AI storage: this line is crucial—but what’s really worth watching now is where the money is spreading.
Over the past two days, I’ve re-sorted the storage industry chain, and I’ve noticed a change: this current AI storage cycle may be more than just an HBM story.
Looking further down the supply chain, it’s quite clear: HBM → DRAM → NAND → enterprise SSDs.
So right now, I’m focusing on two directions: SK hynix and SanDisk.
What is hynix looking at? Certainty.
As AI compute continues to expand, HBM remains the first segment to benefit, and hynix’s advantage in HBM has already been repeatedly validated by the market. On top of that, South Korea’s chip export data has stayed strong, indicating that AI semiconductor demand is gradually moving from capital-market expectations into real industry data.
But here’s the issue: hynix is strong enough, and the market has known it for a long time. So it’s more like an opportunity defined by certainty after earnings are delivered.
Meanwhile, I’ve been more interested lately in researching SanDisk. The reason is simple: AI models are getting bigger, training data is growing, and inference scales are increasing. GPUs handle compute power, and HBM handles high-speed data exchange—but ultimately, all that massive data still needs to be stored.
That elevates the importance of enterprise SSDs and NAND.
In the past, the market mainly looked at NAND through the cycles of smartphones, PCs, and consumer electronics. But in the AI era, data centers are becoming a new variable in demand. If enterprise SSD demand keeps growing, the impact won’t just be on unit sales—it could further transmit to NAND prices, inventory levels, capacity utilization, and manufacturers’ profit margins.
So I’m paying attention to SanDisk not because it is guaranteed to rise today, but because I want to see whether the market has started to reprice “NAND + AI storage.”
If capital continues to spread from HBM toward NAND and SSDs, then the logic of this cycle may be changing. Hynix offers certainty, while SanDisk is about the expectation gap. What’s truly worth watching is the moment when certainty starts to diffuse into the expectation gap. @K线猎人-
Over the past two days, I’ve re-sorted the storage industry chain, and I’ve noticed a change: this current AI storage cycle may be more than just an HBM story.
Looking further down the supply chain, it’s quite clear: HBM → DRAM → NAND → enterprise SSDs.
So right now, I’m focusing on two directions: SK hynix and SanDisk.
What is hynix looking at? Certainty.
As AI compute continues to expand, HBM remains the first segment to benefit, and hynix’s advantage in HBM has already been repeatedly validated by the market. On top of that, South Korea’s chip export data has stayed strong, indicating that AI semiconductor demand is gradually moving from capital-market expectations into real industry data.
But here’s the issue: hynix is strong enough, and the market has known it for a long time. So it’s more like an opportunity defined by certainty after earnings are delivered.
Meanwhile, I’ve been more interested lately in researching SanDisk. The reason is simple: AI models are getting bigger, training data is growing, and inference scales are increasing. GPUs handle compute power, and HBM handles high-speed data exchange—but ultimately, all that massive data still needs to be stored.
That elevates the importance of enterprise SSDs and NAND.
In the past, the market mainly looked at NAND through the cycles of smartphones, PCs, and consumer electronics. But in the AI era, data centers are becoming a new variable in demand. If enterprise SSD demand keeps growing, the impact won’t just be on unit sales—it could further transmit to NAND prices, inventory levels, capacity utilization, and manufacturers’ profit margins.
So I’m paying attention to SanDisk not because it is guaranteed to rise today, but because I want to see whether the market has started to reprice “NAND + AI storage.”
If capital continues to spread from HBM toward NAND and SSDs, then the logic of this cycle may be changing. Hynix offers certainty, while SanDisk is about the expectation gap. What’s truly worth watching is the moment when certainty starts to diffuse into the expectation gap. @K线猎人-