The Oracle
8.2K posts
The Oracle
@CifrOracle
The Oracle of Wall Street — None of my posts are financial advice 🇰🇷 Seeing tomorrow’s winners today.
Three Earnings Ahead Bergabung Kasım 2023
288 Mengikuti38.3K Pengikut
I hope you guys bought the dip
No panicans!
$AAOI $ASTS $NBIS $MRLN $FCEL
English
@zephyr_z9 For those who need a cheat sheet btw this is VERY BULLISH $MU
English
The Oracle me-retweet
The fall is so stupid
U don't really need 1.5TB per CPU for training, chatbot interactions, and light agentic workloads
Provisioning that much DRAM for them will just lead to underutilization and waste of resources (500GB-750GB per CPU is enough)
My biggest fear was Nvidia being unable to sell finished Rubin NVL72 racks becuz of the DRAM shortage
Now they can handle demand in a much better way
Jukan@jukan05
SemiAnalysis: NVIDIA is reducing the SOCAMM capacity for Rubin. via @TMTBreakout
English
The Oracle me-retweet
Micron crossed $1 trillion in market cap and it is still undervalued (Save this).
Every AI chip ever built needs high bandwidth memory stacked directly on the die, physically bonded to the processor itself at the package level, because modern AI compute has gotten so fast that the only way to feed it data is to fuse the memory directly to the chip.
You cannot run a GB300, Vera Rubin, Google TPU or an Amazon Trainium without HBM sitting on top of it.
And there are only three companies on Earth that can make it at scale, Samsung, SK Hynix, and Micron.
Right now, every unit of HBM Micron will produce in 2026 is already contracted.
HBM4 for 2026 is sold out too.
Micron's own management stated publicly that the memory crunch will outlast 2026 and push deep into 2027, with demand continuing to outpace every capacity addition they can bring online.
Elon Musk laid out the exact structural reason for this in a recent interview, and his framing is worth looking at.
He said the limiting factor in AI is not the model, not the software, not even the GPU, it is the ability to manufacture chips, specifically logic, memory, and packaging, at the volumes the AI buildout demands.
He then noted something that most people do not know: there is not a single high-volume computer memory fab operating in America right now.
The first Micron fab in Boise, Idaho, part of a $50 billion expansion will not reach volume production until 2027 while the second Boise fab comes online in 2028.
And the New York megafab is 2029 and 2030.
And Elon's conclusion is that even if you take every memory maker's most optimistic production assumptions and stack them together, it still does not come close to meeting the demand that AI infrastructure is projected to generate across that same window.
That is why he said Micron's market cap going to $1.2 trillion or higher is the just the start.
William Blair estimates Micron's HBM revenue grows 164% in 2026 and another 40% in 2027, as AI servers continue consuming more memory per rack with every new generation of hardware.
Despite that trajectory, the stock currently trades at just 7.2x projected 2027 earnings.
A company growing at triple digits, with its entire production sold out for two consecutive years, sitting on the only American controlled fab pipeline capable of addressing the supply gap over the next decade and it is trading at 7x forward earnings.
That is the undervaluation, we are buying more Micron here.
Come join Milk Road Pro and get our full breakdown on where Micron sits in the AI infrastructure stack, including how we are thinking about HBM pricing through the supply crunch and all of our AI thesis.
Link below.
English
Elon Musk on the AI memory crunch:
Demand is so extreme that $MU (Micron) hit ~$1T market cap on HBM capacity fully sold out through 2026.
Yet the U.S. has zero high-volume memory fabs today. New supply won’t ramp meaningfully until 2027–2028.
CapitalFlow@CapitalFlowApp
Never ending $MU calls!!!!
English
@Oculustrade Micron Technology $MU — $9,888,596
$3,118,850 → $1590 calls (Dec 2027)
$1,955,000 → $2160 calls (Dec 2027)
$790,020 → $1650 calls (Dec 2028)
$686,875 → $2000 calls (Nov 2026)
$642,590 → $1260 calls (Oct 2026)
All these call buyers disagreed with you today
English
@demian_ai @CifrBunny Not to mention… look at volume on Micron calls today
Micron Technology (MU) — $9,888,596
$3,118,850 → $1590 calls (Dec 2027)
$1,955,000 → $2160 calls (Dec 2027)
$790,020 → $1650 calls (Dec 2028)
$686,875 → $2000 calls (Nov 2026)
$642,590 → $1260 calls (Oct 2026)
English
This week everything sold together, but that's not a thesis breaking. That's correlation going to one.
The bottleneck book was down 3.7% this week while the S&P was down 2.6%.
I'm not going to dress that up, red is red 🟥, and a week like this is where you find out whether you actually believe the thesis or just liked it when it was working.
So believe it on the evidence, not the mood.
Zoom out to six months:
- book is up 97%
- S&P is up 7%
Same names, same chokepoints. The half-year that contains this ugly week also contains an 89-point spread over the index.
One week did not undo that, because one week is not the unit the thesis runs on.
The cleanest tell isn't the return, it's the dispersion. This week, the spread between the best and worst basket collapsed to 10%.
Memory Supercycle, Cooling, Photonics, Custom Silicon, they all moved together, down, as a single risk-off blob. That is what a selloff is: correlation going to one. The market stops pricing companies and starts pricing exposure.
Nobody this week sold HBM because they reassessed memory bandwidth. They sold it because it was up the most and they wanted out of everything.
Over six months, that same set of baskets dispersed 376%. InP & Substrates up nearly 4x while Connectors sat flat. That gap is the thesis doing its job, the market separating the names by what they physically own and what's actually tight. Dispersion is where fundamentals show up. Selloffs erase it. Time restores it.
And the fundamentals genuinely did not move this week. HBM is still the constraint on what a frontier model can produce per chip-hour. The substation still takes years to energize. The packaging line, the cooling loop, the optical transceiver supply, none of those have a five-day clock. They run on procurement time, construction time, permitting time. A red tape can't shorten a transformer lead time and a green one can't lengthen it. The chokepoints are indifferent to the week you're having.
That's the whole point of buying the chokes instead of the boxes. The boxes trade on sentiment. The chokes trade on physics, eventually. This week, sentiment won. Over any horizon that matters, physics does.
So: nothing changed this week except the price. The thesis was never a bet on five good days. It was a bet that the binding constraints in AI are physical, slow, and owned by a small number of people, and that the market would keep figuring that out unevenly, in fits, with weeks like this one in the middle.
Zoom out
English
$MU So it's been awhile that I covered Memory stocks because of the over extensions we had for too many days
We're back to approaching appropriate levels for another great opportunity upside move from 775-820 to 1000+
My thesis remains bull for 1200+
Will share Alert with Subscribers 🏆
English
The Oracle me-retweet
Quant has spoken, Monday lookin green
The Kobeissi Letter@KobeissiLetter
BREAKING: President Trump says “stocks should go up, not down” after today’s jobs report. “Growth does not mean inflation,” Trump says.
English
The Oracle me-retweet
🤷♀️ here’s what is happening / will happen
- memory is too expensive so firms are designing to use less of it
- margins are incredible so every memory firm will double capacity in 5 years and the Chinese players will 10x lower memory grades
- in 18 months, AI applied to AI research will reduce overall cost per unit intelligence by 99%, this will include reductions in memory
- market for memory stocks will tank as this happens
- Jevons paradox will strike then, token demand will 1000x
- more datacenters will be announced
- there will be another memory shortage in 24 months
- memory stocks will again soar
This will keep repeating over and over with shorter cycle times because humans are retarded and cannot keep exponentials in their long term planning frame.
Paradis Labs@ParadisLabs
You need to be more bullish on $SNDK, $MU, SK hynix & Samsung... Semiconductor Market Forecast (WSTS): -> 2026: $1.5 trillion (+90% YoY) - due to memory demand -> 2027: $1.9 trillion (+27% YoY) - mainly memory-driven again -> Memory forecast: 2026: $800B (+250% YoY) 2027: $1.1 trillion (+32% YoY) I don't see a viable situation rn where memory slows down. Huge demand + supply shortage = pricing power = rocket fuel. -> Also in 2027: Optoelectronics: accelerating 5% growth to $46.4B (vs. 2.7% fcst in 2026). Includes some of your favourite photonics names like $LITE / $COHR / $AAOI / $SIVE at chip level. And $IQE / $AXTI at substrate level. Always good to see other data sources + forecasts to backup overall bullishness.
English
REALLY comfy with 50% cash today. I am itching to buy
sassy@sassy420_
Sitting comfy with my port at 50% cash
English
@TheAIShrink Very well said, it’s not often that he endorses a company and says in plain sight it’s a trillion dollar company, it’s free money $MRVL
English
@CifrOracle $MRVL joining the S&P 500.
It’s the index committee admitting that NVDA’s biggest customer is now too big to ignore. A polite backdoor to holding AI exposure without buying the leader.
English
The Oracle me-retweet
@optionscjp I’m at 120k brotha I feel you! I think people don’t understand how bullish this red day is tho lol
English
Days like today are exactly why you want to build conviction in the stocks you hold
Does it suck seeing -260k in my portfolio today?
Absolutely, but I’ve learned time after time that days like today or just days to screenshot to look back at and laugh in a few months
Of course, there will be a time where the music stops playing but I still believe we are in the greatest buildout of our lifetimes with AI and my conviction is strong in my holdings so I’m not really worried today.
English