The world's memory chipmakers are scrambling to build more capacity as demand for artificial intelligence hardware keeps climbing. The Korea Herald reports that memory giants are racing to expand AI chip capacity, and the money at stake is enormous: Memeburn reports Samsung is eyeing an 18-fold profit jump as AI memory demand explodes.
Samsung has already moved from planning to production. According to Firstpost, citing details on the chip, Samsung has begun mass production of its PM1763 solid-state drive built for Nvidia's upcoming Vera Rubin platform, a drive the company says delivers 1.8 times better power efficiency. That makes Samsung a direct supplier into Nvidia's next generation of AI systems.
The demand picture, by most accounts, is not cooling. Executives quoted by CNBC describe AI demand as "almost unlimited," even as enterprises get more cost-conscious in what the outlet calls "valuemaxxing." Moomoo similarly reports the AI infrastructure boom shows no signs of slowing.
Analysts are split on the risks. Morgan Stanley, according to Yahoo Finance, warns of "chipflation" as hyperscalers pour money into compute capacity, driving prices up. Nomura, also via Yahoo Finance, argues that worries about AI memory demand are overdone because supply constraints are likely to persist.
The common thread is that memory and storage, not just Nvidia's marquee processors, have become a bottleneck and a battleground in the AI buildout. The suppliers that can add capacity fastest stand to capture outsized profits.
Why it matters: the cost and availability of memory chips helps set the price of AI itself, shaping how quickly companies can build and afford the systems powering the current AI wave.