How AI Bought Up the Hard Drives
Training and running AI needs more than GPUs—it needs massive storage: training datasets, model weights, logs, caches. Data centers' appetite for high-capacity drives is staggering. When global AI infrastructure surges all at once, this demand quickly swallows drive makers' capacity, and Western Digital's "sold out for the year" is a direct signal. This is yet another chain-reaction shortage triggered by the AI investment frenzy, after GPUs, memory, and power—when AI squeezes some link upstream, price and supply propagate all the way down.
The Ripples of the Shortage
The impact of drives selling out spreads layer by layer: rising storage costs propagate to every business that depends on high-capacity storage—cloud services, backup solutions, NAS, and even the budget of ordinary consumers building their own PCs. This again confirms a repeatedly validated judgment: AI's true cost is far more than "buying compute"—it's systematically squeezing the entire hardware supply chain, and chips, memory, drives, and power are none of them spared, while it's ultimately all the downstream users who foot the bill. For businesses, the pragmatic reminder is: when planning capacity and budget, don't assume storage will always be cheap and available on demand—AI-driven shortages may lengthen lead times and raise unit prices. For the industry, this is another concrete example of AI's "externalities": one field's frenzy makes many seemingly unrelated people share the cost, in the form of price hikes and shortages.
via: Hacker News