nan

[湖若深 Shen HU @henryhuchina]

沃什他不是经济学家，而是个“宏观行为艺术家”

网上吹沃什新官上任的文章很多，但看透这波操作的人很少。他上台整出这一堆花活，其实全是无奈之举。嘴炮打得极鹰，却死活不肯直接加息，美其名曰“修复机构信誉比短期调息更重要”。

问题是，他头顶上站着的是特朗普。一个自带华尔街复杂背景、被民粹总统硬推上台的美联储主席，要给国际市场硬讲“美元信用”，这本身就是个笑话。

既然没法硬刚，沃什选了条极其聪明的路：拿“程序正义”当挡箭牌。删掉前瞻指引、砍掉60%的会议声明篇幅、一口气成立五个工作组。这一套动作，就是演给全世界看：“我们在跟白宫解耦，这叫制度自我纠偏。”特朗普那边也得配合，高喊支持央行独立。两人演了出绝妙的双簧，非要让市场相信“这次真的不一样”。

这就是高债务和民粹时代央行的死局：越是被政治裹挟，越需要拼命表演独立性。

就算我们真把沃什当圣人，全盘相信他的剧本，他能成吗？他接盘的美国，通胀、资产价格和债务全在天花板上，是个彻底“难回头”的死结。他想当硬汉，面对的却是Fed史上最无解的三重封锁：后危机时代戒不掉的流动性毒瘾、财政部根本停不下来的赤字狂飙、以及随时可能翻脸的政治干预。

所以别拿这些“独立性改革”太当真。沃什既不是政治家，也不是纯粹的经济学家，他就是跟特一起在表演宏观行为艺术。

只是面对华尔街对流动性的依赖和华盛顿停不下来的发债狂热，他这出戏，到底能演多久？

[Andreas Steno Larsen @AndreasSteno]

This is the chart that everyone should be watching.

If the Token Pricing rolls over, everything from the memory trade to the broader hard-ware and data-centre trade is over for this cycle imho.

The whole setup depends on this..

[nan @gremmn]

@Mimiwftt 证券 食品饮料

[川沐｜Trumoo🐮 @xiaomustock]

人类很难能获得远视来看清当下。

近三年的英伟达单卡显存容量实现了接近 3.6 倍的增幅。
而未来三年HBM端的需求会暴增5倍。

现在绝大多数人对存储股的恐惧是涨太多和能否真正跨越周期，这里可以直接跟英伟达绑定，如果存储股不能跨过周期，那么说明英伟达销量没有持续爆涨AI需求被证伪。
很显然，英伟达的增长依然强劲，AI的需求从全球token用量的增幅可以直观看到。
黄仁勋说未来AI对算力的需求是千倍，同理对存储也一样是这个倍数。

你相信AI革命是真的会发生，因为AI的消费品是算力，存储，光互连，那么可以变相理解为存储会转变成长股，同理AI革命确实在发生。

那么就可以简单理解如果你能接受英伟达因为Ai革命的而持续增长，因为英伟达的估值是成长股估值，实际上就承认了存储股已经转变成长股pe应当享受成长股估值，俩者本质上不可能单独发生，只是绝大部分人不愿意相信。

一、 英伟达侧的需求杠杆与虹吸效应
英伟达对 DRAM（尤其是 HBM）的绝对需求量正在经历代际间的级数放大，这构成了对全球存储产能的强制重组。
1. 代际规格的容量/频宽消耗弹性

Blackwell Ultra 时代（2026年上半年）：单卡对 HBM3e 的消耗量直接推升至 288GB。

Vera Rubin 时代（2026年底 - 2027年）：全面切入 HBM4 标准，单卡起步容量维持在 288GB，但后续 Ultra 版本预计将上探至 384GB 或 512GB。

系统 DRAM 的二次放大：与 Rubin 架构配套的 Vera CPU 将单插槽的 LPDDR5X 主存支持容量从 Grace 的 480GB 垂直拔高至 1.5TB。这意味着单台高级 AI 机柜对普通大容量 DRAM 的抽血效应同样在翻倍。
2. 资金与排他性协议的重组机制
英伟达正在利用其庞大的现金流，通过大额预付款与“不取货制裁”（Take-or-pay）长期协议，将三大厂未来两到三年的高阶产能进行了深度锁死。这种市场机制导致：

三大厂的产能能够基本支撑英伟达的绝对核心订单（即英伟达拿走了绝大部分有效供给）。

但这种锁死极大地挤压了其余云巨头（如谷歌、亚马逊自研芯片）以及标准服务器/消费电子市场的 DRAM 分配额度，造成整体行业的结构性整体短缺。

一、 三大存储厂的产能扩张规划与刚性硬约束（2026 - 2028）
三大厂目前正在进行有史以来最大规模的资本开支扩张，但实际有效位元产出（Bit Growth）的释放受到物理周期的严格硬约束。

1. SK 海力士 (SK Hynix)

产能现状与规划：其 2026 年的 HBM 产能已全部售罄。目前正在加速推进韩国清州 M15X 厂以及龙仁集群首座 Fab 的建设，预计新产能要在 2027 年底至 2028 年初才能大规模实现片级产出。

核心变量：SK 海力士明确指出，未来三年客户对 HBM 的确切订单需求已经超过了其理论供货上限。由于其正在向定制化 HBM（与英伟达共同开发）转型，这部分产线实质上已从标准存储市场中永久抽离。

2. 三星电子 (Samsung)

产能现状与规划：2026 年 HBM 销售目标拟实现同比 3 倍以上的增幅。目前三星已对外宣告启动 HBM4 的量产出货，并针对平泽 5 厂（P5）规划了高达 60 万亿韩元的总投资，但该晶圆厂的正式投产节点挂钩在 2028 年。

核心变量：前期的先进封装良率瓶颈（如 TC-NCF 工艺在高层数堆叠中的表现）虽在逐步收敛，但其 4nm 逻辑工艺底座（Base Die）的代工产能分配与前道良率，是决定其未来两年 HBM4 整体出货斜率的关键变数。

3. 美光科技 (Micron)

产能现状与规划：美光将 2026 财年的资本开支大幅拉高至 250 亿美元以上，其采用 1-gamma 制程节点的 HBM4（36GB 12高堆叠）已于 2026 年 Q1 进入高产出阶段，专为英伟达的 Vera Rubin GPU 平台定制。

核心变量：美光在 2026 及 2027 年的 HBM 供货已被长期绑定条款（Binding Terms）完全锁定。其在美国本土规划的超大型先进封装与制造基地，产能量产窗口同样落在了 2028 年。

4. 行业共同的刚性硬约束

无尘室（Cleanroom）天花板：当前行业共识表明，制约产能扩张的核心瓶颈不是设备采购意愿，而是无尘室的建设周期。从打桩到达到合规标准存在 18 到 24 个月的刚性物理周期，这意味着 2027 年年中之前，全球不会有大批量的全新 DRAM 厂房投入使用。

HBM4 晶圆消耗倍率：由于 HBM4 引入了 2048-bit 接口位宽，其底部的控制底座必须转由台积电等晶圆厂的先进逻辑制程（如 N4/N5）代工。这种前道逻辑工艺与后道三维堆叠的异质整合，导致其制造周期拉长，单片 HBM4 对晶圆的消耗量达到了传统标准 DRAM 的 3 至 4 倍。

[nan @gremmn]

@NullableX @Yue16518515 好认真啊 感谢分享

[一碗屌丝汤 @TSLA50T]

Damn 山区埋雷 无人机偷袭 死人不可避免

[Ben Kizemchuk @BenKizemchuk]

The Square-Root Law and the Death of Market Rotation:

One of the most important empirical discoveries in modern market microstructure is the square-root market impact law, developed by researchers such as Jean-Philippe Bouchaud and later connected to broader work on market structure by Xavier Gabaix. The relationship is simple but powerful:

I(Q) ∼ σ √(Q/V)

Where:

• I(Q) is the expected price impact of a trade
• Q is the size of the order
• V is typical daily trading volume
• σ is the volatility of the asset

The law states that price impact increases with the square root of the fraction of daily trading volume executed. In practical terms, when investors attempt to move large amounts of capital into or out of a market segment, prices must adjust in order to locate liquidity on the other side of the trade.

The key implication is that liquidity is far thinner than it appears. Institutional flows must move through layers of liquidity, and as those flows grow relative to available trading volume, price movements accelerate. In modern markets, this means that flows themselves increasingly drive price changes, often independently of fundamentals.

This dynamic interacts with another defining feature of today’s equity markets: the dominance of passive investment vehicles.

Passive funds now represent roughly half of equity ownership and an even larger share of marginal trading activity. These vehicles allocate capital mechanically according to index weights rather than valuation or macro views. Inflows are distributed proportionally across constituents, reinforcing the dominance of already-large companies and sectors.

Because passive capital does not rotate, the shrinking active cohort is responsible for nearly all discretionary allocation decisions. But the amount of capital controlled by active managers is now small relative to the total market, which makes large reallocations increasingly difficult.

Under this structure, the traditional concept of sector rotation becomes mechanically unstable. Active managers attempting to move large amounts of capital between sectors are effectively trading against limited liquidity, and the square-root impact law implies that such flows will produce outsized price movements.

As a result, the primary mechanism for adjusting risk is not rotation but degrossing. Funds must now reduce total exposure across portfolios rather than reallocate capital from one sector to another.

The market behavior of the past several months illustrates this dynamic clearly.

The process began in September, when multi-strategy and equity funds began reducing gross exposure. The most obvious place to cut risk was in the largest and most crowded positions: the U.S. mega-cap technology complex, particularly the Mag7.

As funds trimmed these long positions, they simultaneously covered short positions in smaller and mid-cap equities. That short covering produced a sharp rally in those segments of the market, which many observers interpreted as the beginning of a long-awaited rotation away from mega-cap technology.

In reality, the move was largely the mechanical result of balance-sheet contraction. Short covering requires far less capital than building new long exposure, so smaller and mid-cap stocks moved quickly even though relatively little net capital was being allocated to them.

At the same time, some capital leaving mega-cap technology sought destinations capable of absorbing large flows. The precious metals complex, particularly gold and silver, fit that requirement. These markets are among the most liquid in the world and were already in strong uptrends, making them natural targets for systematic and momentum-driven strategies.

As a result, early degrossing flows were split between three effects: selling of mega-cap leaders, short covering in small and mid-caps, and inflows into highly liquid trending assets such as precious metals.

But the initial selling in mega-caps began to trigger additional pressure through portfolio risk systems.

Because these stocks dominate index weightings and factor exposures, declines in the mega-cap complex quickly pushed portfolios toward value-at-risk and exposure limits. Once those limits were approached, risk systems forced further reductions in exposure.

At that point, selling began to propagate down the market-capitalization hierarchy.

Positions that shared similar factor exposures such as software and financials began to experience sustained selling pressure. Software companies sit immediately behind the mega-caps in many institutional portfolios, sharing similar growth and quality characteristics while also exhibiting high index correlation. As funds continued to reduce risk, these positions became natural candidates for further trimming.

The result was a broadening of the selloff from the very largest technology names into the wider growth complex, particularly software.

Meanwhile, the short-covering rally in smaller equities continued to ripple through the market. The relative strength of these names created the impression that capital was rotating into cyclicals, staples, and international markets such as Europe and emerging economies.

But the scale of capital embedded in the mega-cap complex makes a true rotation mathematically implausible. Reallocating even a small portion of that capital into smaller markets would overwhelm their liquidity and cause extreme dislocations.

What appeared to be sector rotation was therefore largely a statistical illusion produced by degrossing: the selling of crowded longs combined with short covering in thinner segments of the market.

Because those thinner markets require far less capital to move, they can rise sharply even when net flows are relatively small.

Eventually, however, the same dynamic that drove the initial rally began to reverse.

As volatility rose and risk limits tightened further, funds that had previously covered shorts or rotated exposure began reducing positions across those same sectors. Selling pressure spread into cyclicals, staples, and international markets, including Europe and emerging markets.

The pattern that initially looked like a broad rotation now began to unwind.

From the perspective of market microstructure, this sequence is exactly what the square-root impact law would predict. When large pools of capital attempt to adjust exposure simultaneously in markets with uneven liquidity, price movements become dominated by the interaction between flows and available trading volume.

Thin markets can rise quickly during inflow periods because relatively small amounts of capital push prices higher. But when flows reverse, those same markets can fall just as rapidly because there are few natural buyers available to absorb the supply.

This produces the air-pocket behavior increasingly visible across modern markets: sharp rallies followed by equally abrupt declines as flows change direction.

These dynamics have also interacted with the volatility market in important ways.

Earlier in the process, rising dispersion across individual stocks allowed traders to hedge index exposure through dispersion trades (typically long single-stock volatility while short index volatility). But as selling pressure spread across sectors and correlations began to rise, that dispersion collapsed.

As stocks began moving together again, dispersion trades lost their effectiveness and hedging activity shifted back toward index-level protection, particularly S&P 500 options and instruments linked to the VIX.

If volatility itself begins trending higher, this could introduce a further feedback loop. The volatility complex is structurally smaller than equity, commodity, or bond markets, meaning that large inflows into long-volatility strategies could produce outsized price moves under the same square-root impact dynamics.

Dealers hedging those exposures would likely increase their use of equity or volatility-linked options, which could further amplify volatility in the underlying market.

If equity indices keep weakening, volatility rises further, or the precious‑metals trend continues to fade, the next likely destination for capital would be the few markets with both depth and emerging strength: energy and U.S. Treasuries.

These are among the only venues able to absorb large reallocations without immediate square‑root‑style impact. However the Iran conflict and the resulting spike in crude volatility may narrow that list, effectively leaving Treasuries as the only major asset class with liquidity, lower volatility, and a developing uptrend.

Taken together, the sequence of events over the past several months illustrates a broader structural shift in how markets function.

In an environment dominated by passive ownership and large institutional flows, capital cannot rotate smoothly across sectors as traditional market narratives suggest. Instead, adjustments occur through degrossing, short covering, and reallocations into the few markets capable of absorbing large flows.

The square-root impact law explains why these adjustments increasingly produce abrupt and sometimes counterintuitive price movements.

In modern markets, capital does not necessarily move wherever investors would prefer it to go.

It moves wherever liquidity allows it to go.

[Frank trading @Franktradinglog]

本周第二重要的宏观事件，联储主席人选与Lisa Cook案的一个简单场景模拟：

场景A：Lisa Cook 留任，主席人选概率继续向 Kevin Warsh 倾斜，白宫在人事上试探，但没有突破制度红线。

解读为政治干预存在，但 Fed 仍是一个可预测的央行，使得tail risk 收缩。

利好美元（Fed reaction function 仍可建模），长端美债（恶性通胀溢价被压缩），高质量股票（折现率预期稳定）

利空黄金 / BTC（尾部对冲资产），极端通胀交易（steepener、inflation swap）

场景B： Cook 留任，主席为Hasset。

解读为理事会仍有法律保护，但主席高度行政化，美联储进入主席主导、理事会牵制的张力状态。

利好股指（尤其是 financial conditions 宽松受益者），信用（IG > HY），短久期资产

利空长端美债，美元（相对 G10）

场景C：Cook 被罢免 ，主席为Warsh，制度受损，但仍保留“技术官僚外壳”。

解读为“for cause” 被实质性掏空，理事会独立性遭到破坏，但主席仍尊重市场与通胀约束

利好黄金（开始获得结构性买盘），通胀对冲资产

利空长端美债，美元（信用溢价下降），高估值成长股

场景D：Cook 被成功罢免，主席人选转向更明显的行政导向（Hassett 或同类）。

解读为Fed 已经不再是独立央行，而是政策工具。

利好黄金，通胀资产（能源、商品），实物资产、海外资产

利空长端美债（结构性利空），美元，成长股（折现率 + 风险溢价双杀）

[nan @gremmn]

@NullableX 期待

[nan @gremmn]

@NullableX @qiu_benming 真厉害 阿里已经有消息了 。

[nan @gremmn]

@NullableX @qiu_benming 楼主，今天的无差别下跌，您怎么看～ 谢谢

[Frank trading @Franktradinglog]

以下是对于本周各种事件的详细分析：格陵兰，伊朗，联储人选，以及我的仓位分享，欢迎阅读。
https://t.co/BEC9yvbTRd

[Frank trading @Franktradinglog]

Below is a detailed analysis of this week’s key developments—Greenland, Iran, the Fed chair succession—along with a breakdown of my current positioning. You’re welcome to read on.
https://t.co/vs8kQHgVsb

[nan @gremmn]

@aWIV9o3FOYiDFRV 好久不见 欢迎回来～

[nan @gremmn]

@XXY177 来喽 互关

[nan @gremmn]

@lincoln0110 求回关

[nan @gremmn]

@0xAladdinPrince @grok 这里 回关 谢谢

[nan @gremmn]

@zyc9431 @okxchinese 求回关

[nan @gremmn]

@xq167168 来了 求回关

[nan @gremmn]

@MissOllie2020 求回观~谢谢~