Kioxia $285A.JT reported underwhelming earnings, putting a dent on the memory rally

Seems like some idiosyncratic poor execution by Kioxia as it’s the only memory company that missed this season; not sure how their ASPs even went down when memory prices are rising everyday.

Added to $MU @ $246

Basically this lol

Closed 50% of $SNDK at $263, added more $MU @ $242 and $INTC @ $36.8-$37

$NVDA @ $186 makes sense as a short hedge to the AI bubble bursting if you have duration.

Growing competitiveness from TPUs/accelerators + $MSFT CEO mentioning that they’re limiting GPU capex because they don’t want to run the risk of tech obsolescence due to Nvidia’s 1 year product cycle.

If PPW improvements slow down from one gen to the next, then hyperscalers are less inclined to spend on next gen GPUs as the legacy ones that are fully depreciated would work almost as fine with higher resulting “cash flows”.

Ugly day no two ways about it.

AI spend concerns were exacerbated by OpenAI’s comments about wanting govt’s backstop, CRWV delays dragging down the neoclouds & power providers, Satya comments as above, Kioxia’s print dragging down memory which was one of the last few green AI sectors, and Fed comments on high inflation and a potential push out of cuts. Maybe also some sell the news on the lifting of govt shutdown

Actually all things considered, I would have expected the markets to be down more. This reminds me of a few months back when MSFT reportedly cancelled some DC leases, which caused the AI stocks to chop around violently for awhile. Essentially the market has run out of bullish catalysts in the near-term and is now overly jittery on headlines.

In times like these important not to overtrade, stick to highest conviction ideas and touch grass. Let the markets resolve to its equilibrium before being aggressive again.

Pretty interesting. The Big Short is one of my favourite movies. Not agreeing with Michael Burry’s views (for now) but the parallels here for him are uncanny.

https://x.com/onechancefreedm/status/1988823273287635403?s=46

Closed all remaining $SNDK @ $252-$257, bought more $MU @ $242

$MU is my only memory position now, with a little bit of $HY9H

Seems like Kioxia’s poor print was because they signed a Long Term Agreement with AAPL right before NAND spot prices went ballistic, causing them to not benefit from rising memory prices.

While it might not have been public information in this case, it’s a good example of the importance of examining contract structures of B2B businesses.

Closed all remaining $CRCL short @ $81.5 for +33% gain on the entire position. Borrow fees spiked to >40% p.a. yesterday, too high to remain short

MS downgrades DELL to UW from OW, cuts TP from $144 to $110

MS note below is in line with our thesis from a week ago. Memory is 40% of the BOM of traditional servers and 25% of PCs.

Hardware OEM/ODMs face growing risk of margin pressure as the memory supercycle intensifies amidst a tepid hardware demand backdrop. Secular opps in Gen Al/Data Proliferation remain in-place but cyclical headwinds are growing. We're stepping to the sidelines on a number of OEM's as a result.

Key Takeaways
• We are in an unprecedented memory cycle, with spot NAND and DRAM prices — key components in hardware devices - up 50% and 300% in the last 6 months.
• History tells us these cyclical memory supercycles begin to play out via gross margin and multiple compression 6+ months after costs first increase.
• We model Global OEM/ODM gross margins down a median 60bps Y/Y, in 2026 vs. Street up ~10bps Y/Y. MSe now 11% below CY26 Consensus EPS across coverage.
• Global PC and Server (more DRAM) OEMs/ODMs more at risk than Storage OEMs (more
NAND), with Dell, Lenovo and Asustek most 'at-risk
= Downgrading DELL, HPQ, Asustek, and Pegatron to Underweight, HPE, Lenovo, and Giga-byte to Equal-weight.

Canary in the mine

Not yet short btw -- didn’t want to short at the macro bottom last week. Also picking pennies but want to wait to see if there’s a post-print spike for entry

Shorted $QCOM too. Similar thesis to $DELL and $PSTG

Think either the sales or the margins of low/middle-end smartphones are going to get squashed.

@ $171.71

Added $FIX long @ $977 and will continue to add on this correction

From what I gathered on recent podcasts and events, MEP is likely an overlooked bottleneck in the power/DC construction narrative. It takes up to 7 years to train an apprentice to become well-versed in these fields, so supply will likely remain tight.

Closed all $COIN short @ $267 for 15% gain

Not one to call bottom but my gut feel is that this has the makings of one, especially w all the doomposting

Closed 40% of PLTR short @ $168 for 10% gain

$NVDA saved the day (for now) with a strong print that met all buyside bogey numbers. So far, every bullish news have been sold into — so how long this mini rally lasts will tell us a lot about the strength of this market.

Though I am still not convinced that there hasn’t been a massive pull forward of expectations from NVDA GTC that is already priced into the stock price today. I still think competitive pressures and future growth are not being fairly discounted by the market.

My overall view however is that the AI buildout (and thus “bubble”) will continue, but more so in an idiosyncratic way where fundamentals matter more than the past 3 years. So honestly I would prefer not to short AI-related names, but to hedge the case where I am wrong and that there is a bubble, I still lean relatively bearish on NVDA, and neoclouds — and will look to short these on rallies.