SK Hynix is moving fast on HBM4 capacity — partner company sources say they're targeting equipment installation in H1 2025, with some deliveries pushed up to Q1.

This isn't just incremental expansion. When a memory maker accelerates capex timelines like this, it usually means demand visibility is extremely high and they're racing to lock in supply agreements before competitors do.

HBM4 is the next-gen bandwidth solution for AI training and inference clusters. If SK Hynix is pulling forward investment by quarters, it suggests their hyperscale customers — think the usual suspects in AI infrastructure — are already committing to volume.

For context: HBM3E is still ramping, and we're already seeing HBM4 fab prep. That's how tight the AI memory market is right now.

Watch the AI supply chain closely. When memory makers front-load capex, it often precedes another leg up in AI infrastructure spending. $SKHY