Orbit Post Sitemap
$SNDK $NVDA $AMD Recently, the US chip sector suffered a collective setback, essentially a "bloodbath": the Philadelphia Semiconductor Index retreated more than 20% from its all-time high, officially entering a technical bear market, marking the largest weekly drop in nearly a year. Leading companies like Nvidia, AMD, and SanDisk generally fell more than 5%, wiping out over $3 trillion in market capitalization.
This round of sharp declines is not triggered by a single event: the previous AI market overdrew overvaluations, leading the market to question the sustainability of demand. Coupled with rising expectations of hawkish Fed rate hikes, the collapse of capital groups, and the competitive impact from technological breakthroughs by domestic chip makers, institutions have concentrated on reducing positions in high-leverage tech stocks under multiple pressures.
Currently, sector valuations have fallen rapidly, but short-term panic is still brewing. Going forward, it is important to closely monitor the upcoming financial reports from tech giants to verify the true resilience of AI computing power orders, as blind bottom-fishing risks remain high.美股芯片大跳水，对A股带来多大冲击
财联社7月28日美股收盘，三大指数走势分化，道指收涨0.51%，标普500微涨0.02%，纳指下跌0.18%。苹果大涨创出历史新高，总市值逼近5万亿美元，再度反超英伟达登顶全球市值第一。但半导体赛道遭遇重挫，费城半导体指数大跌超2%，创出5月20日以来收盘新低。
英伟达单日大跌接近5%，创下6月5日之后最大单日跌幅。光通信、存储板块成为重灾区，闪迪大跌超11%；铠侠ADR、SK海力士下跌超7%；Lumentum大跌超6%；西部数据、希捷科技跌幅超4%。不少高位AI硬件标的自6月底高点出现大幅回撤：铠侠ADR回撤超57%，闪迪回撤超45%，迈威尔科技回撤超42%，SK海力士更是直接跌破美股IPO发行价。
一、行情背后深层逻辑解读
1、市场内部风格剧烈切换
资金从AI算力硬件高位赛道出逃，回流消费龙头苹果。AI硬件前期积累巨大涨幅，短期获利盘丰厚，出现集中兑现，并不是单纯基本面彻底恶化，属于高位板块的估值消化。
2、存储+光通信双杀
SK海力士破发，海外存储、光模块龙头普遍暴跌，意味着海外资金对于算力硬件周期预期开始分歧。部分资金开始担忧AI资本开支不及前期乐观预期。
3、分化是关键词
大盘指数并没有崩盘，只是前期爆炒的AI硬件板块杀估值，消费龙头苹果继续创新高，属于结构性调仓，并非全市场系统性熊市。
二、映射A股盘面影响预判
按照强关联性映射原理，隔夜美股半导体大跌，会对A股光模块、光芯片、存储芯片开盘形成情绪压制。
但A股半导体有国产替代独立逻辑做缓冲，不会简单复刻美股的巨大跌幅。短期板块大概率会承压震荡，前期涨幅大的高位标的回调压力更大。
三、相关赛道投资价值与6家A股核心标的剖析
光通信赛道
1、中际旭创
全球光模块龙头，海外客户占比高，深度绑定海外AI大厂，直接受益全球算力建设。海外股价剧烈波动会带来情绪扰动，公司基本面订单依旧扎实。
风险：海外AI资本开支不及预期，海外同行价格竞争。
2、源杰科技
光芯片国产核心企业，高速光芯片实现批量供货，国内光模块产业链核心上游。
风险：海外高端光芯片竞争，研发投入持续增加。
存储芯片赛道
3、兆易创新
国内存储芯片龙头，NOR闪存全球市占率领先，同时布局DRAM国产替代。海外存储大厂大跌，会带来板块情绪冲击，但国产替代长期逻辑不变。
风险：存储产品周期价格波动。
4、江波龙
企业级存储模组龙头，AI服务器存储增量的直接受益标的，海外业务占比较高。
风险：海外地缘政策风险，上游原厂供给变化。
半导体设备材料
5、长川科技
半导体测试设备龙头，国产设备核心标的，国内晶圆厂扩产带来长期需求。
风险：晶圆厂资本开支节奏变化。
6、雅克科技
半导体材料平台企业，覆盖光刻配套材料、电子特气、存储芯片材料，深度受益国内存储厂建设。
风险：行业竞争加剧，下游客户资本开支放缓。
四、实操操作建议
1、短期外围情绪偏空，光模块、存储板块会承受压力，当前科技硬件板块处于调整阶段，严禁盲目追高，务必严格控制仓位，不要轻易抄底。
2、区分短期情绪杀跌和中长期基本面。如果后续海外大厂财报、订单数据进一步恶化，才是基本面拐点信号，不能只看单日行情。
3、优先规避短期累计涨幅巨大的高位个股，等待板块充分消化风险之后再做评估。
免责条款：本文仅为公开市场行情解读，不构成任何投资建议。半导体板块波动剧烈，受海外股市、周期需求、地缘多重因素影响，股价存在较大不确定性，股市投资风险自担。#长鑫科技上市，全球存储竞争添变量 $SNDK Capital entry = supply increase = cyclical pressure. Historically, the decline of storage giants is often not due to technological lag, but because peers expand capacity excessively through financing for huge profits, leading to overcapacity. Changxin has secured over 60 billion RMB in massive financing, directly locking in expansion capacity for the coming years. Additionally, the $950 billion agreement order signed by Samsung, SK Hynix, and Nvidia means the market is already trading on expectations of future storage oversupply. 今天SPCX跌的没有存储三傻凶，可能是之前跌到位了吧，也可能是美股需要一个新题材扛大梁。虽然8月就要解锁了，但这是IPO之前就知道的利空，那么从反身性逻辑来看，所有人都知道的利空便不是利空，甚至可能反向轧空
$SPCX 📊 $BNB Liquidation Overview
$874,600 liquidated in 24 hours, with long positions liquidated at $676,100 accounting for 77.3% of the total, short positions liquidated at $198,500, making longs 3.4 times the shorts. In 1 hour, long liquidations were $86,400 (99.99%), shorts faced almost zero resistance; in 4 hours, long liquidations were $316,200 (96.8%), indicating escalating long position liquidations; in 12 hours, long liquidations reached $464,100 (91.8%), marking the most severe long liquidation window of the day; although shorts counterattacked to $198,500 (22.7%) in 24 hours, longs still dominated absolutely. Liquidations concentrated in the 12-hour period (57.8%), with the 24-hour total being 1.73 times the 12-hour amount, and shorts strengthening their counterattack in the latter 12 hours.
In summary: $BNB experienced a concentrated 12-hour main downtrend wave with continuous long liquidations and a decisive short victory.
🔥 Market Indicator | July 27
Today's three hot topics point to the same theme: AI narrative entering the "validation season"—from the valuation frenzy of domestic storage, to the Federal Reserve's interest rate decision, to the earnings tests of tech giants.
📈 ChangXin Technology IPO: The 3.66 trillion yuan "domestic substitution" frenzy
On July 27, domestic DRAM leader ChangXin Technology officially listed on the STAR Market, surging 471.59% at open, with market value briefly surpassing 3.66 trillion yuan, overtaking ICBC as the largest A-share by market cap. Expected net profit exceeded 50 billion yuan in the first half, with global market share rising from 3% to 8%. However, controversy remains significant: technology still lags about two generations and three years behind South Korean and American giants. Whether the 3.66 trillion yuan valuation marks the start of a super cycle or a peak moment is sharply debated. After ChangXin's listing, Samsung Electronics and SK Hynix each dropped about 4% intraday.
🏛️ Federal Reserve Interest Rate Decision: Rate hike expectations simmering
The Federal Reserve will hold its policy meeting from July 28 to 29. Economists unanimously expect no change, but interest rate futures price in a 36% chance of a hike. The divergence stems from oil prices—Brent crude has surpassed $100/barrel, with US-Iran tensions raising geopolitical risk premiums and inflation pressures reemerging. Whether Fed Chair Powell will deliver a "surprise hike" will be revealed early Thursday.
📊 Microsoft, Meta, Amazon Earnings: AI "burn rate" model under scrutiny
This week, Microsoft, Meta, and Amazon release earnings, sharing a core question: can massive AI capital expenditures translate into real revenue? Google and Tesla have already sounded alarms with their first-ever negative cash flow—AI spending is faster than expected. Whether Microsoft Azure can maintain over 40% growth, Meta's capital expenditure guidance raised to $125-145 billion with AI potentially eroding ad profits, and Amazon AWS growth breaking 30% will determine if the "AI narrative" can continue to support tech stock valuations.
💎 Summary
ChangXin Technology's 3.66 trillion yuan valuation is an extreme pricing of "domestic substitution + AI demand"; the Fed's rate decision is a tense game over "whether inflation will return"; tech giants' earnings are the ultimate test of "whether AI spending can be profitable." The AI narrative is moving from "storytelling" to "answering the test." #长鑫科技上市，全球存储竞争添变量
#美联储周四凌晨公布利率决议
#财报观察员：微软Meta亚马逊能稳住AI叙事吗？ BTC leads the gains, but knockoffs lack the willingness to follow, and the market is showing structural divergence rather than a full recovery.
The question is: if new funds are not widely entering the market, which forces are mainly driving the current price increase?
Core Facts of the Original Text: The market is generally rising, but participation is selective. Open interest cooled, trading volume remained stable, and funds were concentrated in a few assets ($JELLYJELLY, $OPG, $SLX, etc.), while many altcoins ($BEAT, $EDGE, etc.) remained weak. BTC is the liquidity anchor, ETH represents institutional participation, SOL is high beta L1 exposure, HYPE is a risk appetite indicator, and DOGE/ZEC is a thermometer of retail investor sentiment.
Structural change: The decline in open interest accompanied by price increases indicates short covering or leveraged withdrawal rather than new long positions. This explains why the rally failed to spread—existing funds are focused on the game rather than incremental entry. The relative strength gap between BTC and ETH has widened, indicating that the recovery in risk appetite remains highly uneven.
Pricing impact: If open contracts continue to shrink while prices persist, a short-term short squeeze may form, forcing short sellers to close positions and push prices higher. However, if trading volume shrinks simultaneously, the rise will lack sustainability. The weakness of altcoins means the current structure favors BTC and a few strong narrative assets rather than broad longing.
Bullish path: Bearish squeeze continues, BTC breaks through key resistance to attract hedge funds to buy long, followed by ETH, driving some altcoin rotation. Condition: Open interest rebounds after a decline + BTC weekly volume breakout.
Bearish risk: The current rise is only a brief rebound from short covering; after the funding rate turns positive, bears re-enter the market and prices fall back. Counterfeit funds continue to flow out, making BTC's solo rally unsustainable. Condition: Open interest continues to decline + volume shrinks + counterfeit index confirms breakout.
Conclusion: Wait for open interest to stabilize or volume amplification for confirmation; maintain position flexibility until then.
Risk warning: Derivatives data lags behind, and squeeze paths may reverse. $BTC $ETH $SOL#多数党领袖称CLARITY休会前难通过
Senate Majority Leader Toon has temporarily shelved the CLARITY Act to focus on government official nominations and the Russian sanctions bill. On Tuesday and Wednesday, the late Senator Graham's funeral was occupied, leaving the Senate with no time to handle controversial bills. The bill may only enter the voting process until next week, the last few days before the recess on August 8.
The time window is closing, and there aren't enough votes.
Galaxy Digital's research director has already cut the bill's approval probability within the year from 50% to 30%. The probability of passing on Polymarket is now only 30%-33%. The Republican Party can actually mobilize about 50 votes, while breaking the lengthy debate requires 60 votes. On the Democratic side, seven negotiators have publicly said "no," with the key disagreement being the moral clause. Democrats believe that the DOJ's monopolization of enforcement power means Trump's DOJ enforces the provisions against Trump, and that the provisions will automatically expire until January 20, 2029.
Within a week, the shift went from "finally coming" to "basically no chance this year."
My view is:
The CLARITY Act is very unlikely to pass this year. After the August recess, the September return agenda will be completely crowded with the budget, the National Defense Authorization Act, and midterm election preparations. The legislative window for 2026 is effectively close to permanent closure.
The biggest risk in the industry has never been overly strict regulation, but that regulation never comes. Uncertainty is the real killer. Friday's FOMC was already headache-inducing, and now there's another policy variable hanging overhead. Yesterday, watching the sharp drop in storage reminded me of a topic I discussed with friends earlier this year: the vast majority of people are still at the surface. Blockchain = decentralized coin speculation, AI Agents = model capability competition. But essentially, the bottleneck of AI entity-free finance has never been AI computing power. AI can think autonomously, make autonomous decisions, and trade autonomously, but it has no legal entity, no customer service appeal channels, and no boundaries of responsibility. Once unattended robotic financial activities encounter disputes, interceptions, tampering, or violations, traditional systems are almost unable to provide a safety net. Blockchain is not the protagonist here, but rather the essential trust foundation and security sandbox for AI finance—decentralization is just a byproduct; controllability, auditability, and traceability are the core values. This top-level understanding is still seriously undervalued in the market. At the beginning of the year, after communicating with teams deeply involved in agent-driven finance, I realized: big companies are competing on computing power, and the industry lacks a trusted sandbox. Internet giants (Alibaba, Google, etc.) are fully investing in large models and Agentic RL to solve AI's "whether it wants to work and whether it can work." But what truly determines whether AI entity-free finance can be scaled has never been more computing power or storage chips, but trust, compliance, security boundaries, and execution certainty. What the entire industry generally lacks is precisely this more critical foundational infrastructure: the secure sandbox for AI finance, trust red lines, compliance isolation, and unmanned enforcement guarantees. Through three companies—Circle, ETH, and OKX—SanDisk crashed through 1300, what happened?
First, the market started worrying that AI investments are too large and returns are not keeping up, so chip and storage stocks that surged the most early on were sold off first. On the same day, Micron, Western Digital, Seagate, and SK Hynix all fell simultaneously, indicating that funds are withdrawing from the entire storage sector, not just hitting SanDisk.
Additionally, high oil prices and US Treasury yields are suppressing high-valuation tech stocks. SanDisk is due to report earnings on August 5, so funds chose to reduce positions early. SanDisk had the strongest gains earlier and the most crowded chips, so the decline was amplified.
In the short term, watch if 1223 can hold; the resistance above is now between 1318 and 1325. Until it breaks back above 1320, I consider this just a rebound and am not in a hurry to bottom-fish.
If it effectively breaks below 1223, the next target is 1170; if weakness continues, then 1120, with 1000 only as an extreme scenario.
Just recording my personal market view.
$SNDK 1. U.S. Stocks: Chips Lead Declines Again, Funds Await Tech Giants' Earnings Reports and FOMC On July 27, U.S. stocks diverged: the Dow Jones rose slightly, while the S&P 500 and Nasdaq fell about 0.3% and 0.6% respectively. Nvidia plunged about 5.1% to $196.51, dragging down the AI and semiconductor sectors; SPY closed at about $739.09, and QQQ at about $682.12. Opportunities and trends: This week's earnings reports from Microsoft, Meta, Amazon, and Apple will verify cloud businesses, agent commercialization, AI capital expenditures, and free cash flow; If revenue growth can cover CapEx, the power chains for hashrate, storage, servers, and data centers may recover. Risks and warnings: The market no longer unconditionally rewards massive AI investments. If tech giants continue to raise their investment levels but fail to provide clear return cycles, semiconductor and platform tech stocks may remain under pressure. Key data changes: The yield on the US 10-year Treasury fell to around 4.65%; Last week, US equity funds saw a net outflow of $7.34 billion, but technology sector funds flowed in for the fourth consecutive week, with about $2.12 billion in cash flowing in a single week, indicating that funds are still undergoing structural rotation within AI. ⸻ 2. AI/Large Models: Anthropic launches Opus 5, competition shifts to "close to flagship, cost halved" Anthropic releases Claude Opus 5, with performance close to its flagship Fable 5 but with roughly lower operating costs盘面突然安静了几分钟，交易员们都在等那个数字。
你猜到了吗？美股巨头财报其实已经悄悄变成了币圈的晴雨表。
Alphabet 交出了一份看似漂亮的 Q2 成绩单：营收 1198 亿美元，Google Cloud 继续稳健增长。但盘后 $GOOGL 跌了 4% 以上。为什么？因为市场真正在交易的，不是过去，而是未来。
关键点在这里：
- Alphabet 把 2026 年资本支出预期从 1800-1900 亿美元上调到 1950-2050 亿美元，自由现金流却转负了。
- 华尔街越来越焦虑：AI 故事很性感，但烧钱速度让人不安。
- 谷歌、微软、Meta、亚马逊预计 2026 年合计资本支出 7250 亿美元，比去年暴涨 77%。
市场奖励的不再只是"超预期"，而是"健康的预期"。未来指引、现金流、AI 投入效率——这些才是定价的核心。
Tesla 讲了另一个故事。它手里还握着 11509 个 BTC，从 2022 年起没动过。虽然 Q2 因为比特币前期下跌计提了 1.12 亿美元亏损，但既没恐慌抛售，也没趁机加仓。就只是 HODL。安静，但很有态度。
这对我们 crypto 市场意味着什么：
- BTC 继续吃 ETF 资金流入的稳定红利，但这不是唯一的叙事。
- 币圈和纳斯达克 100 的关联度仍然很高，科技巨头财报会影响情绪传导。
- 接下来微软、Meta、亚马逊的财报和指引，可能会通过预期差直接塑造美股和 crypto 的风险偏好。
一个被很多人忽略的细节：美股收盘后，crypto 是 24 小时交易的。像 OKX 上的代币化美股 $XGOOGL、$XTSLA 可以用 USDT 在财报发布期间和周末继续交易。这意味着，传统市场的定价间隙，正在被 crypto 填补。
现在的问题是：下一波巨头财报，会强化 crypto 的上行信心，还是带来新的脆弱点？
我的感觉是，市场正在从"AI 狂热"切换到"AI 算账期"。如果巨头们继续上调资本支出但现金流转弱，crypto 的高 beta 属性可能会在同一条船上颠簸。但反过来，如果哪家给出超预期的效率提升信号，整个风险资产的情绪都可能被点燃。
别只看价格，多盯着指引和烧钱节奏。这才是这一轮真正的交易主线。
$BTC $ETH $GOOGL $TSLA
#EarningsRealityCheck #AIvsCrypto1. Direct Trigger: Changxin Technology's A-share listing brings long-term competitive expectations and repricing. Changxin surged on its first day, raising tens of billions in funds, and plans to fully expand DRAM production and develop high-end memory in the future; Foreign traders begin pricing in DRAM: Global DRAM has shifted from a three-oligopoly → a four-strong competitive landscape. In recent years, Samsung, SK Hynix, and Micron have proactively controlled production capacity to create tight supply, continuously driving up chip prices and earning ultra-high gross margins. Market concerns: After Changxin's capacity continues to ramp up, overseas manufacturers' "price control is weakened," and the sustainability of long-term storage price increases is questionable. 2. Underlying Internal Causes (Most Important! Without this, there would be no sharp drop) 1. The sector saw huge gains in the early stages, with profit-taking and extremely crowded chips. In this AI storage bull market, SanDisk and SK Hynix's largest gains this year have doubled or even multiplied, and their stock prices have already priced into optimistic expectations for storage price increases and AI demand over the next 1~2 years. The high-level sector is a stock game, with no continuous new capital flowing; Once bearish sentiment appears, a large amount of unprofitable funds concentrate to take profits, easily leading to a sell-off and crowded market. 2. Market Divergence: Storage Price Increase Cycle May Approach Turning Point Morgan Stanley and other investment banks warn: The current DRAM/NAND spot price increase rate is very likely to slow and peak in the fourth quarter. Capital begins to gamble on marginal changes in the cycle: The optimistic expectation is "AI continues to consume storage, extending the cycle"; The pessimistic outlook is: terminal PC and mobile phone demand remains weak, and relying solely on server demand alone is insufficient to sustain chip price increases. The capital market fears marginal weakening of prosperity and capital increases隔夜美股存储赛道迎来明显调整，引发不少投资者讨论。SK海力士美股ADR盘中回落，股价跌破IPO发行价；闪迪股价持续走弱，成交量显著放大，近一个月自前期高点出现明显回撤。美光科技、西部数据等多家海外存储企业同步震荡下行，费城半导体指数也受到板块情绪拖累。 恰逢国内DRAM产业迎来关键发展节点，国产存储龙头持续推进产能扩张与技术迭代，在全球市场份额稳步提升。很多人自然而然将两件事关联起来，认为海外存储股价走弱直接由国产厂商冲击造成。这里首先厘清一个核心观点：二者不存在简单直接的因果关系。国产存储突破是长期产业变量，只能影响市场远期预期；本轮海外个股调整，是高位获利盘兑现、行业预期重估、宏观资金环境多重因素共振的结果。本文仅客观复盘盘面与产业逻辑，不构成投资建议。 一、先看懂隔夜美股存储板块真实行情，避开片面解读 北京时间7月28日凌晨收盘的美股市场，存储板块成为科技赛道调整的重点方向。 SK海力士ADR于7月上旬登陆纳斯达克，发行价149美元，上市初期股价冲高，积累了不少短线获利盘。隔夜交易时段抛压集中释放，股价正式跌破发行价。作为全球HBM高带宽内存领域的头部厂商，千亿市值龙头出现破发，隔夜美股半导体板块遭遇了一场血洗级别的抛售。SK海力士的ADR跌破了发行价，这个全球HBM内存的绝对霸主，这个站在AI算力需求金字塔最顶端的存储巨头，居然在美股被砸到所有IPO投资者全部套牢。消息传到亚洲，所有半导体群里一片哀嚎，仿佛明天开盘A股存储芯片板块就要被活埋。 恐慌到这个程度，反而值得冷静下来看一看，这场抛售潮的背后到底是什么在驱动。SK海力士ADR破发，拆开来看有三层驱动因素。 第一层是短期交易层面的被动抛售。SK海力士的ADR上市时间不长，流动盘偏小，容易被大资金的调仓行为剧烈冲击。隔夜美股整体科技板块因为利率预期变化和获利盘回吐出现系统性回调，许多对冲基金和量化策略在科技板块上集中减仓，流动性偏弱的标的首当其冲被砸穿关键价位，触发更多的止损盘和程序化卖单，形成了连锁踩踏。这种流动性驱动的抛售，和SK海力士的基本面没有任何关系，是交易结构问题被市场系统性风险放大了。 第二层是存储芯片行业预期的修正。过去几个月市场对HBM的追捧过于狂热，把SK海力士的股价推到了一个提前透支了未来几个季度业绩的高度。当英伟达的出货节奏出现短期波动，市场立刻开始担心HBM的供需缺口会不会没有之The groundbreaking ceremony for the foundation has concluded, and now the load-bearing walls are being poured—ETH validator exit queue reset to zero, which is equivalent to dismantling all temporary scaffolding on the construction site. The backlog of 2.6 million ETH exit requests from September has now been completely cleared, allowing any staking teams wanting to withdraw to leave without waiting. But don’t celebrate too soon; next to the material yard, 2.48 million ETH are still queued to enter, with an estimated construction period of 43 days—this means new rebar and concrete are continuously being delivered to the site, and in greater volume than those exiting.
From a structural mechanics perspective, currently 40.9 million ETH are staked in the consensus layer, accounting for 33.55% of the total supply, with about 885,000 active validators on duty, yielding an average annualized return of 2.64%. This ratio is already close to the critical load limit designed for the load-bearing walls—too low and the foundation is unstable, too high and the floors crack. 33% is exactly the critical reinforcement ratio for the core tube of high-rise buildings. The exit channel is cleared while the entrance queue is long; net flow has reversed from outflow to inflow, like the final vibration before concrete sets—volatile but with a clear direction.
Many so-called “decentralized” public chain projects are just temporary scaffolding solutions. ETH’s staking channel design is the true structural blueprint—scalability is reflected in the elasticity of exits and entries. Exits have no wait, but entries queue up, just like building codes require evacuation stair widths and entrance lobby capacities to match. Now the evacuation stairs are empty, while a long line waits in the lobby to check in. According to architectural plans, this kind of asymmetric flow often signals the next phase will add a podium or widen the lobby.
Look at that 2.64% APR, even lower than bank loan interest rates. When building load-bearing walls, does anyone ignore the corrosion rate of the rebar inside? This yield is the durability indicator of the building materials; too low means maintenance costs exceed returns. Yet capital keeps flowing in, indicating the market is betting not on rental income but on the land’s appreciation—ETH as the “deed” of the digital world, where the underlying protocol’s security and scalability represent the true value of the building site.
The linkage between the US stock token XDELL and ETH is like the relationship between a steel frame and concrete infill walls—mutually supporting during market resonance, but also causing stress concentration during volatility. But true architects never focus on cracks in the infill walls; they only watch the stress state at beam-column joints. #ethexitqueuezeroSK海力士跌破发行价：“中国对手”“AI泡沫焦虑”引爆芯片抛售潮
7月27日，美股芯片板块遭遇了一场“黑色星期一”。费城半导体指数暴跌超5%，创下自5月19日以来的最低水平。而最受关注的“受害者”，是刚刚以265亿美元创下今年全球最大IPO之一的SK海力士（SKHY.US）。
当日，SK海力士ADR盘中一度重挫10%，最低触及139.01美元，不仅远低于149美元的发行价，最终收报143.02美元，正式宣告破发。这已是纳指连续第四个交易日下跌。由AI狂热催生的半导体泡沫，正在经历一场剧烈的挤兑。
一、破发真相：当“中国对手”站上牌桌
SK海力士跌破发行价，市场给出了两个核心归因。
最直接的冲击，来自大洋彼岸的竞争对手。 7月27日，中国DRAM龙头长鑫科技（688825.SH）登陆科创板，首日暴涨466%，市值飙升至约4840亿美元。作为全球第四、中国第一的DRAM厂商，长鑫科技的上市意味着全球存储芯片市场迎来了一个不可忽视的“中国变量”。更令市场警惕的是，另一家存储巨头长江存储也预计将在今年晚些时候启动IPO。当中国存储双雄开始在全球资本市场上攻城略地，投资者对既有巨头未来市场份额的担忧迅速转化为抛售。
第二重压力，来自AI基础设施投资的“泡沫焦虑”。 报道称，英伟达正在洽谈总规模超过7500亿美元的AI基础设施项目。巨额投资让市场开始质疑：AI的“烧钱”速度是否已远超其创造回报的能力？“循环融资”的担忧再度升温。英伟达企业债的违约保护成本因此创下历史最大单日涨幅，投资者风险偏好急剧降温。英伟达当日股价下跌近5%，市值被苹果反超。当“利好”被解读为“利空”，即便是SK海力士宣布与英伟达合作建设超2吉瓦AI数据中心的“王炸”消息，也未能挽救股价。
与此同时，高估值半导体板块的整体获利了结也在加速。
二、从首尔到台北：一场席卷亚洲的“芯片海啸”
美股的抛售只是第一块倒下的多米诺骨牌。对于极度依赖半导体产业的亚洲市场而言，这场风暴才刚刚开始。
· 韩国首当其冲。 韩国KOSPI指数权重股高度集中于半导体板块。作为KOSPI的“压舱石”，SK海力士和三星电子的任何风吹草动都将直接决定大盘的生死。SK海力士股价自5月以来一度飙升132%，积聚了大量获利盘，回调压力本就巨大。
· 台湾地区同样难以幸免。 台积电、联发科等权重股大概率将同步承压，拖累大盘表现。
· A股存储板块短期承压。 虽然中概股周一逆势走强，但从情绪传导来看，A股存储芯片相关概念股开盘将面临压力。
· 连锁反应或将波及整个亚洲市场。 随着投资者持续撤离此前涨幅较大的半导体个股，这种避险情绪可能从芯片股蔓延至整个科技板块。
当全球资本开始重新审视AI投资的回报周期，当中国存储芯片力量正式登上全球牌桌，一场由“中国对手”和“泡沫焦虑”联合引爆的芯片股大回调，正在从华尔街向首尔、台北和上海蔓延。SK海力士的破发，或许只是这场全球半导体估值重构的第一块多米诺骨牌。
你觉得SK海力士的破发是“黄金坑”还是泡沫破裂的开始？欢迎在评论区聊聊你的判断。
觉得分析有深度？点个赞支持一下，转发给炒股的朋友看看。关注我，每天带你读懂全球市场背后的深层逻辑。#长鑫科技上市，全球存储竞争添变量 $SKHYNIX Friends, overnight US stocks and overseas markets have sent clear signals of risk appetite shifts. U.S. stocks were mixed (Dow +0.51%, Nasdaq -0.18%), but beneath the surface calm, undercurrents surged—the semiconductor sector was hit by concentrated sell-offs, with the Philadelphia Semiconductor Index down over 2%, SanDisk plunging over 10%, Nvidia plunging nearly 5%, and SK Hynix plunging 7.5%, falling below its IPO price. Capital is accelerating its withdrawal from high-valuation chip stocks and shifting to super-heavyweight leaders like Apple (+1.17%), pushing its market value close to $5 trillion and reclaiming the top spot globally. This is not simply a sector rotation, but an explicit expression of capital risk aversion—selling flexible assets and embracing certainty assets. Meanwhile, the U.S. and Iran paused their attacks for several consecutive days. Trump said he had "enough patience" for negotiations, and the rapid fading of geopolitical premiums combined with demand concerns caused international oil prices to plunge 8% in a single day, with WTI crude falling below $82. A sharp drop in oil prices temporarily eases inflationary pressures and theoretically benefits risk asset valuations, but if the decline is due to expectations of a global economic slowdown, caution is warranted for negative feedback links. Insights for the crypto market: Whether semiconductor capital outflows into digital assets like Bitcoin is worth closely monitoring—if U.S. stocks remain volatile, some liquidity may spill over into the crypto market; However, if the oil price crash triggers a chain reaction in commodities, putting overall risk assets under pressure, crypto will not be immune to this. Currently, the macro environment is intertwined with bullish and bearish. It is recommended to closely monitor the linkage between Bitcoin and the Nasdaq, respond with a range-bound consolidation approach until the direction becomes clear, and strictly control position risk. External uncertainties have not yet disappearedThe Federal Reserve meets this Wednesday. If they raise interest rates, the AI stocks you hold will be the first to get hit.
The new Fed Chair Warsh is recognized on Wall Street as a hawk, and since taking office, he has been emphasizing one thing: bringing inflation back to target. Unfortunately, oil prices have surged again this week due to Iran, so inflationary pressure hasn't eased.
Therefore, this Wednesday's FOMC meeting is the biggest risk event of the week.
Why say AI stocks will be the first to get hit? Because most AI stocks are high-valuation growth stocks. The growth stock logic is "not making money now, betting on big profits in the future." But rate hikes mean money becomes more expensive, so future money discounted to the present is worth even less. When interest rates rise, it kills stocks that rely on future expectations to support their valuations.
Nvidia, Micron, Tesla, SpaceX—all follow this logic.
I know what you're thinking now. You're wondering, "Should I exit early and wait for the Fed decision before re-entering?"
I advise you not to rush. The market has likely already priced in some of the rate hike expectations. If the decision day just maintains the status quo, it might actually be a relief rally. If you exit now, you might be selling at the darkest moment before dawn and then watch it rebound.
If you're using spare money and holding good companies, this week's volatility doesn't concern you. If you're leveraged and your mindset collapses at the first dip, then you should have reduced your position long ago; that has nothing to do with the Fed.
$BTC Rebound ≠ reversal, $ETH surged 4%, $QQQ was dazzlingly green, and the market was waiting—whoever showed weakness first would set today's tone.
Look at the numbers
$BTC 65,283 +1.45% $ETH 1,952 +4.14%
$QQQ -1.12% $SPY +0.10% $IBIT -0.82%
$DXY -0.15% $GLD +0.10%
Hormuz and crude oil are still adding variables to inflation expectations, while the shadow of US Treasury yields and Fed tightening continues to weigh on valuations. The dollar is not a backdrop; a simple adjustment of the exchange rate line can disrupt the rhythm of $QQQ$SPY. Today, it's not surprising if any switch gets touched on this plate.
$ETH is clearly more elastic than $BTC, short-term risk appetite is rising, but $QQQ is sinking downward, and money is shrinking into defense. $IBIT Weaker than spot $BTC, a weakness in ETFs means the spot market isn't that strong; $DXY Only when risk assets can breathe a sigh of relief can they catch their breath, but once tightened, they quickly turn hostile; $GLD Still quietly rising, haven't fully withdrawn safe-haven funds, don't be fooled by the surface buzz.油价跌了，美债涨了，芯片股全线承压。
国际油价急刹车，WTI跌破68，布伦特逼近70关口。地缘溢价在快速出清——美伊从升级到暂停互袭，霍尔木兹油轮触雷也没能托住。国内能化期货跟着降温，SC原油、燃油、PTA全线绿，炼厂开工率反而还在升，供需基本面回归主导。
另一头，30年期国债期货创近8个月新高，收益率跌破2.15%。债市告别"三低"震荡（低波动、低收益、低配置），资产荒逻辑没变，只是定价锚从"降息预期"切到了"避险配置+资产荒"。银行存款利率密集下调，宽货币向宽信用传导，长端利率中枢下移还有空间。
最硬核的信号在亚洲时段：韩股单日跌6%，日经跌超3%，费半指数逼近关键支撑。AI烧钱疑虑重燃，美股财报季前夜，资金选择先撤一步。SK海力士、三星电子、台积电ADR盘前全绿。这不是简单的技术回调，是市场在重新给"AI Capex无限外推"定价。
长鑫存储IPO在即，阿里账面浮盈超1500亿。合肥国资产业投资总浮盈破万亿。这才是确定性在哪里——产业链上游卡位完成、国家队真金白银陪跑、技术突破有时间表。
盘前看三条线：
1. 油价能否企稳70下方，决定通胀预期拐点
2. 30年国债收益率能否守住2.15%，决定资产定价锚
3. 费半指数支撑位（5000点附近）能否站住，决定AI产业链信心恢复节奏
逻辑在跑通，等市场给确认。#美军暂停对伊空袭，国际油价开盘大幅下跌 $CL I really didn't sleep well last night. On one hand, China's storage IPOs were setting new records for A-share single-share market cap ceilings; on the other, various crashing stocks in the US storage industry were taking over.
From SpaceX to the super correction in the storage sector (SK Hynix, Samsung, Micron, SanDisk—either plunged or crashed in flashes), the myth of super IPOs in the US stock market is gradually cooling down this year.
Chinese companies represented by Changxin are on the rise and overpower.
As AI semiconductor valuations fall and capital takes profits, high-valuation assets are beginning to be repriced by the market.
Musk's net worth evaporated by $130 billion in five days, reflecting that the capital market's risk appetite for the "AI narrative" is cooling down.
Suddenly, a DOGE coin related to Musk caught my eye in the crypto world.
Looking at this round of super correction, DOGE is actually still in a relatively slow decline. In other words, although SPCX and Tesla have both dropped significantly recently, DOGE hasn't dropped much.
This is something I find very unusual.
I think either the crypto world hasn't reacted yet, or DOGE has its own advantages, or it's gradually breaking free from Musk's IP constraints. But DOGE's narrative has always followed Ma's memes. Musk's decline in net worth will inevitably and will continue to affect DOGE's future trend.
So around 2 a.m. last night, I got up and opened a short DOGE position, filling up the little funds I had left.
I think this round of correction is #doge will not be absent, only be late.
#长鑫科技上市, global storage competition adds variables #美联储周四凌晨公布利率决议 $DOGE 美股7.27（周一）总结：美伊谈判预期升温推动油价暴跌，但英伟达循环融资争议重创芯片板，资金从高波动算力硬件切换至稳健软件与消费科技龙头。
一、市场核心逻辑
1、地缘风险阶段性缓和，油价大幅下行带来短期利好，但难以对冲科技板块负面情绪
特朗普宣布暂停对伊朗空袭，谈判窗口打开，WTI原油暴跌超8%，创两月最大单日跌幅。油价回落压低通胀预期，10年期美债收益率小幅下行。但利好被半导体板块深度调整完全抵消，指数呈现严重分化格局。
2、AI循环融资叙事发酵，市场重新审视算力需求真实性
据称英伟达为OpenAI提供大2500亿融资担保，市场担忧大量订单依靠表外循环融资催生，并非纯粹终端真实需求；英伟达CDS价差创历史最大单日涨幅，信用风险担忧升温。
3、半导体多重利空共振，中国替代预期压制
国内EUV技术研发持续推进的长期预期，持续挑战阿斯麦独家垄断壁垒。市场交易远期假设：成熟制程DUV率先实现替代，中长期国内持续攻坚EUV先进光刻技术，自上而下压缩海外设备龙头增长空间。
4、资金风格切换明显：硬件承压，软件逆势走强
资金规避高资本开支、需求存疑的AI硬件赛道，转向现金流稳定、商业化落地清晰的软件企业，Palantir大涨印证这条主线；苹果凭借终端AI预期走出独立行情。
5、超级事件周临近，资金提前避险
美联储议息会议不确定性极高，加息概率34%-38%；叠加微软、亚马逊、Meta、苹果集中披露财报，交易员主动降低半导体等高风险仓位。
二、科技巨头
苹果 +1.17%：资金避险首选，折叠iPhone、新一代Mac芯片预期持续发酵，市值逼近5万亿美元。
微软 +1.94%：等待财报资本开支指引，相对算力硬件韧性更强。
谷歌A +2.13%：参与美国高算力AI模型监管框架制定，迎来阶段性催化。
Meta -0.22%：大型数据中心融资成本持续上行，震荡偏弱。
亚马逊 -0.31%：跟随大盘震荡，缺乏新增催化。
英伟达 -4.99%：循环融资争议发酵，CDS飙升；伯里加仓空头，算力龙头遭遇集中抛售。
特斯拉 -1.22%：伯里继续维持空头仓位，市场担忧毛利率持续承压。
SpaceX -1.36%：星链用户增速放缓，星舰商业化一再延后，持续大额亏损压制估值。
三、半导体
费城半导体指数大跌2.23%，板块全线走弱，存储、设备跌幅靠前：
1）存储
美光科技 -2.25%：伯里新增空头仓位，国产存储竞争压力持续存在，高位持续消化估值。
闪迪 -11.02%：AI推理存储架构不确定性叠加国产NAND追赶，延续深度回调。
SK海力士 -7.47%：股价跌破IPO发行价，即便和英伟达达成算力合作，仍难扭转资金出逃趋势。
2）算力
AMD -5.17%：多家投行上调目标价，看好Helios平台长期潜力，但短期受板块情绪拖累大幅调整。
台积电ADR -1.03%：下游云厂商资本开支预期摇摆，股价震荡下行。
3）设备
阿斯麦 -5.80%：国内浸没式DUV光刻机落地消息冲击情绪，叠加市场持续关注国内EUV长期研发突破预期，阿斯麦独家垄断远期逻辑松动#长鑫科技上市，全球存储竞争添变量 $MU 长鑫科技上市，全球DRAM竞争格局正式进入“三强时代”
全球存储产业，迎来了过去十年来最大的变量
7月27日，国产DRAM龙头长鑫科技正式登陆科创板，上市首日市值突破3.3万亿元。这不仅是今年A股最受关注的IPO之一，更意味着中国存储产业正式进入全球资本市场视野，也让长期由三星电子、SK海力士和美光主导的DRAM市场迎来新的竞争者。
过去两年，AI服务器需求爆发推动HBM和高端DRAM价格持续上涨，三星、SK海力士凭借技术优势几乎包揽全球AI存储红利。就在长鑫上市前，Anthropic分别与三星、SK海力士签订存储供应协议，英伟达也加码投资韩国AI生态，市场一度认为全球AI存储产业链将进一步向韩国集中。
而长鑫的上市，意味着这一格局开始出现变化。
对于全球客户而言，DRAM供应首次拥有更具规模的第三个选择；对于产业链而言，中国厂商获得资本市场支持后，研发投入和产能扩张能力将进一步增强，未来有望在消费级、服务器及工业级DRAM市场持续提升份额。长期来看，全球存储行业的竞争逻辑也将从过去的“双寡头博弈”，逐步演变为“三强竞争”。
资本市场已经开始对此进行定价。韩国KOSPI指数当天冲高后回落，反映出投资者开始重新评估未来全球存储行业利润分配。随着中国产能持续释放，DRAM价格周期、各大厂商资本开支以及HBM供需平衡，都将成为决定下一轮行业景气度的核心变量。
AI时代带来的存储需求仍在增长，但未来最大的变化，可能不再是谁拥有最多订单，而是谁能够掌握下一轮全球存储产业的话语权。
$BTC $ETH $KAITO
#长鑫科技上市，全球存储竞争添变量 隔夜美光、闪迪冲高回落、最终收跌，外围存储情绪偏弱，或将传导至港股盘面，进而影响北向资金流向。受此联动，兆易创新今日大概率低开。
虽然兆易创新前期已经深度回调，继续深跌的空间有限，但目前仍处于漫长震荡筑底阶段。磨底行情最考验心态、耐心与持仓定力，炒股本身，就是一场漫长的修行。
现阶段最关键的核心不是猜涨跌，而是仓位管理。当下行情不适合重仓博弈，最优策略是节奏化小仓位、分批低吸、慢慢建仓，稳字当头。
再看市场核心变量——长鑫科技。
上市首日超66%超高换手，开盘与收盘价格基本持平，筹码充分交换、分歧充分释放。
可以明确预判：
今日长鑫换手率将大幅回落，成交量至少萎缩一半以上，预估缩至700亿级别；明日成交会进一步衰减。
这意味着，长鑫对全市场、对半导体板块的资金虹吸效应正在快速减弱，对科技、芯片赛道的流动性压制逐步解除，对整个半导体企稳修复形成积极托底作用。
整体来看，科技板块最煎熬的虹吸压力期正在过去，震荡磨底、缓慢修复仍是主节奏，耐心等待趋势反转即可。#长鑫科技上市，全球存储竞争添变量 $SNDK 1. The Event Itself As of July 28, BTC was about $63,680, ETH about $1,886, and SOL about $74, fluctuating within a narrow range throughout the day, but at one point during the day it plunged due to liquidations and a flash crash in Binance US. Technical View: BTC support at 63,500 to 64,000, resistance at 66,500 to 67,000; ETH supports at 1850, strong support at 1800, resistance between 2000 and 2100; SOL supports at 74, resistance ranges from 82 to 88. The market as a whole has entered a 'wait-and-see compression zone' ahead of the Fed's rate decision, with bulls and bears waiting for a remark. 2. Why is this the hottest trend? A week ago, when BTC just surged above 66,000, CryptoQuant warned it could be a bull trap—open interest surged to $23 billion, but spot trading volume remained weak. The rise was driven by short squeezes and derivatives leverage after funding rates turned negative, rather than genuine accumulation. Now, 150,000 liquidators are just a testament to this 'fragility.' This pre-FOMC structure of "spot market cool, futures hot" pattern is exactly what veteran traders are most wary of, and it's also where retail investors are most likely to chase the highs. 3. Extended Analysis: All three signals should be examined simultaneously. (1) Structural divergence: High OI but spot buying lacks determination; once macro negative factors (rising rate hike expectations), leveraged bulls explode first; (2) Capital Divergence: ETH ETFs attracted $1.85 billion last week (the second highest in history), with funds shifting toward ETH, while BTC is relatively losing its edge; (3) Macro Pricing: CME shows a 55.7% probability of a rate hike in September, with a strong dollar advantageLast night, US tech stocks continued to drop valuations, looking quite alarming. Many opened high at three or four points, plunging deep into the waters, and then dropped more than ten points.
This is the case with US stocks. When they fall, they don't hold back, but there is always a limit. A single wave of 20 or 30 points can basically hold their ground. A few that have dropped sharply, like SanDisk, clearly saw major capital stop losses today, with two stop-loss orders close to 500 million at the bottom. However, many tech stocks have stopped hitting new lows, especially the Big Seven companies, which have mostly rebounded from their lows, with some even hitting new highs.
Don't guess the bottom—just walk out of it. At least the Dow will still be bright red
$SNDK $FIL, that idiot, the expansion of hardware giants like Changxin, will instead become a "catalyst" for the explosion of decentralized low-cost computing?
Changxin Technology (and the domestic semiconductor industry chain behind it) addresses the "physical bottleneck of centralized computing power," while decentralized computing solves the "allocation and cost bottleneck of computing resources." In the AI era, the two are not only non-conflicting but highly complementary.
1. Hardware expansion fills the "gap," decentralization solves the "expensive" and "difficult"
Changxin Technology's recent IPO raised tens of billions, with the core goal of expanding DRAM (memory) capacity to alleviate the "memory wall" problem faced by AI servers. However, even with a significant increase in hardware capacity, the AI computing power market still faces two pain points:
Extremely high computing costs: AI large model training and inference require massive GPUs and memory, and centralized cloud providers (such as AWS, Alibaba Cloud) charge extremely high rental prices.
Monopoly and queuing of computing resources: high-end computing power is often locked by a few tech giants through long-term agreements.
Decentralized computing (such as Render, Akash, etc.) aggregates idle computing resources worldwide, providing small and medium AI developers with costs much lower than centralized cloud providers, breaking the monopoly on computing power.
2. The "data flood" in the AI era requires distributed processing
Changxin's expansion means more AI servers will be manufactured, directly promoting the prosperity of AI large models, autonomous driving, and physical AI. The more advanced AI becomes, the more data it generates exponentially.
If all AI computing relies on a few ultra-large centralized data centers, not only will bandwidth costs be unbearable, but there will also be risks of single points of failure and privacy leaks. Decentralized computing processes massive data through distributed networks and is the inevitable path to support the future AI data explosion.
3. The inevitable trend from "centralized" to "cloud-edge-end collaboration"
The memory produced by Changxin will eventually be assembled into various terminal devices. Future AI computing will not be entirely centralized in the cloud but will move toward "cloud-edge-end" collaboration.
Decentralized computing networks can directly schedule some latency-sensitive and privacy-involved computing tasks to edge nodes or personal devices closest to users. This model greatly reduces the pressure on central data centers, and the massive high-performance memory provided by manufacturers like Changxin is the physical foundation supporting the efficient operation of these distributed nodes.
Hardware manufacturers like Changxin Technology are "building engines," thickening and enlarging the computing power foundation for the AI era; decentralized computing is "building intelligent scheduling systems," enabling computing power to be used by society at large more fairly and at lower costs. Changxin's expansion will not eliminate the demand for decentralized computing; instead, the comprehensive prosperity of the AI industry will provide decentralized computing with massive application scenarios and real demand. 英伟达7500亿美元循环融资事件深度点评
先厘清关键事实边界，避免被标题夸张表述带偏：
1. 7500亿美元是潜在谈判总额，并非已经落地的合同：5000亿SK集团合作项目已经官宣推进；OpenAI2500亿算力租赁担保+3500亿芯片采购融资仍处于洽谈阶段，最终能否落地存在巨大不确定性。
2. 市场争议核心：英伟达这套客户融资+产业链绑定模式，美股机构称之为循环交易（circular financing），通俗理解就是：英伟达出钱给下游AI企业，下游企业资金主要用来采购英伟达GPU、算力设备，资金闭环流转。
一、英伟达这套商业模式的底层逻辑（不能简单定义为“左手倒右手造假”）
黄仁勋这套打法本质是产业链厂商金融赋能，全球高端制造业普遍存在：
1. 海外AI初创企业（OpenAI、各类云算力厂商）普遍没有稳定现金流，无法拿到银行大额低息贷款；英伟达作为算力龙头，通过担保、产业投资、供应链金融，帮助客户锁定算力建设资金。
2. 英伟达获取稳定订单，锁定长期AI算力建设份额，挤压AMD、谷歌TPU等竞品空间；韩国SK海力士同步采购英伟达超算、建设数据中心，双方双向产业链配套，属于上下游协同。
3. 产业上行周期里，这套模式会加速算力基建扩张，AI产业链业绩持续放量，美股AI板块估值持续抬升。
二、机构担忧的三大核心风险（也是本轮舆论炒作的根源）
1. 算力需求存在“人为虚增”，订单含金量打折扣
大量算力采购订单，依托英伟达提供的融资信用支撑，并非企业依靠自身经营现金流自主采购。
如果未来大模型商业化落地不及预期，企业AI业务无法盈利，算力租赁价格持续下行，大量AI公司现金流断裂，无法按期偿还债务：
上游GPU需求会快速崩塌，大量订单取消；英伟达账面大量应收账款、产业投资将会形成巨额减值。
当下北美算力租赁价格已经持续走弱，大量GPU利用率不足50%，产业供需已经出现松动苗头。
2. 产业链形成高杠杆闭环，属于典型的顺周期放大风险
英伟达、OpenAI、SK海力士、软银数据中心多方深度捆绑，形成庞大债务链条。
上行阶段：基建扩张、订单源源不断，AI景气度持续强化；
下行阶段：一旦电力成本上涨、AI盈利兑现不及预期，债务链条出现违约，会引发全产业链戴维斯双杀。
查诺斯、Burry等华尔街空头核心论点就是：这套模式对标2008年次贷扩张逻辑，依靠金融加杠杆托举产业需求，一旦拐点出现波动会极大。
3. 业绩持续性的估值矛盾
当前英伟达估值定价，隐含市场对未来5年全球算力持续爆发的极高预期。
如果大量算力订单属于融资催生的基建投资，而非政企、车企、互联网大厂原生数字化刚需，一旦融资窗口收紧（美联储维持高利率、信贷条件收紧），英伟达业绩增速会出现断崖式下滑。
半导体历史规律：算力基建高峰过后，会迎来漫长的库存下行周期。
三、区分两个关键概念：循环融资≠财务造假
很多散户容易混淆两件事：
1. 合法的厂商供应链金融（英伟达当前模式）
车企、工程机械龙头长期存在厂家贴息、客户融资购车设备；产业链配套合作本身合规，英伟达没有虚增营收、没有伪造合同。监管目前没有定性英伟达交易违规，仅关注交易透明度、债务敞口披露。
2. A股经典左手倒右手关联交易造假
实控人控制上下游公司互相买卖，虚构收入利润，属于违法财务舞弊。二者性质完全不同。
英伟达模式最大隐患不是业绩造假，而是产业需求透支、宏观利率下行周期的债务风险。
四、结合A股算力产业链的传导影响
1. 英伟达产业链博弈分歧显著放大
算力板块内部出现巨大分化：
乐观资金：英伟达持续拉动全球AI基建，光模块、HBM、液冷、服务器产业链中长期景气逻辑不变；
谨慎资金：担忧算力需求依靠融资托底，AI建设高峰临近尾声，上游半导体板块迎来估值杀跌。
2. 国内算力基建逻辑将重新定价
国内东数西算、智算中心建设更多依靠国资、地方产业基金，没有英伟达这种大规模循环融资模式。海外算力订单预期回落，会压制国内出海算力厂商、光模块厂商的海外订单预期。
3. 算力板块波动将会显著加剧
英伟达如果后续交易落地不及预期、美国芯片出口管制持续加码，美股AI板块震荡加大，会同步传导至A股AI算力、半导体赛道。
五、A股实操层面的总结研判
1. 不要简单把新闻解读为“英伟达马上崩盘、AI泡沫彻底破裂”
SK海力士合作项目已经落地，全球AI端侧大模型、工业AI落地仍有真实产业需求；算力需求有一部分是真实的，只是当前远期订单大量依靠融资托底，存在明显预期透支。
2. 本轮事件最大启示：警惕高位算力板块远期业绩博弈
军备竞赛进入中后期，大量算力项目依靠杠杆资金推进。一旦AI商业化兑现速度低于市场一致预期，算力上游硬件板块会率先承压。 #长鑫科技上市，全球存储竞争添变量 $NVDA #长鑫科技上市，全球存储竞争添变量 #美联储周四凌晨公布利率决议 SNDK officially entered a sustained downtrend at 21:30 on July 27, with the price continuously falling and almost no decent rebound, showing a one-sided pressured market.
In the early opening phase, the market still maintained some volatility, and many expected the previous oscillating trend to continue. After 21:30, the trend suddenly changed, with sell orders pouring in continuously, the bulls' support quickly weakening, and the price stepping down gradually.
Not only SNDK, but the entire memory chip sector weakened simultaneously, with SK Hynix, Micron, and Western Digital all under pressure, causing sector sentiment to resonate and further amplify selling pressure.
Behind this, market concerns began to ferment: the market is re-evaluating the sustainability of AI computing power capital expenditures, combined with expectations that the memory price increase cycle has peaked, leading many previously accumulated profit positions to start fleeing collectively. Funds are withdrawing from the high-level memory sector, creating a scenario of multiple sellers driving down prices.
Every minor intraday rebound is very weak, followed by a new round of deeper declines. Many holders originally had hopes for short-term recovery but were continuously worn down by the market.
One-sided downtrends test investor psychology the most; do not easily bottom-fish to gamble on rebounds. During the phase of collective sector weakness and continuous capital outflow, counter-trend buying carries extremely high risk.
This ongoing decline also serves as a reminder to everyone: once the trend reverses, do not subjectively fantasize about support; respect the real choices of market funds.昨晚的美股，与其说是“暴跌”，不如说是一次迟到的“清醒”。
费城半导体指数跌超5%，英伟达、SK海力士这些AI时代的“当红炸子鸡”纷纷重挫，闪迪更是跌了近14%。市场给出的理由很直接：中国国产DUV光刻机量产的消息，像一根刺，扎破了之前对于海外巨头垄断地位的绝对自信。虽然几台的产量对比ASML的百台交付量仍是九牛一毛，但资本市场的恐惧往往不在于当下的损失，而在于未来叙事逻辑的崩塌。
更深层的原因，其实是AI烧钱故事讲累了。谷歌自由现金流转负是个警号，大家开始算账：巨额投入到底何时能听见响儿？资金是很诚实的，它们正从高估值的硬件端撤出，流向苹果这种有真实消费支撑的防御板块。道指涨、纳指跌，这种分化说明市场不是在恐慌出逃，而是在进行剧烈的高低切换。#长鑫科技上市，全球存储竞争添变量 $MU During the Bitcoin cycle, the bottom never appears as a "buy signal"—it appears as "no one wants to touch this thing anymore."
2018 bottom: $19,800→ $3,200, down 84%. At that time, some people shouted, "Bitcoin will reset to zero."
2022 bottom: $69,000→ $15,500, down 78%. At the time, someone shouted, "This time is different."
2026: $126,000 →? A decline of about 50%. If history repeats itself, there is still room for decline.
Current market sentiment: Institutional pessimism has reached a two-year high, and retail investors are waiting for lower prices.
If historical patterns hold, bottoms may occur when most people lose confidence in the market. #RWA永续月交易量4700亿美元 $BTC 日韩股市集体跳水！韩股暴跌6.25%濒临熔断，李在镕：我不背锅！
隔夜亚太股市风声鹤唳，风险集中释放。行情数据显示，日经225指数大跌2.73%；韩国综合指数暴跌6.25%，KOSPI200指数跌幅达到6.67%，盘面剧烈动荡，距离熔断仅有一步之遥。
韩国股市接连大幅下挫，网络上有趣的段子开始流传。很多股民看到韩国指数大跌、屡次触发熔断预警，不由自主联想到三星会长李在镕。
因为读音巧合，不少网友戏称“李在镕=李再熔”，调侃只要行情出现大跌，市场就要再度熔断。
不少普通人分不清人物，经常把三星集团会长李在镕，和韩国总统李在明两个人相互混淆。市场一跌，不分青红皂白直接把下跌责任扣在李在镕头上。
倘若李在镕看到各类评论，想必会深表愤慨，忍不住呐喊：我不背锅！
股市暴涨暴跌、存储板块周期轮回，行情起伏和我并无直接关联，“再度熔断”这个谐音梗带来的大锅，我坚决不接！
行情下跌从来不能归咎于某一个人，核心根源在于全球存储产业链景气预期持续恶化。
此前韩国总统李在明大力推动SK海力士落地9300亿美元半导体合作大单，看似重磅利好，却无法阻挡外资持续撤离。SK海力士股价持续走弱，美光、闪迪、康宁等存储、光通信龙头不断创出新低。市场逐渐清醒，AI催生的算力存储需求含有大量水分，英伟达高达7500亿循环融资“左手倒右手”的商业模式饱受质疑，虚假需求泡沫正在缓慢出清。
海外半导体赛道持续杀跌，日韩股市同步承压；反观A股，长鑫科技上市一日催生2.7万亿账面新增市值，内外行情走势形成巨大反差。
在没有增量资金入场的存量博弈市场，盈亏同源是不变的规律。海外市场提前反应产业周期下行风险，也给国内火热的存储赛道敲响警钟。
日韩股市恐慌情绪蔓延，短期或将扰动A股科技板块情绪。国产替代拥有长期逻辑，但短期情绪炒作催生的估值泡沫，依然需要警惕估值回归的风险。
免责声明：本文仅为盘面行情客观解读，不构成任何投资建议。全球股市波动风险极大，任何交易请理性决策。#长鑫科技上市，全球存储竞争添变量 $SKHYNIX $FIL This dumb fil, yesterday Changxin's IPO blew up the three storage giants. If Changxin's production capacity increases exponentially, what impact will it have on fil coin?
1. The core advantages of FIL (Filecoin) decentralized storage
FIL, as the native token of the decentralized storage network Filecoin, has its core advantage in solving the pain points of traditional centralized storage (such as AWS, Alibaba Cloud, etc.) through cryptographic verification and economic incentive mechanisms. Its main advantages include:
Data security and tamper resistance: Using the IPFS (InterPlanetary File System) protocol, data is located through "content hash (CID)", split into small pieces stored on different nodes, avoiding single points of failure. At the same time, it supports data version control, permanently retaining historical versions, ensuring extremely high data reliability.
Innovative consensus and verification mechanism: It adopts the PoST (Proof of Spacetime) consensus mechanism instead of traditional computing power competition. Nodes must regularly generate "Proof of Replication (PoRep)" and "Proof of Spacetime (PoSt)" to prove that data is truly and continuously stored on local hard drives. If data is deleted or tampered with, valid proofs cannot be generated and staked tokens will be penalized, thus ensuring data integrity.
Low cost and resource optimization: Through a market-based pricing mechanism, it integrates global idle storage resources (such as personal computer hard drives, enterprise backup servers) into a distributed storage network. Its storage cost is far lower than traditional centralized cloud storage, and node energy consumption is extremely low, aligning with carbon neutrality trends.
Programmability and ecosystem expansion: By introducing the Filecoin Virtual Machine (FVM), developers can directly deploy smart contracts on the storage layer, supporting cross-chain interoperability. This makes FIL not just storage infrastructure but also supports complex applications like AI data markets and Data DAOs, transforming data from static assets into programmable resources.
2. Analysis of the potential impact of Changxin Technology's capacity increase on FIL
Sun Yuchen's "always lacking storage" assertion is based on the core logic that in the AI era, data generation speed will always exceed storage technology development speed, making storage a key bottleneck determining AI performance. As a leading domestic DRAM storage chip manufacturer, Changxin Technology's IPO and significant capacity increase impact FIL from two dimensions: demand scenario differences and macro industry resonance.
1. Fundamental differences in demand scenarios (weak substitutability)
Changxin Technology produces DRAM (Dynamic Random Access Memory), mainly serving as the "computing power engine" and high-speed memory for AI servers, high-performance computing, and consumer electronics, solving data "processing speed" and "bandwidth" bottlenecks. FIL provides decentralized, low-cost "massive cold data" long-term storage and backup services. Therefore, Changxin's capacity increase mainly meets the rigid demand for high-speed video memory/memory in AI training and inference stages and does not directly compete with or substitute FIL's low-cost distributed storage market.
2. Macro-level positive resonance (data overflow effect)
Changxin Technology's exponential capacity increase will macroscopically validate and strengthen the "always lacking storage" industry logic, indirectly benefiting FIL:
Improvement of AI infrastructure: Hardware manufacturers like Changxin solve AI computing power and high-speed memory bottlenecks, greatly promoting the explosion of AI large models, intelligent agents (AI Agent), and physical AI. The AI boom will exponentially generate massive data, and when this massive data exceeds the capacity of expensive high-speed storage, it will inevitably overflow to low-cost decentralized storage networks (such as FIL).
Strengthening of market consensus: Changxin Technology's market value topped the A-share market on its IPO day, marking high capital market recognition of the logic that "storage is the core asset of the AI era." This industry-level prosperity will increase market attention and valuation premiums for the entire storage sector (including decentralized storage).
Summary: Changxin Technology's capacity increase will not pose a bearish or competitive threat to FIL. On the contrary, it fills the physical infrastructure gap for AI, accelerates the arrival of the AI era, and creates a larger underlying data demand and broader application scenarios for FIL's decentralized massive data storage market. The two play complementary rather than mutually exclusive roles in the AI era storage ecosystem. AI chip crashes—while others watch the spectacle, we see the tricks
US AI chips plunged across the board, with leading company Nvidia plunging from a pre-market gain of 2.8% to a drop of over 7% during trading, while AMD, TSMC, and Broadcom all took the hit.
The trigger is ironic: Chinese AI chip maker Moore Threads surged 420% on its first day of listing on the STAR Market, but the market changed its tone in an instant—new capacity is coming, will the logic of computing power price hikes be undermined? Panic was triggered.
But the real reason for such a sharp drop lies underneath: Nvidia has risen about 480% this year, its holdings have loosened early, the narrative cracked, and profit-taking positions have fled.
Does this script look familiar? It shares the same core as the high-end crypto sector: high beta built on narrative and capital. When prices rise, supply + sentiment + profit-taking come together, making it faster than anyone else to catch up on declines.
A fourfold increase isn't a safety cushion—it's a hard-hit zone—it's all floating profits trying to escape
#长鑫科技上市, global storage competition adds new variables
$BTC
$ETH
$SNDK Everyone believed that last night's storage crash was caused by Changxin Technology's IPO draining capital?
Let me show you the real reasons behind the drop!
Confused? Is the DeepSeek moment for lithography machines here?
Why were five domestic machines able to make ASML disappear overnight?
Before today's opening, chip stocks were originally opening higher. With two positive factors in mind, the Middle East situation eased, and the Wall Street Journal reported that Nvidia was negotiating to secure $250 billion in financing for OpenAI's data centers. At that time, ASML was still up 2%.
Then a message from The Information appears, and the whole sector turns down.
The news is as follows: a company with Chinese state-owned capital background has begun mass production of self-developed immersion DUV lithography machines, planning to produce 5 units this year and expand to around 20 units by 2027, with the first batch delivered to SMIC, Huahong, and Changxin.
ASML immediately turned negative from a 2% gain, dropping as much as 6% during trading, hitting its lowest level since June. All US storage and equipment stocks were dragged down.
Today, Twitter gave this message a name: the DeepSeek moment of the lithography machine.
After looking for a long time, the first thing I want to say is: first look at the numbers clearly...
Five this year. ASML delivered 131 units last year.
Putting these two numbers together, the difference is not just a little, but an order of magnitude more.
Moreover, The Information's original article clearly states that these domestic machines still lag behind ASML in performance and reliability, requiring months of testing before entering production, and some key components still have to be imported from Japan. As for the more advanced EUV, China is still stuck at the prototype stage and is several years away from producing usable chips.
So if you say these five machines touched ASML's cheese today, it didn't. ASML's machines are still scarce, and SK Hynix's HBM long-term contract should be signed in 2028 or 2028.
So what is the market afraid of?Friends of all planets, after waiting for a whole month, today has arrived. The Federal Reserve's FOMC policy meeting officially kicks off today (July 28). For two consecutive days, interest rate decisions will be announced at 2 a.m. Beijing time tomorrow (July 29), and the new chair, Kevin Warsh, will hold a press conference at 2:30 a.m. This is what I've always called the 'real flipping moment'—tonight is the eve, and the whole audience held their breath. First, let's talk about the counterintuitive key point: many beginners think the market is just waiting for a rate cut to save the situation. Wrong. If you look closely at the betting data, you'll find that no one in the market actually expects a rate cut—the real disagreement is whether to keep things the same or to raise rates. According to CME FedWatch, the probability of holding rates unchanged is as high as 82% to 93%, while the remaining 25% to 30% are betting on rate hikes, not rate cuts. In other words, for risk assets, the best outcome of this meeting is simply to 'maintain the status quo'; there's no option for 'massive liquidation.' You really need to keep this background in mind to understand today's market trends. Then why are you still so nervous? Because the style of meetings is never just about the numbers, but about whether the attitude behind 'maintaining the same' is whether the dove is the hawk or the hawk. The same phrase 'hold stead' can be interpreted in three ways: if Warsh's tone is dovish and implies inflation is under control, the market will see it as good news and risk assets can breathe a sigh of relief and rebound; If he stands firm, emphasizes that 'high inflation will not be tolerated' (he actually said that recently), and even some committee members vote to raise interest rates, then so be itAEON 这次上 OKX，我觉得比普通新币开盘更值得拆开看。 7 月 27 日，OKX 连着放了几条 AEON 相关信息：现货 AEON/USDT 开放交易，USDT 本位永续跟上，OKX Wallet Boost 也把 AEON 放进 X Launch，给出 400 万枚 AEON 的活动奖励。单看 K 线，这就是新币上线；但放到 Web3 支付和 AI Agent 这条线里，它更像是一次“能不能把叙事搬到真实支付场景”的压力测试。 我不会把新币上线直接等同于基本面兑现。币圈老玩家都知道，第一天最吵的地方往往不是项目本身，而是流动性、空投预期、开盘筹码和合约杠杆。OKX 公告里给得很清楚：AEON 现货在 7 月 27 日 11:00 UTC 开放，提现在 13:00 UTC 开放；永续在 11:30 UTC 开放，最高 20 倍杠杆，资金费率按 4 小时结算。也就是说，现货、合约、钱包活动几乎在同一天把热度推起来了，这种节奏短线一定会放大波动。 但 AEON 的看点不只在“新”。OKX 对项目的介绍里，AEON 被定义为面向 AI Agent 和真实商业场景的加密支付基础设施，提到Garlinghouse clearly stated that the CLARITY Act is the final barrier to institutional-level adoption of XRP. 🚧 If the bill passes smoothly, XRP will gain legal certainty, attracting a large influx of institutional funds and potentially significantly enhancing market stability. 📈 But don't be naive—clear regulation has always been a double-edged sword. ⚔️ It brings not only compliance bonuses but also stricter scrutiny and more intense regulatory scrutiny—which will also deter some speculators. 😏 So the question arises: once the rules are implemented, will XRP's volatility converge into stablecoin-like sideways movement, or will a new wave of volatility erupt due to liquidity structure restructuring? 📉📊 This is the core variable truly worth discussing.美股半导体集体崩了！国产DUV落地长鑫，全球存储格局真的要变天！
先看昨夜美股半导体惨烈行情：
• 存储板块领跌：闪迪暴跌13.79%；SK海力士大跌9.39%，创出上市新低；美光科技下跌6.94%，市值跌破万亿关口，即便手握9500亿美元长期订单，依旧难以稳住市场信心。
• 半导体设备同步重挫：阿斯麦大跌7.3%，泛林集团跌7.45%，资金重新评估其在中国市场长期增长空间。
• 全赛道普跌：AMD狂泻8.51%，英特尔跌5.38%，英伟达下跌4.03%，盘中失守关键支撑位。
引发全球芯片市场震荡的核心导火索：外媒The Information爆料，刚刚登顶A股市值榜首的长鑫存储，敲定首批国产浸没式DUV光刻机交付名额，今年8月起陆续收货。
消息梳理：上海国资背景宇量昇科技，正式进入浸没式DUV小批量制造阶段。2026年计划产出5台，2027年扩产至20台，首批设备定向供给长鑫存储、中芯国际、华虹半导体，年内启动交付上机验证。
设备技术定位
这款国产机型采用193nm ArF浸没式架构，原生适配28nm制程，依托DTCO+多重曝光技术可延伸覆盖14~17nm成熟工艺，完美匹配DRAM制造需求。设备大部分核心零部件实现国产化，仅少数组件依赖日本供应，大幅对冲美国《MATCH法案》出口管制风险。
过去国内成熟制程扩产高度依赖ASML浸没式DUV，随时面临供货、售后限制；国产DUV实现从0到1突破，相当于给国内晶圆厂搭建供应链安全缓冲，制裁的威慑力持续下降。
最大受益标的：长鑫存储
长鑫上市首日大涨465.82%，总市值3.28万亿登顶A股，募资足额到位，正全力推进产能扩张。
产能规划：当前12英寸晶圆月产能接近30万片，年底冲刺35万片；合肥、上海两座新厂区建设中，远期目标月产60万片，2030年产能有望追赶美光。
在此之前，扩产节奏受制于海外设备交付周期；国产DUV落地后，产能扩张主动权显著提升。
更关键的战略布局：长鑫选择3D DRAM换道超车路线
EUV设备获取受阻之后，长鑫研发重心从平面制程微缩，转向垂直堆叠架构。
目前依托G5节点量产1a DRAM，自研垂直全环绕栅极（GAA）3D DRAM单元结构，有效改善Row-hammer干扰难题；合肥试验线推进W2W晶圆混合键合技术，实现存储单元与逻辑电路分层制造、垂直整合，同时联合高通、兆易创新开发面向端侧NPU的定制化3D DRAM。
此前研发进度受制于设备供给节奏，国产DUV批量上机，将加速技术试产迭代。
为什么仅仅5台样机，就能引发海外半导体集体杀跌？
本质是三大长期预期被彻底改写：
1、预期差被击穿。海外机构普遍预判国产浸没式DUV规模化交付至少还要2~3年，落地时间大幅提前。资本市场交易远期空间，自主可控拐点确认，直接引发估值下修。
2、阿斯麦核心基本盘承压。浸没式DUV是阿斯麦最核心的现金牛业务，中国市场占据重要份额。国产设备打通技术路线，即便短期体量有限，但长期会持续挤压其在华增长天花板。
3、全球存储周期格局重塑。DRAM属于典型周期行业，供给决定价格与盈利。长期由三星、SK海力士、美光三家垄断产能定价；长鑫获得自主设备保障，成为独立新增产能变量，海外巨头的定价权、周期盈利韧性遭遇长期挑战，这也是存储板块跌幅最深的核心原因。
客观理性看待现状
现阶段国产DUV尚处在小批量样机验证周期，全年仅5台产能，套刻精度、连续稼动稳定性、量产良率对比阿斯麦成熟机型仍存在明显差距，距离大规模商业化替代还有漫长周期。
半导体产业规律向来如此：从0到1最难突破，完成技术打通后，从1到10只是时间问题。一旦产线验证顺利，后续迭代速度大概率超出市场预期。
⚠️风险提示：资讯来源于海外媒体报道，尚未有企业官方公告；设备量产、上机验证存在不确定性。内容仅产业逻辑交流，不构成任何投资交易建议。#长鑫科技上市，全球存储竞争添变量 $SNDK Just saw the news saying Trump supports Wash cutting rates and says the US rate should be the lowest in the world, and criticizes the Fed's "cut blockers"
I think this is more like a game strategy, like a house without windows—if you ask to open a window, you might be refused, but if you propose to tear down the entire door, people will start discussing whether you should consider opening one window first.
Trump may be well aware that pushing for a Fed rate cut in the short term is unrealistic, but by continuing to apply pressure, he can at least influence market expectations and reduce the likelihood of further rate hikes, thus achieving his goal.Reviewing regulatory developments from July 27: New York State Attorney General Letitia James submitted written testimony to the Senate Permanent Inquiry Panel, calling for stronger regulation of the crypto market and arguing that the existing design of the CLARITY Act could weaken state and local enforcement capabilities against fraud cases.
Draw the boundaries first: this is a regulatory opinion, not a law that has already been passed or come into effect. Independent media outlet The Block also confirmed that testimony was submitted to the relevant Senate panel on Monday. The final text of the bill, voting arrangements, and specific jurisdictional provisions will still be subject to parliamentary procedures.
Why is "regulatory division of labor" more important than the headline?
If digital assets are mainly covered by federal rules and state-level powers are largely excluded, changes will not only affect which regulator the platform faces but will also spread along three links:
1. After the user or victim submits a clue to the state-level agency, can local authorities investigate promptly?
2. Whether state and federal agencies can handle cross-state platforms, advertising, or payment chains in parallel;
3. What does the platform need to comply with is a set of unified rules, or must it still bear state-level consumer protection obligations beyond the federal bottom line?
James's office stated that complaints about crypto scams in New York State tripled over three years, with losses approaching $500 million in the past five years; These are the statistical calibers cited by the office in its testimony and should not be extrapolated to the total loss across the United States.
Next, observe three things: whether the bill clearly retains the scope of state-level enforcement; Whether amendments will appear during Senate review; and how regulators explain the collaborative processes for complaints, investigations, and transfers.
Do you think the most important thing in market structure legislation is to clearly define unified federal standards, or to define the boundaries between states and the federal government handling fraud cases simultaneously?Target in range: Microsoft, Amazon, and Meta will reveal their capital expenditure cards on Wednesday and Thursday — a key wind drift parameter in ballistic calculations. Last week, Alphabet was shot down by the market due to increased ammunition (Capex), and Tesla recorded its largest single-week drop since 2022. All indicators point to the same trajectory: will the spending guidance from the three cloud giants freeze the market's nerves or further fuel anxiety?
I have locked onto three lines in the scope: cloud business growth rate is the wind speed vector, AI monetization efficiency is the ballistic coefficient, and capital expenditure intensity is the bullet weight. All three are indispensable. If growth cannot cover spending, it's like a bullet veering off target due to crosswind after leaving the barrel — you must preemptively adjust the trigger. But so far, no bullet has truly hit the bullseye. The market has only learned a lesson from Alphabet's severe drop: a charge without cover will be counterattacked.
Now, tokenized US stocks on OKX provide a 24-hour shooting window — quotes for XMSFT, XMETA, and XAMZN closely track the latest closing prices, denominated in USDT, like thermal imaging in night vision. But remember: snipers don't shoot targets without a positive risk-reward ratio. These derivatives are not bullets but reticle markings in the scope. The real trigger pull timing must wait for the capital expenditure reports from the three giants, combined with relay data confirming cloud growth and AI monetization to pinpoint the impact.
Lie in wait. Wait. Count breaths. Wind speed shifts from 3.5 knots to 4.1 knots — Alphabet's lesson is etched on the ruler: expensive spending without matching revenue feedback is self-destructive trajectory. And once Microsoft's Azure growth exceeds expectations, Meta's AI ad conversion rate rises, and Amazon's AWS profit margin stabilizes, that will be the triple-overlap kill window.
There is only one bullet in the magazine. Without a perfect risk-reward ratio, never pull the trigger. #AIEarningsWatch Betting on South Korea's AI infrastructure! Nvidia has invested $1 billion in equity, becoming NAVER's third-largest shareholder
Nvidia's move into South Korea's AI infrastructure landscape is moving from a promise of cooperation at Jensen Huang's table to a real equity lockdown. On July 27, NAVER announced it had signed an investment contract with NVIDIA through a third-party private placement, with NVIDIA acquiring a 4.5% stake in NAVER for $1 billion (about 1.48 trillion KRW), becoming its third-largest shareholder. At the same time, NAVER has signed a preparatory contract with global asset management firm Bofeng for up to $9 billion in AI factory financing for the expansion of its data center in Sejong City. This marks Nvidia's first time entering a local internet platform as a strategic shareholder in Asia's core AI market—the computing power giant is no longer satisfied with just chip delivery, but is directly embedding its balance sheet into the expansion trajectory of downstream AI infrastructure. From GPU supply to 4.5% ownership: NVIDIA's Korean ecosystem "anchoring" is typically used when introducing strategic partners or quickly injecting capital through targeted placements to third parties. NAVER chose this approach to accept Nvidia, aiming to upgrade the two parties from a computing power buyer to an equity community. On June 8, Jensen Huang visited NAVER 1784's headquarters in Seongnam, Gyeonggi Province, and met with NAVER Chairman Lee Hae-jin. Less than two months later, this $1 billion strategic investment was realized, demonstrating Nvidia's urgent need to accelerate its transformation from "shovel seller" to "partner" in South Korea's AI ecosystem. Bundled cancellation of trillion won in proprietary shares: a two-way signal for controlling interest. While accepting NVIDIA as the third-largest shareholder, NAVER announced it will cancel 4.9 million shares worth about 1 trillion won worth of its own shares on the 3rd of next month to enhance shareholder value. This move, amid concerns over equity dilution that the private placement may spark dilution, sends a signal to the market that management still firmly controls the pace of capital. For NVIDIA, canceling its own shares means a reduction in NAVER's total share capital, which in the long run helps improve earnings per share and asset quality, indirectly increasing the value of its 4.5% stake. Bofeng's $9 Billion Preliminary Contract: The Key Piece of AI Factory Financing NAVER's preliminary contract with Bofeng sets a maximum external financing cap of $9 billion for AI factory construction projects. According to the financing structure, Bofeng will provide a maximum amount for project costs, with the remainder to be raised by NAVER itself. Currently, NAVER's direct investment amount has not yet been finalized. This funding is expected to be fully invested in the expansion of the AI factory at the Sejong City data center "Gak Sejong." Against the backdrop of exponentially growing demands for large model training and inference, the computing power density and scale of data centers have become the core competitive barriers of AI infrastructure. Bofeng's entry has introduced key external capital leverage for this heavy-asset project like NAVER. The Korean pivot of the global AI infrastructure plan: As the operator of NVIDIA's "Global AI Infrastructure Initiative," Bofeng has directly incorporated NAVER's AI factory expansion project into its global investment portfolio. This is not an isolated case—Nvidia has been active in South Korea recently: establishing an AI research center with KAIST, signing a $1.5 billion chip packaging agreement with Amkor, and with NAVER's dual advancement of equity and infrastructure financing, South Korea is becoming one of the densest regions in Nvidia's AI infrastructure layout in Asia. For the capital market, NAVER's AI transformation has received dual endorsement from computing giants and top asset management companies. Whether NVIDIA can replicate the Korean model in other regions through this combination of "equity investment + infrastructure financing" will be a key dimension for observing its ecosystem expansion pace going forward. $NVDA I guess many people didn't understand?
One piece of news. A Chinese state-owned background company has mass-produced a self-developed DUV lithography machine. 5 units this year. 20 units next year.
Then, ASML dropped 6% intraday today and triggered a trading halt.
5 units. ASML delivered 131 units last year.
With just these 5 units, SanDisk fell 13%, Hynix fell 8.6%, Micron fell 6.6%. The semiconductor sector was dragged down by a big bearish candle, even though it opened higher pre-market.
Many people don't understand. What can 5 units do?
The difference is more than an order of magnitude, performance is still behind, parts still need to be imported. Isn't this just a toy?
No.
The market never prices for the "now." The market prices for "possibility."
ASML is worth that money not because it can sell 131 machines a year. It's because it's the only one in the world that can make them. This "only" is the most valuable part of its valuation.
Between 0 and 1 lies the entire Pacific Ocean. Between 1 and 100 lies only time and money. Do you think China lacks these two things?
When DeepSeek came out, the market panicked too. Everyone said it was far behind, a toy, not promising. Half a year later, who dares say that now?
The same script. The same kind of panic. The same group of people making the same mistake: mistaking "gap" for "safety."
Gap is not safety. Direction is.
Once direction is confirmed, gap is just a countdown.
Storage stocks were hit even harder than equipment stocks today, and there's something many people didn't see. The high gross margin in storage over the past two years isn't just from the AI demand boom.
There's also a hidden factor: China can't expand production. ChangXin can't buy ASML machines, the capacity ceiling is locked, global DRAM supply is tight, and pricing power is in the hands of Samsung and Hynix.
Today, someone is starting to make the key for this lock.
Nothing will change in the short term. 5 machines can't enter anyone's financial model. The money to be made this year will still be made.
But the valuation model for three to five years from now needs rewriting. Storage has been treated as a growth stock these past two years, priced for growth. Today the market reminds everyone that at its core, it's a cyclical stock. The one thing cyclical stocks fear most is: someone learning how to make it.
I have storage-related positions. I didn't move a single share today. The logic supported by AI demand hasn't changed, so I won't move.
But I have a line in my mind that used to say "China's advanced process ceiling is physical." Today I crossed out "physical" and changed it to "engineering."
Engineering problems, this country has never backed down from.
$MU
$SNDK
$SKHYNIX 我见过一个哥们，2021年1.2万买的ETH。 跌到8000的时候，他说"我再看看"。跌到6000，他割了。 原因很简单：他受不了账面上每天少好几万块钱的感觉。 然后ETH在4000多待了一阵，反弹到9000，他又追进去了。 这一来一回，他亏了将近40%。而如果他只是什么都不做，他的持仓到现在是赚钱的。 这个故事在加密圈每天都在发生。 止损和"止盈"是两回事 大多数人嘴里的"止损"，其实是一个固定数字。 "跌破5万我就卖。""到了-20%我就跑。" 听起来很理性，对吧？有纪律，有原则。 但你仔细想想——你设这个数字的时候，依据是什么？ 大概率是你亏得起多少，或者你心理上能承受多少。 这跟市场本身没有任何关系。 你设的-20%，不是因为资产在-20%的时候基本面变了，而是因为你的心脏到了极限。 这不是止损，这是"止你心理上的痛"。 我干了十年财经记者，见过08年雷曼倒闭那天华尔街交易员脸上的表情，也见过2022年LUNA归零时社群里的哀嚎。 我发现一个规律：真正亏大钱的人，不是没止损的人，是反复"止损"的人。 他们像洗衣机里的衣服一样被甩来甩去——跌了割，涨了追，再跌再割。 每一次都觉得The US stock market's storage sector has started a crash 😅
Changxin is like a whale rising, everything else falls 😁
I've said many times, the main narrative of this storage cycle downturn is China's production capacity dominating the world
Whether it's SK Hynix or Micron, they have no chance to compete against China's production capacity
China will definitely turn the storage industry into a dead scene like photovoltaics 😁
P.S. The narrative for the next storage cycle might be storage-computing integration
$MU
$SNDK
$SKHYNIX (1) SanDisk (SNDK) Daily Market On Monday (July 27), SanDisk closed down 11.02%, closing at $1,278.23 per share. The transaction volume reached $27.071 billion, with a turnover rate of 14.13%. The stock has already fallen more than 45% from its peak at the end of June. Background of the Decline: SanDisk's sharp drop was a microcosm of Monday's US stock "chip storm." The Philadelphia Semiconductor Index once fell nearly 5%, ultimately closing down 2.23%. The memory chip sector was the main force leading the decline, with SanDisk, Kioxia ADR, SK Hynix, and others all suffering heavy losses. Despite the continued ceasefire between the US and Iran and the sharp drop in international oil prices, the semiconductor sector still faced collective sell-offs. Core Drivers The core reasons for the collective decline in the semiconductor sector include: Nvidia was sold off due to concerns over "circular financing" and news of "plans to provide a $250 billion guarantee to OpenAI," closing down 4.99% and dragging down the entire semiconductor sector; Investors continue to question the level of AI spending; Semiconductor stocks that had surged earlier are now facing profit-taking pressure. Short-term Outlook: SanDisk's short-term performance still heavily depends on overall sentiment in the semiconductor sector and the upcoming earnings season performance of tech giants. Given that SanDisk has already pulled back significantly, attention should be paid to whether there are opportunities for an oversold rebound, but caution is needed regarding overall sector downside risks. (2) SpaceX (SPCX) Daily Market Trends SpaceX briefly fell to $108.66 intraday on Monday, hitting a new low since its listing. By the close, it was quoted at around $110-112, down about 3-4%苹果vs美光：中国芯片让特朗普陷入两难
苹果想用中国芯片降成本，美光全力阻击怕丢单——特朗普后院真着火了。
苹果的算盘很清楚：成本持续上涨，中国存储芯片性价比高，为什么不用？库克和特朗普关系一直不错，苹果也是特朗普最想留住的科技巨头，但这次可能要让老特头疼了。
美光的焦虑也很现实：作为美国唯一的大型存储芯片公司，一直是半导体政策和制造业回流的最大受益者。如果苹果开了先例用中国存储，不仅自己丢订单，更可怕的是帮中国存储打开全球市场——这是战略性的失守。
特朗普现在左右为难：支持苹果，得罪美光和半导体鹰派；支持美光，苹果成本压力无解，还可能把供应链逼得更往外走。
这不是简单的商业纠纷，是美国科技政策内部矛盾的集中爆发：一边要遏制中国，一边要保持企业竞争力和利润。长期看，这种割裂只会加速供应链重构，中国存储厂商的机会窗口可能比想象中来得更快。$MU $AAPL $BTC #美联储周四凌晨公布利率决议 #财报观察员：微软Meta亚马逊能稳住AI叙事吗？ #长鑫科技上市，全球存储竞争添变量 Bottom-fishing or cutting losses? My short position has a floating loss of 1957 USD
Thinking bottom-fishing was an opportunity? My short position is already at a floating loss of 1957 USD. Only when the child is asleep would I dare to secretly watch the market.
The 24-hour trading volume more than doubled, but the price dropped 2.5%. Dog dealers were both selling and buying, and retail investors chasing gains were all buried. Guess how many people are still shouting about the "Golden Pit"?
💔 Market sentiment: Hidden in panic is greed at this level. With no one in the group calling for orders anymore, the screen is filled with phrases like "can't hold on" and "can we still break even?"
But strangely, the funding rate was still positive at +0.0025%, indicating that many bulls were holding their ground and never admitted defeat. Fear and greed are fighting, and at times like this, it's easiest to lose direction—either a rally with a long rally, or keep selling the bulls.
📉 The data speaks for itself: oversold ≠ bottomed out and RSI dropped to 25.22, clearly in the oversold range. Logically, a rebound should be expected, right?
But looking at the moving averages, both MA7 and MA30 are holding down above the heads (64606 and 64987), so prices are lying below, so bearish alignment is reasonable. Trading volume surged by 148%, indicating that selling pressure hasn't eased but is accelerating.
I've seen too many scenarios of oversold followed by continued declines—no matter how low the RSI is, if there's no buying interest, it's pointless. Think about it in reverse: if everyone thinks oversold means bottom-fishing, then why should the main players help you carry the sieve?
Even more ironically, BTC balances on exchanges are still declining, long-term holders are hoarding coins, but short-term prices are falling. What does this indicate?
The shakeout isn't over yet; the chives are cutting leverage, not spot stocks. If a reversal really happens, at least wait until the volume ramps up and it returns to MA7.
The key isn't the indicators, but who admits defeat first. Guess how much longer the bulls can hold out?
🔍 My strategy: Hold onto short positions and hold on, but keep one lot with a floating loss of 1957 USD, an 11.66% loss. Honestly, it's a bit painful.
But with 48x leverage, the position remains within a controllable range, with stop-losses set above 65,750. I don't recommend anyone take on the same deal as I do. Going against human nature trading sounds easy, but doing it is harder than climbing to the sky.
At this level, chasing short sellers isn't cost-effective either, because the oversold zone could trigger a spike at any time. Here's a suggestion: either wait for a pullback to the first support at 63,610 and confirm support before considering going long, or wait for the rebound to close to 64,700 to add a short position.
Don't make decisions when your emotions are at their most panicked. $BTC
#美军暂停对伊空袭, international oil prices opened sharply lower 🩸 A Monster Red Daily Candle Sends Bitcoin Back Under $64K
The reclaim didn't last a day. Bitcoin opened at $65,375, got rejected, and dumped all the way to $63,567 before settling at $63,682. Down 2.59% on heavy volume, one of the largest red daily candles this month. Yesterday's $65K reclaim is already gone.
📉 Where it stands:
Price: ~$63,682 (down 2.59%)
Daily range: $65,722 to $63,567
Lost: the $65K reclaim and the $63,800 line
Next support: $63,281
On the 1D, this is a failed reclaim, and those matter. Price pushed above a level that rejected it twice, held for a single session, then got sold hard. That pattern traps late buyers and usually means sellers were waiting above rather than exhausted. BTC is now back below $63,800, the line analysts flagged as the boundary between a recovering market and a continuing downtrend, and it sits directly on $63,281 support.
The timing explains a lot. The Fed announces tomorrow, July 29. Traders de-risk before binary events, and heavy volume on a red candle the day before a rate decision is often positioning, not conviction. Rising bond yields have quietly lifted hike odds, and stablecoin inflows sit at a multi-year low, meaning thin buying power underneath. Options are clustered at $65K to $70K, so the reaction tomorrow could be violent.
What to watch:
Hold $63,281 and reclaim $65K after the Fed, and this was a pre-event shakeout.
Lose $63,281, and $62,000 then $60,000 open up.
A failed reclaim before a Fed decision is not the moment to force a position. Sit on your hands, let Wednesday resolve it, then act.
Pre-Fed shakeout, or the start of another leg down?
Not financial advice. $ETH $BTC $SOL