· 供应链分析
HBM（高带宽内存）在2025年是一个关键瓶颈，我们从$MU和SK海力士身上看到了这一点。 正在研究未来超大规模云计算厂商ASIC（专用集成电路）部署所使用的组件，整个行业正在向光子学（Photonics）转型。 如果我们深入该供应链的底层，用于INP（绝缘体上铟磷）基板的供应商只有两家公司。想到这一点真是荒谬。 对于像$AXTI这样的公司来说，这可能是一个巨大的市场机会，但仅从国家安全风险的角度来看，这是一个需要被审视的疯狂瓶颈。虽然有像$COHR这样的公司，但远未达到AI基础设施建设所需的规模。
HBM was a critical bottleneck in 2025 and we've seen that with $MU and SK Hynix. Just researching components used for future hyperscaler ASIC deployments, the whole industry is moving to photonics. If we go to the bottom of that supply chain it's just two companies for INP substrates. Which is absurd to think about. It could be a huge market opportunity for investors for stuff like $AXTI, but just in terms of national security risk, it's just a crazy bottleneck that needs to be looked at. There's companies like $COHR, but nowhere near to scale required for AI buildout.