AI Not Only Sells "Stories" – But Also Sells Memory. And Memory Is Where the Real Money Is.

In the current AI wave, many segments are mentioned: GPU, large model software, AI applications… But if you look at the financial reports, the part that is clearly printing money is memory – especially DRAM and HBM.
The difference lies in:
- GPUs are processing centers.
- But AI cannot run without sufficiently large and fast memory.
And as models grow larger, memory demand increases exponentially.
🧠 Why is memory the "money printing machine" of the AI cycle?
1. Capacity grows rapidly
Each new AI generation requires more VRAM and HBM. Training larger models → memory expands.
2. HBM supply is scarce
High-end DRAM manufacturers are few. When demand exceeds supply, prices rise → profit margins expand.
3. Reflected directly in actual profits
Unlike many AI companies still telling future stories, memory has directly shown up in revenue and profit.
Big beneficiaries
- Micron Technology
- SK Hynix
These two giants are directly benefiting from the demand for DRAM and HBM for AI. Notably:
- The market is starting to revalue them, no longer as "cyclical stocks" but increasingly as growth stocks linked to AI.
If this continues, their P/E ratios could fully expand like other tech growth companies.
👉 Changing mindset: From cyclical stocks → growth stocks
If DRAM and HBM truly become "essential infrastructure" for AI, then market revaluation is reasonable.
{alpha}(560x8b6acf6041a81567f012ff6a4c6d96d5818d74bf)