DemonRoj retweeted
JPM: Memory stocks that were once trading at 6-7x P/E have significantly de-rated on AI sustainability/funding concerns, HBM de-specs, disbelief in the elongated upcycle and now trade at <3.5x P/E while the EPS has not seen any correction helped by memory shortages across the industry and LTAs. $MU $SKHY
DemonRoj retweeted
JUST IN: UK growth minister declares Britain will not slow down AI development despite growing backlash in the US.
DemonRoj retweeted
1. 반도체 주가가 그냥 망해버림.
2. 미국 주지사들이 데이터센터 승인을 멈추겠다고 해서임.
3. 그럼 반도체는 얼마나 덜 팔릴까?
4. 막히는 데이터센터를 칩으로 바꾸면 얼마나 되는지 계산해봤음.
5. 정확히 세보면 HBM 한 달 치도 안 됨ㅋㅋ
6. 근데 진짜 문제는 따로 있음.
7. 텍사스는 선거 전부터 이미 데이터센터를 막고있음.
8. 새 데이터센터의 전력망 연결을 보류했음. 데이터센터 허가도 전부 멈췄음.
9. 이번에 새 주지사가 뽑혔음.
1월에 취임하면 새 데이터센터 승인을 멈추겠다고 함.
10. 텍사스는 28년 미국 데이터센터 증설의 30% 가까이를 짓는 곳임.
11. 미국 데이터센터 40%가 막힌다는 분석까지 나왔음.
12. 진짜로 멈출 수 있는 건 얼마나 될까?
13. 최대 11GW임.
14. 40%는 발표만 된 계획까지 넣은 수치임.
중단을 하겠다고 하는 주들에서
아직 안 돌아가는 데이터센터는 대부분 착공 전임.
15. 2030년까지 지어질 물량 중 멈출 수 있는 게 11GW임.
그중 10GW가 텍사스임.
16. 이번에 멈추는 건 새로 내줄 허가임.
27년에 문 열 데이터센터는 이미 공사 중임.
17. 그래서 막히는 건 주로 아직 허가 전인 28년분인 거임.
18. 11GW가 다 멈추면 반도체로 얼마나 될까?
19. 1GW 데이터센터엔 GPU가 40만~50만 개 들어감.
젠슨 황이 직접 밝힌 데이터임.
20. 엔비디아 최신 GPU 하나엔 HBM이 288GB 붙음.
21. 11GW면 GPU 440만~550만 개임.
22. HBM으로 치면 27년 세계 공급의 21~28%.
3달 치 정도임.
23. 작지 않은 양임.
근데 막힌 칩이 그냥 사라질까?
24. 안 사라짐.
25. 텍사스에서 막힌 개발사는 GPU를 다른 주 데이터센터로 옮길 수 있음.
26. 다른 주가 그렇게 받아 가면,
2030년까지 덜 지어지는 건 3GW뿐임.
27. HBM으로 치면 27년 세계 공급의 6~8%.
28. 한 달 치도 안 됨.
29. 막아도 수요가 옮겨가는건 전에도 그랬음.
30. 싱가포르는 19년부터 22년까지,
새 데이터센터 건설을 한 건도 승인 안 했음.
31. 넘친 수요는 바로 옆 말레이시아로 갔음.
25년엔 가동 487MW, 계획까지 5GW가 넘는 시장이 됐음.
32. 아일랜드는 21년부터 더블린에 새 데이터센터 연결을 막았음.
33. 이미 연결된 곳들이 계속 커져서,
25년엔 나라 전력의 23%를 썼음;
34. 암스테르담은 1년 만에 전력 상한을 달고 다시 열었음.
35. 세 곳 다 1~4년 뒤 더 까다로운 조건을 달고 중단을 풀었음.
36. 그럼 반도체 주식은 왜이��게까지 빠졌을까?
37. 시기가 문제임.
38. 27년엔 HBM이 모자라서,
데이터센터 몇 곳이 빠져도 칩은 다른 데로 다 감.
39. 28년은 다름.
40. 27년 HBM을 100으로 놓고 매년 늘어나는 공급과 수요를 계산하면,
28년엔 공급이 35, 수요가 28 늘어서
공급이 7 더 많음.
41. 규제로 덜 지어지는 게 하필 그 28년분임.
42. 28년엔 덜 지은 만큼 HBM이 그대로 남음.
43. 원래 차이가 7뿐이라,
3GW만 빠져도 그 차이가 두 배로 벌어짐.
44. 외국인은 이틀 동안 코스피에서 6조 넘게 팔았음.
그중 절반 넘게가 닉스였음.
45. HBM을 제일 많이 파는 회사부터 판 거임.
46. 주가가 덜어낸 건 반도체 양보다,
28년에 대한 기대였던 거임.
47. 그럼 3년 뒤는 어떻게될까?
48. 3년 뒤 그림은 아직 그대로임.
49. 같은 계산으로 29년엔 수요가 공급보다 12 더 많음.
50. 텍사스가 1년 안에 다시 열면,
28년이 더 나빠질 뿐 29년에 다시 모자란 건 안 바뀜.
51. 다만 싱가포르, 아일랜드, 암스테르담은 1~4년 걸렸음.
52. 텍사스가 그만큼 길어지면 29년 물량까지 밀림.
53. 규제가 바꾸는 건 반도체 양보다 언제 팔리느냐임.
✅️3줄요약
1. 미국 주지사들의 데이터센터 승인 중단으로 멈출 수 있는 건 최대 11GW이고,
다른 주가 받아 가면 2030년까지 3GW만 덜 지어짐.
2. 3GW를 반도체로 바꾸면 27년 HBM 공급의 6~8%로 한 달 치도 안 되고,
11GW가 다 멈춰도 3달 치 정도임.
3. 다만 막히는 게 HBM이 남기 시작하는 28년분이라 28년이 더 나빠지고,
텍사스가 1년 안에 다시 열면 29년에 다시 모자란 건 그대로임.
DemonRoj retweeted
If Texas or other states don’t want the economic benefits of data centers, Alabama will gladly take the business.
alreporter.com/2026/10/16/op…
DemonRoj retweeted
Cyclical money doesn’t count. It’s fake money. Let’s do a fun comparison between Micron and Wal Mart. But I’m going to give Wal Mart a really big advantage. Let’s see what happens.
Let’s look at current cash and 2027 profits. Wal Mart is actually negative $45 billion dollars because they borrow a lot of money to pay for inventory and stuff. But for this exercise let’s count their inventory as cash. Including inventory and all debt, Wal Mart has about $11.5 billion dollars in cash. They will make about $22 billion dollars in profit in 2027. They grow their profits around 5% a year every year. Their market cap is right around $1 Trillion dollars.
Micron has about $68.3 billion dollars in net cash. They will make at least $200 billion dollars in profits in 2027. Their market cap is about $780 billion dollars ! I can’t help but laugh when I type that. Crazy.
Here is the fun part. The current 30 year bond pays 5.26%. Let’s compare Wal Mart growing their cash pile and their profits at 5% a year. By Jan 1st 2040, that cash and total cumulative profits will be worth $401 billion dollars. That’s assuming Wal Mart is still in business and growing at their average growth of 5% a year. That’s cumulatively. All together from now until 2040. That’s assuming Ai doesn’t disrupt them and the world still shops at Wal Mart in 2040. I’m not saying Wal Mart won’t be here. But I am saying anything can happen by then! Just like memory might still be under supplied by then!
Now, let’s get to Micron. Micron has $68 billion in cash and they will make $200 billion in 2027. Let’s pretend Micron closes shop in 2028. They make zero dollars of new money from 2028 to 2040, but they put that original $268 billion dollars in a govt 30 year bond and get 5%. I won’t even bother using the extra .26% let’s just use a flat 5% rate. That $268 billion dollars growing at 5% until Jan 1, 2040 will be worth $481 billion dollars. Thats over $80 billion dollars more than Wal Mart by 2040. That’s nearly 4 full years of Wal Mart’s profits today.
So, Micron could literally shut down their business from 2028 to 2040 and they will still have more cash than WalMart at that same time. I don’t give a fuck about cyclical bullshit. The cash is real and no one has a clue when the so called “cycle” will end. At some point their actual business and cash has to matter.
The idea that these crazy “real” profits that MU is making right in front of our eyes doesn’t matter, doesn’t make any sense. I could understand applying all of the bullshit hypothetical reasoning that they come up with if MU were trading at 50x forward PE, but it’s not. It has way more than enough risk built into the price. Way too much built in, IMO! There is no reason MU shouldn’t trade at an average market multiple of 20x forward PE. Or $3,600 a share.
Why are we crushing a stock doing everything right by pricing in a fabricated future without any evidence of that? Why don’t we price in the actual reality and then punish the stock if it ever deserves to get punished? We give Wal mart the benefit of the doubt for its future, when no company’s future is certain. If Wal Mart has a sudden drop in business and profits sink, the stock will get punished, as it should after that news. But not before it happens! That’s how it has always worked. Why is this time different? Why do we now have psychics that think they can see the future?
I keep saying that I’ve never seen anything like this. I’ve never seen a stock get punished for a completely made up story about its future without one piece of evidence that points to it. Usually it’s the other way around. Usually the market prices in hopes and dreams in the future. Not demise! Make it make sense!
DemonRoj retweeted
중국이 2030년까지
AI 데이터센터 5배 더 짓는다고
어제 KBS 뉴스 보도됨
AI서버, MLCC, 반도체, 태양광 소재 등
5개 공장 증설중인
GRT 매출 폭증한다는 이야기
미국은 데센 막고, 중국은 5배 더 짓고
내러티브는 어디로 쏠릴까?
DemonRoj retweeted
We are hearing from internal sources that a bipartisan Senate bill could fast-track AI data center expansion and the power infrastructure behind it.
The bottleneck for AI was never chips alone. It's permits, interconnection and power.
If this passes, watch grid equipment, independent power producers and data center developers first.
We are not Fox News or CNN but we trust our plug.
Fintwit got this wrong (again) as we suggested a few days ago.
DemonRoj retweeted
Worlds Migration & New Collection Update - IMPORTANT!
We’re sharing how existing Worlds will carry into the reboot. Migration and redemption dates have not been set, and this is not an imminent launch announcement. The purpose of this update is to give holders, buyers, and sellers clarity ahead of time.
We previously said any snapshot would use a historical date and that newly acquired Worlds might not qualify. We’re changing that approach to a 1:1 migration. We recognise this differs from our earlier guidance and want to clearly explain what it means.
The replacement claim follows the World:
- Own an eligible, unredeemed World when claiming: You can receive its replacement in the new collection, with relative rarity preserved and the allocation mapped to your legacy World’s rarity and stats.
- Sell before redeeming: The claim transfers to the buyer. Historical ownership alone does not reserve a separate claim.
- Buy an eligible, unredeemed World: You receive its future replacement claim, including purchases made after our earlier announcements.
- Redeem your World: The legacy World is permanently retired from circulation as part of redemption, and its replacement is released to you. Each legacy World has one standard migration claim.
This keeps the replacement opportunity attached to the World someone owns, rather than leaving it with a previous owner. It also avoids making assumptions about who read earlier announcements or why someone sold. We are not treating historical sellers as having acted dishonestly.
The September Magic Eden / Limit Break exploit will be handled separately.
Our on-chain review identified incident-linked movements involving 123 Worlds. Those movements include both suspected theft and rescue activity, so a zero-price transfer alone will not be treated as proof of an unrecovered loss.
Verified thefts will not count as voluntary sales. Verified, unrecovered losses will receive separate recovery allocations from within the planned total supply, without automatically removing legitimate downstream buyers’ claims. Rescued or returned Worlds will be accounted for to prevent duplicate recovery awards. Disputed Worlds and recovery claims will require review.
Replacement Worlds will be created and reserved at launch.
Unclaimed replacements will remain in a dedicated migration vault until redeemed. They cannot be sold or redistributed by the team and will not enter circulation while they remain reserved.
We’ll actively support migration throughout the first year after redemption opens, but claims will remain available indefinitely. Missing that first year will not cause a reserved replacement to expire.
Staked Worlds and legitimate wallet migrations will also be accounted for. Before redemption opens, we’ll publish the full instructions, the rules for exceptional cases, and a way to check each World’s eligibility and claim status.
In practical terms: buying an eligible legacy World includes its future migration claim; selling it transfers that claim to the buyer. Ownership will be checked at redemption, with exploit-related and disputed cases handled under the separate review process.
The legacy collection this migration applies to is opensea.io/collection/topia-…
We’ll share more about the new collection’s art, features, and differentiation separately.
Launch and redemption dates will also be announced separately.
For now, this update establishes how existing Worlds will carry forward.
DemonRoj retweeted
If you’re wondering why this selloff is so aggressive, save this! !
NASDAQ is down 5.1%. S&P 500 is down 3%. $MU −17%. $SNDK −18%.
This is what forced selling looks like.
It isn't happening in one sector. It's literally everywhere you look at once from financials to AI infrastructure and the best businesses like $MU $AMZN and $NVDA are falling hardest because they're the easiest things to sell when funds need cash…
It didn't have to be this way folks.
AI is the golden goose of this cycle…record earnings, contracted demand into 2030 and a market that was finally starting to price it.
But the war with Iran sent oil above $100, pushed inflation back up, forced the Fed into a hike and set the stage for last night's republican humiliation.
The cost of that war is now showing up in every portfolio in America including all the pension funds & 401Ks.
But what hasn't changed is the thesis.
3 days ago $AMD guided above everyone’s expectations.
$SMCI did $19.9bn in a quarter.
$MU and $SNDK are sold out for years, with margin floors written into their contracts. None of that has changed at all…in fact every earnings we get from here gets even better.
The fundamentals are intact. But every chart looks broken rn and has been broken for a couple of months.
These so called analysts spent this morning cutting their 2027–28 AI capex estimates because of the election.
The outcome had been telegraphed for months…these guys have no idea what they’re talking about…just as always chasing the news cycle.
Hyperscalers sign multi year capacity contracts and they don't rewrite them because control of the House changed hands.
What the market needs now is for the macro to stop getting worse.
Trump has accepted the results which takes a constitutional crisis off the table.
An Iran deal and lower oil would open the door for the Fed to cut. When this nonsense “war” fades, earnings are what's left and the earnings are still extraordinary and that’s where the opportunity is…
This is a flush mainly algo driven not the end of the AI trade or American capitalism...stay hard!
DemonRoj retweeted
Think about how wild the cyclical memory thesis for Micron is. The thesis is that we should ignore the hundreds of billions of dollars in profits that Micron will make as far out as orders can go. They don’t matter. Even if they make $500 to $600 billion dollars of profits over the next 2-3 years. It doesn’t matter. All because one day memory prices will fall so much that they will bring down the margins. No one knows if this will actually happen, but the ones that claim it will, are certain. But, they don’t know what the future margins will be or how much the prices will fall. Yet, they assign basically a complete collapse. All based on past cycles that had nothing to do with today’s business. There was no Ai in the past. The worst part is There literally is no way to prove or disprove this hypothesis. Or when that may happen. Micron can beat earnings for the next 8 quarters in a row and they can keep saying the same thing. Every prediction is a complete guess. These are just baseless predictions without any evidence on today’s current data. If Micron were based on current evidence with all of the information we actually have today, Micron would trade at a premium multiple to the market. Not even an average multiple. Definitely not a discount. It would probably command a 50 x forward PE. If anyone out there can prove me wrong with actual current data, please do. Not past bull shit stories. Not predictions without facts. Real numbers for facts. And not technical charts. Pure real fundamentals. Actual data. I’d love to read what you can come up with.
DemonRoj retweeted
$MU $SKHY TSMC “considering” 3D DRAM. Fake news.
100% manipulation. TSMC is a logic foundry, not a DRAM designer, and it has zero 3D DRAM IP.
-3D DRAM took Micron, SK hynix and Samsung decades of research. TSMC doesn’t even have the building blocks for high performance planar DRAM.
-The cell is still 1T1C DRAM: retention, refresh, sense amps, row hammer. That takes decades of DRAM design know how, and TSMC has never built a DRAM, let alone performance DRAM.
-The stacking is 3D NAND style: 100+ layer Si/SiGe superlattice and high aspect ratio etch. TSMC has never run NAND at volume.
-The horizontal capacitor is the hardest step in the flow. Nobody has done it at competitive cost yet.
-Samsung, SK hynix and Micron own the patent wall.
-The incumbents are years ahead. SK hynix showed 56% yield on a 5 layer stack back in 2024. Micron already has two 3D DRAM generations, 3D1 and 3D2, on its roadmap.
-Even the leaders are targeting the 2030s. TSMC has no fab, no team and no IP.
DemonRoj retweeted
On today's stock market crash:
As an investor, on days like today, you have to ask yourself one question: What fundamentally changed over the last 24 hours?
The shift in the fundamental backdrop cannot be attributed solely to the Midterm Election results; markets had already largely priced in the outcome.
So, what changed between yesterday and today? What is the fundamental reason for today's decline?
One of the most likely catalysts is growing concern about a potential "ban" on AI, or at the very least, severe regulation of the industry.
Over the last 24 hours, Democratic lawmakers have publicly called for restrictions on advanced AI development, including an outright "ban on artificial superintelligence."
Frankly, an outright ban will not happen.
Such a ban, if even possible to enforce, would effectively mean the US surrendering its position in the global AI race to China.
Does this mean AI is "over" or today's stock market crash marks "the end?" Absolutely not.
Rather, we believe we are witnessing a sudden shift in sentiment colliding with excessive levels of leverage, potentially triggering forced liquidations by several major institutional investors.
Therefore, today's downturn is more technical than fundamental. This is not to discount the concerns around regulation, they are legitimate. But price action right now does not look orderly or rational.
We are in the midst of the biggest technological revolution in human history.
Zoom out and remain calm.
DemonRoj retweeted
I’m pretty sure we’ve never had this many companies growing earnings better than 40% YoY while having a NTM PEG ratio below 0.5x … and after today many of them now have a NTM PEG ratio below 0.3x with memory stocks below 0.1x
Meanwhile stocks like $AAPL, $WMT, $COST and $KO trade at an average of 38x NTM earnings with EPS growing ~10% YoY (average).
Either many of the companies growing EPS above 40% are going to have huge EPS misses over the next 12-18 months or we have 50-100 companies that are severely mispriced and undervalued.
DemonRoj retweeted
Here is the problem I have with the memory trade right now. I personally believe we are still very early on the memory supply catching demand. I honestly think it might never happen. At least not in this decade. I also think memory prices will stay high forever. Just like new GPU or TPU prices will never drop in price. Every new GPU, TPU, and ASICS will get a new version of HBM. HBM will always be upgrading, and demand a higher price than the previous version. We are so early. Personal interference bots are about to explode. We are not even at the stadium with those, let alone starting the game. So much more memory will be needed for all of these bots. Then we get physical robots in a few years. Don’t even get me started on that. Space and other planets will follow, but that’s way out there. Literally no end to the build out. There are many people that completely disagree with me. Some think we are at the memory peak right now. Some think it’s less than a year away. And some think it’s a year or 2 away. Getting back to my problem with memory: how do we prove who is right? Is memory near peak and about to crash? Or are we in the early innings?
Every earnings that goes by, MU keeps crushing and guiding higher. But investors don’t believe it will continue. Next earnings, MU will make at least $44 billion dollars of profits in one quarter, and guide higher for the following quarter. But that doesn’t seem to matter. Micron stock no longer goes up because people don’t believe the earnings will continue. You see, when facts and data don’t matter, anything goes. Any story or thesis can be used against MU. So, if actual earnings and guidance don’t matter, because they are not believed to continue, how does that get proven to be true or false? If earnings after earnings continue to beat and raise, but the imaginary end of memory is thought to be right around the corner, then what? How cheap does MU have to be for investors to believe it can continue. With their thesis, the current valuation doesn’t matter because they believe memory prices will fall. so the actual earnings don’t matter! How long does this continue? What breaks this thinking? I honestly don’t know. I am 100% convinced Micron will make hundreds of billions of dollars of profits the rest of this decade. I absolutely believe earnings will go up every year for the rest of this decade. But if the market doesn’t believe that, when does the stock reflect the actual earnings? As every earnings report goes by, MU just gets cheaper and cheaper! I’ve never experienced this before. I’ve never seen actual earnings doubted to continue without any signs of a slowdown!
I honestly don’t know what it will take to convince investors that these earnings are durable and lasting.
Maybe the buyback is the only way! I would love to hear everyone’s answers as to what might eventually break this thinking. Obviously if earnings crash, we get that answer, but what if earnings continue to go up every quarter for the next few quarters?
DemonRoj retweeted
Everyone is worried about hyperscalers spending too much money on AI infrastructure, but I think investors are seriously underestimating just how much cash these companies are going to generate over the next several years.
Alphabet, Microsoft, Meta, and Amazon are collectively expected to generate more than $600B in annual free cash flow by 2030, even after accounting for their massive capital expenditures. And this is exactly why I think a lot of the concerns surrounding hyperscaler spending are getting overblown. We're currently witnessing one of the largest infrastructure buildouts in history, with hundreds of billions of dollars being invested into data centers, GPUs, networking equipment, memory, power infrastructure, and everything else needed to support the next generation of AI. Naturally, this is putting enormous pressure on free cash flow in the near term. Some of these companies could even see their free cash flow decline substantially over the next couple of years as capital expenditures continue accelerating.
But here's what I find interesting. These companies aren't borrowing hundreds of billions of dollars to keep struggling businesses alive. They're investing aggressively while their existing businesses are already generating enormous amounts of cash. Microsoft has Azure, Office, and its enterprise software ecosystem. Alphabet has Search, YouTube, and Google Cloud. Amazon has AWS, advertising, and its massive e-commerce business. Meta has one of the most profitable digital advertising businesses in the world. These businesses aren't disappearing because AI infrastructure spending is increasing. In fact, AI could eventually make all of them even more profitable. And keep in mind, the infrastructure being built today isn't just supporting today's AI workloads. It's laying the foundation for AI agents, autonomous systems, robotics, enterprise automation, and entirely new applications that haven't even reached commercial scale yet.
This is also why I remain bullish on the broader AI infrastructure trade. As hyperscalers continue expanding their computing capacity, the demand for semiconductors, optical networking, memory, power equipment, and data center infrastructure should remain substantial.
We're only in the second or third innings of this massive AI infrastructure buildout and we're not going to see the full profitability of these investments until the eighth or ninth inning.
One of the biggest concerns on Wall Street right now is the enormous amount of capital being poured into AI. Companies like Microsoft, Amazon, Google, and Meta are spending and raising hundreds of billions of dollars building data centers and I fully expect some of these companies to experience negative free cash flow for the next couple years as they continue ramping up their capital expenditures. But I don't necessarily think that's a bad thing, especially when you consider the scale of the opportunity ahead.
We're essentially watching the construction of an entirely new computing infrastructure, and building that infrastructure requires enormous upfront investments before the full economic benefits become visible. Think about it. We're still in the earliest stages of AI agents, autonomous systems, enterprise AI adoption, and advanced reasoning models. As these technologies become more capable and widely adopted, the demand for computing power will increase dramatically.
The important distinction is that the spending happens today, while much of the potential profitability may not materialize until years down the road. That's what makes this investment cycle so difficult for Wall Street to evaluate. Investors are looking at declining free cash flow and questioning whether these companies are spending too aggressively, while the companies themselves are preparing for demand they believe will be substantially larger in the future. Of course, there's no guarantee that every company will earn an attractive return on these investments. Some will overspend, some will struggle with execution, and others will emerge as major winners. But I think it's a mistake to judge a multiyear infrastructure buildout solely by the negative cash flow we're seeing during its earliest stages.
Thanks for having me on the show, @cvpayne! Always a pleasure discussing the markets and the incredible opportunities ahead in AI.