Memory prices climb 500% in 12 months, up to 10x the lowest ever tracked prices

Commenters saw the memory surge less as a normal component cycle than as AI infrastructure demand spilling into every machine people own. The practical anxiety was immediate: RAM, SSDs, hard drives, GPUs, used servers, and replacement parts are all becoming expensive enough that people are postponing upgrades, stretching hardware lifetimes, and worrying that a single failed module could force an unaffordable purchase. Several readers pointed to resellers stripping RAM from used servers and selling it separately as a sign that even secondhand infrastructure is being hollowed out by the price spike. The AI-specific anger centered on frontier labs and hyperscalers locking up supply. Readers questioned how loss-making frontier-model businesses can justify vastly costlier memory than a few years ago, and argued that enormous advance purchases protect incumbents while denying smaller entrants the hardware needed to compete. A minority framing cast higher local-compute costs as beneficial because they steer customers toward cloud AI, but that was met with disbelief: centralized access may please major providers and their investors, yet it turns AI from something people can run and build independently into another metered utility. Some still hope an AI downturn eventually creates cheap surplus hardware, though others noted that data-center memory may not translate cleanly into consumer bargains.