Samsung Foundry just landed $AMD orders for HBM base dies — production headed to their second Taylor, Texas fab.
This is more than a supplier win. It's about diversification in the AI memory supply chain. TSMC has dominated advanced packaging, but Samsung's Texas expansion gives AMD another credible option for scaling HBM capacity.
Watch how this plays into AMD's MI300 ramp and datacenter GPU roadmap. Memory bandwidth is the bottleneck in AI training — whoever controls HBM production controls a key piece of the infrastructure.
Samsung's U.S. fab buildout also aligns with CHIPS Act incentives and domestic supply chain priorities. Not just about geopolitics — it's about redundancy and lead times in a market where every quarter of capacity matters.
If you're tracking the AI supply chain, this is a signal that the memory layer is getting more competitive. That usually means better pricing power for hyperscalers and tighter margins for the foundries — but also more optionality for chip designers like $AMD.
This is more than a supplier win. It's about diversification in the AI memory supply chain. TSMC has dominated advanced packaging, but Samsung's Texas expansion gives AMD another credible option for scaling HBM capacity.
Watch how this plays into AMD's MI300 ramp and datacenter GPU roadmap. Memory bandwidth is the bottleneck in AI training — whoever controls HBM production controls a key piece of the infrastructure.
Samsung's U.S. fab buildout also aligns with CHIPS Act incentives and domestic supply chain priorities. Not just about geopolitics — it's about redundancy and lead times in a market where every quarter of capacity matters.
If you're tracking the AI supply chain, this is a signal that the memory layer is getting more competitive. That usually means better pricing power for hyperscalers and tighter margins for the foundries — but also more optionality for chip designers like $AMD.
