I don't think IBM's collapse is just a company-specific issue. To me, management's explanation is actually one of the most important takeaways from this earnings season. If customers are delaying software and mainframe spending because they are prioritizing AI servers, storage and memory, it suggests AI budgets are increasingly flowing toward the infrastructure layer first. Companies supplying GPUs, HBM memory, networking and storage could continue to benefit before enterprise software fully catches up.

That also helps explain why semiconductor and memory stocks reacted so strongly. SK Hynix's surge, together with gains in Micron, SanDisk and Nvidia, reinforces my view that the AI infrastructure buildout is still in its early stages. As long as hyperscalers continue expanding AI capacity, demand for high-performance memory and storage should remain robust, even if software spending becomes more selective in the near term.

For my own portfolio, I remain constructive on AI hardware. I continue to focus on companies exposed to memory, storage and semiconductor infrastructure because that's where I see the strongest earnings momentum today. Software will likely have its next leg higher once this infrastructure wave matures, but for now I believe the market is rewarding the businesses that are building the foundation of the AI ecosystem.