P/E ratios for memory stocks have fallen to single digits, and many people think this is the bottom.
I’m bullish on memory stocks, but I’m not so sure I’d say that.
$MU closed at 594.51 on Friday, down from above 1,250 just six months ago—more than a 50% drop. Sandisk fell 16% on the first day after its 8-for-1 split. Valuations certainly look cheap, but memory is a cyclical industry, and cyclical stocks often have their lowest P/E ratios when earnings are at their peak.
That’s exactly what happened in 2018. Micron posted record earnings that year, with a P/E of only three or four, but its stock fell from around $64 in May to around $28 in December—losing more than half its value. Later, when earnings actually declined, the P/E rose, and only then did the stock hit bottom.
There’s one thing that’s different this time: much of HBM is locked in under long-term contracts, and AI demand won’t disappear just because the Nasdaq falls 10% in a week. But on the NAND side, research firms are already saying prices could fall 20% in 2027, and spot prices have already weakened.
Being bullish on memory and buying in right now are two different things. I’m still bullish on the industry over the long term, but I’d rather wait for prices to show me that earnings expectations have been cut enough. I won’t buy just because the P/E is low.
There are similar indicator traps in crypto. BTC falling below miners’ costs or below Strategy’s average purchase price may sound like a bottom signal, but those numbers can’t stop prices from falling further—they’ll only make things harder for some people.
#存储芯片 #美股 #比特币 $BTC
I’m bullish on memory stocks, but I’m not so sure I’d say that.
$MU closed at 594.51 on Friday, down from above 1,250 just six months ago—more than a 50% drop. Sandisk fell 16% on the first day after its 8-for-1 split. Valuations certainly look cheap, but memory is a cyclical industry, and cyclical stocks often have their lowest P/E ratios when earnings are at their peak.
That’s exactly what happened in 2018. Micron posted record earnings that year, with a P/E of only three or four, but its stock fell from around $64 in May to around $28 in December—losing more than half its value. Later, when earnings actually declined, the P/E rose, and only then did the stock hit bottom.
There’s one thing that’s different this time: much of HBM is locked in under long-term contracts, and AI demand won’t disappear just because the Nasdaq falls 10% in a week. But on the NAND side, research firms are already saying prices could fall 20% in 2027, and spot prices have already weakened.
Being bullish on memory and buying in right now are two different things. I’m still bullish on the industry over the long term, but I’d rather wait for prices to show me that earnings expectations have been cut enough. I won’t buy just because the P/E is low.
There are similar indicator traps in crypto. BTC falling below miners’ costs or below Strategy’s average purchase price may sound like a bottom signal, but those numbers can’t stop prices from falling further—they’ll only make things harder for some people.
#存储芯片 #美股 #比特币 $BTC