1016 SanDisk (SNDK): Recent Strategy
It’s time for everyone’s absolute favorite moment: the three storage stocks.
Let’s start with SanDisk.
This downturn isn’t over yet; the short sellers just got out early.
Long positions still have to weather more downward pressure.
I’ve mentioned this logic before: there are three fundamental elements to training large and small AI models:
1. Compute: continuously iterating and upgrading AI models through more training steps.
2. Power: demand for compute keeps rising, and a stable power supply is essential. A typical example is the one we often trade, $VST .
3. Storage: Each round of model iteration leaves behind massive amounts of intermediate data. It’s like a child learning to walk—every step leaves a footprint. A model may run from its initial iteration all the way to 24801 steps or even more. Power is a fundamental, essential energy need, so its demand is more predictable.
Although storage is an indispensable part of AI infrastructure,
the sector comes with a huge valuation bubble.
Many people wonder where the storage bubble comes from:
The core issue is a mismatch between supply and demand, combined with expectations being priced in too far ahead. When the AI boom takes off, the market extrapolates growth in a straight line, assuming AI storage demand will keep surging exponentially forever, and prices decades of future growth into today’s stock prices all at once. But the storage industry has one hard constraint: expanding wafer-fab capacity takes a very long time. When the market is booming, manufacturers see prices rising and pour money into building factories. But once a factory is built, its capacity keeps coming online and can’t be shut down quickly. When that new capacity arrives, if growth in AI compute capital spending slows even a little, supply can suddenly outstrip demand and prices can plunge. Put simply: in a good market, investors value storage stocks as high-growth tech companies; but once supply and demand reverse, they’re fundamentally cyclical businesses, and earnings can fall off a cliff. Expectations are most inflated when the bubble is biggest.
After all that, SNDK has already fallen quite a bit, but it still needs one final push down.
The level I’m watching is daily-chart support at 1197 below. Let’s not quibble over a few points.
Just place a limit order a little above 1200, with a stop-loss at 1165.
$SNDK : same old approach, medium to long term.
#SNDK #老郑策略
It’s time for everyone’s absolute favorite moment: the three storage stocks.
Let’s start with SanDisk.
This downturn isn’t over yet; the short sellers just got out early.
Long positions still have to weather more downward pressure.
I’ve mentioned this logic before: there are three fundamental elements to training large and small AI models:
1. Compute: continuously iterating and upgrading AI models through more training steps.
2. Power: demand for compute keeps rising, and a stable power supply is essential. A typical example is the one we often trade, $VST .
3. Storage: Each round of model iteration leaves behind massive amounts of intermediate data. It’s like a child learning to walk—every step leaves a footprint. A model may run from its initial iteration all the way to 24801 steps or even more. Power is a fundamental, essential energy need, so its demand is more predictable.
Although storage is an indispensable part of AI infrastructure,
the sector comes with a huge valuation bubble.
Many people wonder where the storage bubble comes from:
The core issue is a mismatch between supply and demand, combined with expectations being priced in too far ahead. When the AI boom takes off, the market extrapolates growth in a straight line, assuming AI storage demand will keep surging exponentially forever, and prices decades of future growth into today’s stock prices all at once. But the storage industry has one hard constraint: expanding wafer-fab capacity takes a very long time. When the market is booming, manufacturers see prices rising and pour money into building factories. But once a factory is built, its capacity keeps coming online and can’t be shut down quickly. When that new capacity arrives, if growth in AI compute capital spending slows even a little, supply can suddenly outstrip demand and prices can plunge. Put simply: in a good market, investors value storage stocks as high-growth tech companies; but once supply and demand reverse, they’re fundamentally cyclical businesses, and earnings can fall off a cliff. Expectations are most inflated when the bubble is biggest.
After all that, SNDK has already fallen quite a bit, but it still needs one final push down.
The level I’m watching is daily-chart support at 1197 below. Let’s not quibble over a few points.
Just place a limit order a little above 1200, with a stop-loss at 1165.
$SNDK : same old approach, medium to long term.
#SNDK #老郑策略
