SK Hynix shipped samples of its new 12-layer HBM4E chips with 16Gbps per pin speed and 20% better power efficiency to Nvidia for AI systems. Coherent inked a letter of intent for up to $50m in government funding to expand an optical networking plant tied to Nvidia’s AI datacenter infrastructure.
SK Hynix has shipped samples of its 12-layer HBM4E high-bandwidth memory chips to major AI customers including Nvidia, featuring speeds of 16 gigabits per second per pin and over 20% improved power efficiency compared with prior models.
Coherent signed a letter of intent to receive up to $50 million in government funding to expand a manufacturing facility producing optical networking components that support Nvidia’s AI datacenter infrastructure.
The introduction of HBM4E chips and expanded optical networking capacity should bolster Nvidia’s ability to deliver next-generation AI GPUs by easing memory bandwidth constraints and increasing data transport throughput, though ramp timelines and integration will determine short-term production gains.
