keep buying more memory even as prices rise, because it remains the most cost-efficient lever for system-level throughput.
On the supply side, the analyst highlights three factors keeping the market tight regardless of how aggressively Micron, SK Hynix, and Samsung try to add capacity: rising manufacturing complexity, increasing capital intensity for new fabs, and the difficulty of technology node migration. First-hand observations from a recent Asia trip reinforce this view, with supply chain companies reporting extended order visibility and growing discussions around long-term agreements.
The conversation also touches on the analyst’s actively managed ETF, TEKY, which holds both large CapEx spenders building AI infrastructure and the hardware supply chain companies receiving that capital. Looking further out, the analyst sees physical AI — autonomous transportation and humanoid robotics — as the next multi-trillion dollar demand wave sitting above the current infrastructure build-out. SK Hynix receives particular attention as a company the analyst flagged early as critical to the broader AI infrastructure story.
📺 Source: Bloomberg Technology · Published June 24, 2026
🏷️ Format: News Analysis
