AI data centers trigger memory supercycle
Kingston founders David Sun and John Tu add $14bn each, hottest asset is DRAM and packaging capacity not chatbot narratives
Images
A quiet corner of the tech supply chain is suddenly producing some of the loudest wealth effects of the AI boom: memory.
Business Insider reports that Kingston Technology cofounders David Sun (74) and John Tu (84) have each added roughly $14 billion to their fortunes since the start of January—about a 44% jump—putting them near $45 billion apiece on Bloomberg’s Billionaires Index. The pair, who split ownership 50/50, now outrank celebrity names of capital such as MacKenzie Scott and SoftBank’s Masayoshi Son in personal net worth.
This is less a feel-good immigrant success story (though it is that: Sun and Tu emigrated to the US in the 1970s, met in Los Angeles, and bonded over basketball) than about where scarcity pricing actually lives. AI model builders can talk about “intelligence” all they want; what constrains deployment is increasingly physical: DRAM, NAND, and—most painfully—high-bandwidth memory (HBM) stacked close to GPUs.
The industry’s bottleneck is not just wafer starts but advanced packaging capacity and the ability to qualify memory at scale for data-center-grade workloads. Hyperscalers are “clamoring for memory chips” to build out AI data centers, Business Insider notes, driving a “severe global shortage” and pushing prices higher. In that environment, the value migrates from the companies selling abstractions (“platforms,” “agents,” “models”) to the firms that can ship modules, drives, and validated configurations into procurement pipelines.
Kingston is not a household name because it doesn’t need to be. It sits in the unglamorous middle: memory modules and storage products that turn upstream silicon into deployable systems. When shortages hit, that middle layer becomes a toll booth. The market, with its usual talent for rediscovering the obvious late, is now repricing that reality.
The founders’ history also illustrates how capital cycles reward resilience more than narrative. Bloomberg’s profile, cited by Business Insider, recounts that Sun and Tu sold their first memory-device company (Camintonn) in the 1980s, lost their savings in the 1987 Black Monday crash, then rebuilt—founding Kingston, selling 80% to SoftBank for $1.5 billion in 1996, and buying it back for $450 million in 1999. Few business plans include “sell high, buy back low” as a repeatable strategy; fewer still execute it.
The broader “memory supercycle” is already visible in public markets. Business Insider highlights Micron, whose stock has more than quadrupled over the past year, valuing the company around $470 billion—an eye-catching number for a business that, in calmer times, is treated as a commodity supplier.
If this cycle persists, it will be because AI’s promised productivity gains still require old-fashioned inputs: electricity, data centers, and the kind of memory that doesn’t care about your pitch deck.