There's a chance the market is underestimating the demand side for HBM. The idea that more advanced AI reasoning models will always require less compute might be off.
More reasoning-heavy models can actually lead to higher requirements for tokens, inference steps, memory bandwidth, and HBM capacity. While the cost per token might look lower, the total compute cost per completed task could increase.
This reminds me of how AI optimization is getting more application-specific, similar to ASICs. Improving one area often creates new demands in another. The race isn't just about software; it's also about the underlying chips, memory, and infrastructure needed to run these models.
That points to ongoing demand for AI GPUs, HBM memory, data centers, and networking. Names like $NVIDIA(NVDA)$ , $Micron Technology(MU)$ , $SK hynix(SKHY)$ , and DRAM.X are positioned in this infrastructure layer.
Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.
