Storage capacity & data movement have become as important as data processing in AI infrastructure.
Kioxia and SanDisk's new 2Tb QLC flash uses a six-plane architecture and a 33% faster interface to keep pace with the demands of massive language models:
AI demand is tightening global memory chip supply as makers shift to high-margin AI hardware.
Apple is pushing for permission to buy chips from China’s CXMT, facing pushback over security and domestic industry concerns.
Read more: techstrongsemi.com/article/ai-dri…#AI#Semiconductors
Intel's Starfire processor is a radiation-hardened system-on-chip that lets satellites run AI inference directly in orbit, eliminating the need to shuttle data back to Earth.
The chip also acts as an early showcase for Intel's 18A manufacturing process:
SK hynix is spending $38 billion on two new fabs in South Korea to expand HBM, DRAM, & NAND production.
The capacity arrives in 2028 or 2029, but the company projects 19% annual memory demand growth through 2030, keeping supplies tight and prices high.