哈达哥
475 Posts
哈达哥
Japanese Hada上海人在日本长居，以太坊比特币基金投资人，在华语币圈分享一些浅见的东西。
6 Following
80 Followers
290 Liked
Posts
All
Videos
Traditional mechanical-style value investors often fall into cognitive misconceptions: “If you don’t understand, don’t act; shut out market noise.” What is supposed to be a reliable risk-control principle is frequently distorted into rigid thinking that rejects new things.
Rules that should protect investment bottom lines become shackles that refuse new logic and sidestep emerging sectors. This is also the core reason why traditional value investors are currently struggling. It’s not a mismatch in the logic of the times—most people simply misunderstand value investing in a one-sided way.
Real value investing is never about closing oneself off. After fully grasping the underlying logic, it makes rational choices—never a blanket rejection of emerging areas.
Buying Coca-Cola is value investing; positioning in Nvidia and Micron is also value investing.$NVDAB
Rules that should protect investment bottom lines become shackles that refuse new logic and sidestep emerging sectors. This is also the core reason why traditional value investors are currently struggling. It’s not a mismatch in the logic of the times—most people simply misunderstand value investing in a one-sided way.
Real value investing is never about closing oneself off. After fully grasping the underlying logic, it makes rational choices—never a blanket rejection of emerging areas.
Buying Coca-Cola is value investing; positioning in Nvidia and Micron is also value investing.$NVDAB
AI infrastructure:
Nvidia, Samsung, Micron, Broadcom, Marvell, TSMC, Arista, ASML, and others. Over the past year, the performance and stock prices across the entire upstream and downstream supply chain have been fully realized. While certainty is high, it also means the asymmetry (EV) isn’t as good.$NVDAB
Nvidia, Samsung, Micron, Broadcom, Marvell, TSMC, Arista, ASML, and others. Over the past year, the performance and stock prices across the entire upstream and downstream supply chain have been fully realized. While certainty is high, it also means the asymmetry (EV) isn’t as good.$NVDAB
CZ said that to prevent IBM's quantum computer from cracking the Big Boss' cake, you should lock down a few million Big Boss' cakes. This shows two problems: first, that the Big Boss' cakes really might be cracked; second, that his so-called locking down of the Big Boss' cakes is very flimsy, which also makes the security of the Big Boss' cakes highly questionable. $BTC
This AI revolution will last at least 10 years. The period of greatest risk I perceive is in 2033, which is around the 10th anniversary of ChatGPT’s release. I don’t really believe that the AI era will end just after a few years…$NVDAB
Will storage surpass GPUs? Will Micron surpass NVIDIA? Can you believe it?
The AI compute power race has shifted from “competing on compute” to “competing on memory/storage capacity.” The memory wall is a long-term, insurmountable physical bottleneck. The underlying logic of the AI industry has flipped—storage has changed from a supporting role to becoming the limiting factor for compute capability. Even if compute power is stronger, if the data can’t be fed in, it’s just empty talk. In AI training, over 60% of power consumption is used for data movement rather than computation. This “memory wall” physical bottleneck makes storage the main character that determines the upper limit of compute power. Memory manufacturers are leveraging capacity scarcity to gain the initiative. HBM consumes 3 times the wafer capacity of standard DDR5, and with advanced packaging capacity in short supply, demand far outstrips supply. Bernstein predicts that by 2027, HBM prices may rise by 2 to 2.5 times.
The AI compute power race has shifted from “competing on compute” to “competing on memory/storage capacity.” The memory wall is a long-term, insurmountable physical bottleneck. The underlying logic of the AI industry has flipped—storage has changed from a supporting role to becoming the limiting factor for compute capability. Even if compute power is stronger, if the data can’t be fed in, it’s just empty talk. In AI training, over 60% of power consumption is used for data movement rather than computation. This “memory wall” physical bottleneck makes storage the main character that determines the upper limit of compute power. Memory manufacturers are leveraging capacity scarcity to gain the initiative. HBM consumes 3 times the wafer capacity of standard DDR5, and with advanced packaging capacity in short supply, demand far outstrips supply. Bernstein predicts that by 2027, HBM prices may rise by 2 to 2.5 times.
Ahead of the earnings report release, Micron’s share price surged from below $100 a year ago to above $1,000. The market has started pricing it using a growth-stock logic, which sparked this doubling-level rally. However, capital still has reservations: investors are worried that storage cycles will repeat—after the stock runs up, it could fall again. This very disagreement is the core opportunity.
Comparing with Nvidia’s case clearly validates the logic: Nvidia’s past several consecutive quarters of earnings beat expectations, yet its stock price response afterward kept weakening—sometimes even falling when good news was confirmed. The key reason was that the market had already priced in all the positive news in advance, the gap in expectations disappeared, and the stock lost its upside momentum. Micron is not in that stage right now.
There is still a large amount of capital in the market assessing Micron from the perspective of a cyclical stock. Institutional target prices are only $1,500. The market is still debating whether its forward P/E can reach 20x, and there is a significant gap in expectations. Going forward, as long as the company continues to deliver through its financial results, its valuation is expected to complete the transition from a 20x PE to a 40x PE, with ample room for upside.
Nvidia’s experience has already proved that the best entry window is the period when the market is divided. Once the entire market reaches a unanimous bullish consensus and analysts collectively raise their target prices by a large margin, the rally is left with only a tail-end move. The stage Micron is in today is equivalent to Nvidia two years ago: performance keeps exceeding expectations, but the market has not yet finished switching its valuation logic. It is still using traditional cyclical thinking to price a brand-new growth stock driven by AI. $MUB
Comparing with Nvidia’s case clearly validates the logic: Nvidia’s past several consecutive quarters of earnings beat expectations, yet its stock price response afterward kept weakening—sometimes even falling when good news was confirmed. The key reason was that the market had already priced in all the positive news in advance, the gap in expectations disappeared, and the stock lost its upside momentum. Micron is not in that stage right now.
There is still a large amount of capital in the market assessing Micron from the perspective of a cyclical stock. Institutional target prices are only $1,500. The market is still debating whether its forward P/E can reach 20x, and there is a significant gap in expectations. Going forward, as long as the company continues to deliver through its financial results, its valuation is expected to complete the transition from a 20x PE to a 40x PE, with ample room for upside.
Nvidia’s experience has already proved that the best entry window is the period when the market is divided. Once the entire market reaches a unanimous bullish consensus and analysts collectively raise their target prices by a large margin, the rally is left with only a tail-end move. The stage Micron is in today is equivalent to Nvidia two years ago: performance keeps exceeding expectations, but the market has not yet finished switching its valuation logic. It is still using traditional cyclical thinking to price a brand-new growth stock driven by AI. $MUB
The U.S. Senate plans to advance key legislation for the crypto market in July—the “Clarity Act.” However, with the current congressional agenda crowded, other higher-priority bills, including the National Defense Authorization Act, agriculture, housing, and more, are squeezing the review window, creating significant time pressure for the bill’s on-schedule vote.
Senator Cynthia Lummis, who leads the legislation, said the full text of the bill will be released around U.S. Independence Day. Her team is still working hard to complete a Senate vote in July. As of now, the bill’s text has not been finalized, and multiple parties continue to negotiate over three core provisions: first, detailed ethical constraints governing government officials holding and trading crypto assets; second, a unified anti-money-laundering compliance standard for the crypto industry; and third, defining a regulatory safe harbor for decentralized, non-custodial developers. Disagreements among the parties have not yet been resolved.
Industry observers predict that if, over the coming weeks, the parties cannot finalize the terms and reach cross-party consensus, the bill may miss the July window and the review could be postponed to the “lame-duck” session after this year’s election. At that time, the Congress would be reconstituted, lawmakers’ attitudes would shift, and legislative progress would likely slow significantly—meaning the timeline for implementing a clear regulatory framework for the U.S. crypto industry could be delayed again.
Senator Cynthia Lummis, who leads the legislation, said the full text of the bill will be released around U.S. Independence Day. Her team is still working hard to complete a Senate vote in July. As of now, the bill’s text has not been finalized, and multiple parties continue to negotiate over three core provisions: first, detailed ethical constraints governing government officials holding and trading crypto assets; second, a unified anti-money-laundering compliance standard for the crypto industry; and third, defining a regulatory safe harbor for decentralized, non-custodial developers. Disagreements among the parties have not yet been resolved.
Industry observers predict that if, over the coming weeks, the parties cannot finalize the terms and reach cross-party consensus, the bill may miss the July window and the review could be postponed to the “lame-duck” session after this year’s election. At that time, the Congress would be reconstituted, lawmakers’ attitudes would shift, and legislative progress would likely slow significantly—meaning the timeline for implementing a clear regulatory framework for the U.S. crypto industry could be delayed again.
Significant gains in storage chip stocks this year to date.
Kioxia’s share price surged 760%, with a market cap of $312.5 billion.
$SNDK rose 736%; SK Hynix rose 315%; Micron rose 270%; Samsung rose 168%. $MUB
Kioxia’s share price surged 760%, with a market cap of $312.5 billion.
$SNDK rose 736%; SK Hynix rose 315%; Micron rose 270%; Samsung rose 168%. $MUB
The market’s valuation of Micron (MU) has always been confined to a short-term cycle perspective. The $1,280, $1,500, and even UBS’s raised $1,625 are all conservative price targets anchored to this round of storage price upcycle optimism; they have not yet materialized the industry’s long-term growth certainty. The stock’s reasonable target price should instead look toward $5,790.
The underlying logic is clear and straightforward: the explosion of AI compute creates a massive, high-bandwidth memory (HBM) and high-bandwidth DRAM demand. Storage chips have evolved from being complementary components for consumer electronics into core hardware infrastructure for AI. On the supply side, capacity expansion is subject to strict constraints; the oligopolistic market structure remains solid. Micron’s technological barriers continue to rise in areas such as HBM and automotive storage, and its bargaining power is on a long-term upward trajectory.
The valuation discount driven by past cycle fluctuations will gradually be eliminated, and the earnings center of gravity will be lifted steadily as AI deployments continue. Current institutional price targets only reflect a temporary earnings rebound, completely ignoring the long-term premium of the growth trajectory. While $5,790 may seem aggressive, it is in fact a fair valuation that reflects the long-term certainty of the storage industry, its technology moat, and incremental upside from AI. Short-term price volatility will not change the core path through which its long-term value is realized. $MUB
The underlying logic is clear and straightforward: the explosion of AI compute creates a massive, high-bandwidth memory (HBM) and high-bandwidth DRAM demand. Storage chips have evolved from being complementary components for consumer electronics into core hardware infrastructure for AI. On the supply side, capacity expansion is subject to strict constraints; the oligopolistic market structure remains solid. Micron’s technological barriers continue to rise in areas such as HBM and automotive storage, and its bargaining power is on a long-term upward trajectory.
The valuation discount driven by past cycle fluctuations will gradually be eliminated, and the earnings center of gravity will be lifted steadily as AI deployments continue. Current institutional price targets only reflect a temporary earnings rebound, completely ignoring the long-term premium of the growth trajectory. While $5,790 may seem aggressive, it is in fact a fair valuation that reflects the long-term certainty of the storage industry, its technology moat, and incremental upside from AI. Short-term price volatility will not change the core path through which its long-term value is realized. $MUB
It will be difficult for today’s smartphones to see price cuts in the near term. The key reason is that supply of mid-range phones in the 1,000-yuan class has been significantly reduced. Models with more RAM and larger storage are generally priced higher, and the cost increase of complete units is no longer limited to just DRAM and NAND flash. The new 2nm process chips to be mass-produced in the second half of the year also have procurement costs that remain high, directly lifting the price floor for flagship models.
Market reports indicate that when manufacturers purchase a single Qualcomm Snapdragon 8 Elite Gen6 Pro chip, the cost falls in the range of $300 to $330, which is roughly RMB 2,040 to 2,240. Taking a mainstream top configuration as an example—16GB LPDDR5X memory paired with a 1TB UFS 4.1 flash chip—the combined materials cost for that memory and storage setup alone reaches RMB 2,300.
Considering only three major core hardware components—processor, operating memory, and in-device storage—the manufacturers’ total procurement cost already exceeds RMB 4,300. Before even factoring in additional costs such as the display, imaging modules, battery, chassis components, R&D, contract manufacturing, distribution channels, marketing, and more, the base cost of the complete device has already been substantially raised.
With rigid increases in core hardware costs, it’s hard to lower the terminal price of flagship models. Instead, they have a basis to raise prices. Wanting to wait and buy a large-storage flagship at a lower price is not realistic in the short term.$MUB #苹果股价跌6.1%
Market reports indicate that when manufacturers purchase a single Qualcomm Snapdragon 8 Elite Gen6 Pro chip, the cost falls in the range of $300 to $330, which is roughly RMB 2,040 to 2,240. Taking a mainstream top configuration as an example—16GB LPDDR5X memory paired with a 1TB UFS 4.1 flash chip—the combined materials cost for that memory and storage setup alone reaches RMB 2,300.
Considering only three major core hardware components—processor, operating memory, and in-device storage—the manufacturers’ total procurement cost already exceeds RMB 4,300. Before even factoring in additional costs such as the display, imaging modules, battery, chassis components, R&D, contract manufacturing, distribution channels, marketing, and more, the base cost of the complete device has already been substantially raised.
With rigid increases in core hardware costs, it’s hard to lower the terminal price of flagship models. Instead, they have a basis to raise prices. Wanting to wait and buy a large-storage flagship at a lower price is not realistic in the short term.$MUB #苹果股价跌6.1%
Microsoft says that starting August 1, the price of the Xbox Series S with 512GB of storage will be increased by $100 to about $500; the 1TB version will be increased by $150. Meanwhile, the entry-level Xbox Series X will have its starting price raised to about $750. Microsoft stated that the main reason behind this round of price increases is the sharp rise in storage and memory costs. “The prices of storage and memory used in game consoles have risen by more than 2.5 times, and we expect these costs to double again by the fall of 2027.”
MSFTonAlpha
MSFTUS+5.49%
This round of storage chip price increases has completely overturned the industry’s prior operating logic and is fundamentally different from the traditional cycles driven by smartphones and PCs. In the past, the storage industry was affected by fluctuations in consumer electronics inventory, showing short-term oscillation cycles occurring roughly every 2 to 3 years. The market swings were rapid but weak in terms of sustainability, representing periodic fluctuations driven mainly by passive inventory replenishment.
Now, the industry’s valuation and growth logic have been comprehensively rebuilt. The core driving force has shifted to the massive capital expenditures for global AI compute. With iterative improvements to large models and the rollout of intelligent computing centers, the amount of storage used per AI server has surged several times compared with traditional devices. Storage chips are no longer just consumer accessory components; they are a core, indispensable asset for deploying AI compute.
Leading technology companies continue to intensify their compute infrastructure investments, driving sustained explosions in storage demand. At the same time, overseas major manufacturers are focusing on high-end HBM production capacity, while general-purpose storage supply is tightening, causing the industry’s supply-demand gap to keep widening. Institutions have clearly stated their outlook: in the second half of 2026, storage prices will see a significant jump, and the price-reduction turning point will not appear until 2028.
A consensus has formed in the market: from 2026 to 2028, the storage industry will enter a three-year super upcycle—fully breaking away from short-cycle volatility and moving into a golden age of AI long-term prosperity marked by high certainty and high sustainability. $MUB
Now, the industry’s valuation and growth logic have been comprehensively rebuilt. The core driving force has shifted to the massive capital expenditures for global AI compute. With iterative improvements to large models and the rollout of intelligent computing centers, the amount of storage used per AI server has surged several times compared with traditional devices. Storage chips are no longer just consumer accessory components; they are a core, indispensable asset for deploying AI compute.
Leading technology companies continue to intensify their compute infrastructure investments, driving sustained explosions in storage demand. At the same time, overseas major manufacturers are focusing on high-end HBM production capacity, while general-purpose storage supply is tightening, causing the industry’s supply-demand gap to keep widening. Institutions have clearly stated their outlook: in the second half of 2026, storage prices will see a significant jump, and the price-reduction turning point will not appear until 2028.
A consensus has formed in the market: from 2026 to 2028, the storage industry will enter a three-year super upcycle—fully breaking away from short-cycle volatility and moving into a golden age of AI long-term prosperity marked by high certainty and high sustainability. $MUB
The Nasdaq is hit hard as Apple’s price hikes weigh on it, with Micron delivering an earnings report that exceeds expectations—highlighting the AI industry chain’s payoff landscape: Microsoft, Meta, Amazon, and China’s major internet companies continue investing in computing power infrastructure. The real beneficiaries reaping the dividends are upstream “shovel sellers” such as memory chips.
Institutional pricing forecasts are clear: Jefferies expects memory chip prices to rise 40%-50% quarter-on-quarter in the third quarter this year, followed by another 30%-40% increase in the fourth quarter; the turning point for price declines is set in 2028. Citic Securities has formed a consensus that from 2026 to 2028, the industry will enter a three-year consecutive super-upcycle. Unlike previous short-term inventory cycles driven by consumer electronics, this memory bull market is a long-cycle行情 shaped by AI computing-capex restructuring. $NVDAB
Institutional pricing forecasts are clear: Jefferies expects memory chip prices to rise 40%-50% quarter-on-quarter in the third quarter this year, followed by another 30%-40% increase in the fourth quarter; the turning point for price declines is set in 2028. Citic Securities has formed a consensus that from 2026 to 2028, the industry will enter a three-year consecutive super-upcycle. Unlike previous short-term inventory cycles driven by consumer electronics, this memory bull market is a long-cycle行情 shaped by AI computing-capex restructuring. $NVDAB
The AI computing power industry boom is pushing up global memory and flash storage chip prices, causing storage component costs to surge to an unprecedented degree. Even Apple, which holds significant leverage across the supply chain, is unable to continue absorbing the costs internally. It has officially announced an increase in the global prices of the Mac and iPad lineups, while the iPhone product line has not been repriced at this time.
The price hikes across multiple models are notable: the starting price of the MacBook Neo has been raised from $599 to $699; the 11-inch iPad Pro has increased from $999 to $1,199, and the iPad Air has also been raised to $749. The MacBook Air price has been increased to $1,299, while the entry-level MacBook Pro has risen to $1,999. The maximum increase for a single model is $300, and the overall increase range falls between 15% and 25%.
Apple’s official explanation attributes the price increase to the large-scale expansion of AI data centers. Computing-power vendors are securing a substantial portion of storage production capacity, tightening consumer memory supply and driving spot prices higher. Previously, the company had long absorbed cost fluctuations on its own; now, raising terminal prices has become an inevitable choice. Industry analysis suggests that storage original equipment manufacturers prioritize capacity allocation toward higher-profit AI computing orders, and that chip supply for consumer electronics devices—and the associated cost pressure—are unlikely to ease in the near term. Apple may still have further repricing moves later on. $MUB
The price hikes across multiple models are notable: the starting price of the MacBook Neo has been raised from $599 to $699; the 11-inch iPad Pro has increased from $999 to $1,199, and the iPad Air has also been raised to $749. The MacBook Air price has been increased to $1,299, while the entry-level MacBook Pro has risen to $1,999. The maximum increase for a single model is $300, and the overall increase range falls between 15% and 25%.
Apple’s official explanation attributes the price increase to the large-scale expansion of AI data centers. Computing-power vendors are securing a substantial portion of storage production capacity, tightening consumer memory supply and driving spot prices higher. Previously, the company had long absorbed cost fluctuations on its own; now, raising terminal prices has become an inevitable choice. Industry analysis suggests that storage original equipment manufacturers prioritize capacity allocation toward higher-profit AI computing orders, and that chip supply for consumer electronics devices—and the associated cost pressure—are unlikely to ease in the near term. Apple may still have further repricing moves later on. $MUB
The Nasdaq dipped 1.32%, SpaceX plummeted 16%, and Google took a nearly 5% hit due to talent drain; meanwhile, Intel and Micron surged 5.19%, both hitting new all-time highs.
The market split confirms our predictions: the second-tier tech giants in the US stock market are rapidly catching up to frontline heavyweights like Nvidia and Google, with the core value of AI inference and storage continuously being realized. Storage has broken out of the traditional cycle, entering a super cycle driven by rigid demand. Intel's wafer process has made breakthroughs, and Micron's stock keeps hitting new highs; the market is being re-priced to determine who the true hardcore king of the AI era really is. $NVDAB
The market split confirms our predictions: the second-tier tech giants in the US stock market are rapidly catching up to frontline heavyweights like Nvidia and Google, with the core value of AI inference and storage continuously being realized. Storage has broken out of the traditional cycle, entering a super cycle driven by rigid demand. Intel's wafer process has made breakthroughs, and Micron's stock keeps hitting new highs; the market is being re-priced to determine who the true hardcore king of the AI era really is. $NVDAB
Micron's PE is only 8, and they’re seeing solid growth, while META's PE is at 17 and growth seems to be peaking. It’s clear who’s overvalued and who’s undervalued; don’t let stock price thinking mislead you. Always check the PEG ratio. Good companies with high momentum get cheaper as they rise. Old tech that’s cheap and lacks growth is just a value trap. $MUB
MLCC is a core foundational component of the electronics industry. The market landscape is highly concentrated, with the global top ten manufacturers divided into three tiers: Korean/Japanese, Taiwan (China), and mainland China, effectively monopolizing global mainstream production capacity.
Korean and Japanese companies control high-end industry technology and high-end markets. Murata maintains the global No. 1 position, monopolizing ultra-miniature, server, and high-end automotive-grade MLCC; Samsung Electro-Mechanics leads in production scale, focusing on high-capacity and automotive MLCC with strong cost performance. Taiyo Yuden has deep roots in consumer electronics, with clear advantages in high-frequency, low-impedance products; TDK and Kyocera specialize in high-reliability special MLCC for industrial, defense, and automotive applications, with deep barriers.
Taiwanese manufacturers dominate the mid-tier general-purpose market. Yageo is the global leading supplier of general-purpose MLCC; it has acquired and completed automotive-grade product lines, giving it extremely strong industry influence. Lite-On? (Huaxin Ke) and Darfon Electronics rely on large-scale production capacity, focusing on consumer electronics and industrial control, delivering clear strengths in both supply and cost.
Mainland Chinese manufacturers are the key force driving domestic substitution. Fenghua Hi-Tech is a domestic MLCC leader, serving both consumer and industrial control markets; Huanqiu Group focuses on high-reliability ceramic capacitors, continuously breaking through mid-to-high-end technologies and accelerating the replacement of imported products.
Korean and Japanese companies control high-end industry technology and high-end markets. Murata maintains the global No. 1 position, monopolizing ultra-miniature, server, and high-end automotive-grade MLCC; Samsung Electro-Mechanics leads in production scale, focusing on high-capacity and automotive MLCC with strong cost performance. Taiyo Yuden has deep roots in consumer electronics, with clear advantages in high-frequency, low-impedance products; TDK and Kyocera specialize in high-reliability special MLCC for industrial, defense, and automotive applications, with deep barriers.
Taiwanese manufacturers dominate the mid-tier general-purpose market. Yageo is the global leading supplier of general-purpose MLCC; it has acquired and completed automotive-grade product lines, giving it extremely strong industry influence. Lite-On? (Huaxin Ke) and Darfon Electronics rely on large-scale production capacity, focusing on consumer electronics and industrial control, delivering clear strengths in both supply and cost.
Mainland Chinese manufacturers are the key force driving domestic substitution. Fenghua Hi-Tech is a domestic MLCC leader, serving both consumer and industrial control markets; Huanqiu Group focuses on high-reliability ceramic capacitors, continuously breaking through mid-to-high-end technologies and accelerating the replacement of imported products.
Target A10
Log in to explore more content