crypted Labrador

SCJP to CFA
NFA

[crypted Labrador @Tom20200714]

@GaoHeadmaster 船都刻漏了

[crypted Labrador @Tom20200714]

@yangyue992125 很深刻。有体量的上游模块，都是靠拿大血库亏越周期杀出来的。 日本人也轴，但是没有韩国又轴又耿的厉害。
至于岸这边能不能一起玩大周期，这个可能也没那么轻而易举。这两个新玩家刚开始启动时的贵人现在是哪里免费吃喝？面板时代扛大周期时的相助贵人们又在哪里免费吃喝？路长且阻。

[crypted Labrador @Tom20200714]

同意没到崩的时候，但也难再翻倍上去。明年HYPERSCLAER是要拷问roi的，所以市场不会为短期1-2年的涨价而把估值贴现搞上去。后面更可能的会是宽幅震荡，并在某个临界点把韭菜套在里面。
从风险收益比，个人拙见是逐步减仓，落袋为安了。在某个临界点，就是会出现上涨潜力30%，但是下跌可能腰斩或者1/3斩

[crypted Labrador @Tom20200714]

@yangyue992125 @BigbirdflyChan 2、情绪顶。所有的高潮过后总有不应期，所有的牛市激情总是会剩下韭菜的一地鸡毛。这轮牛市已经持续了很多年，最近的价格上涨斜率也告诉我们已经是情绪赶顶阶段了。顶部派发可能需要一个过程，但现在再持有已经风险收益不成比例。密切关注insider抛售事件。hyperscaler roi大考可能在明年q2-q3.

[crypted Labrador @Tom20200714]

@yangyue992125 @BigbirdflyChan 这里有两个顶：
1、基本面，存储是强周期产业，一直都是牛短熊长，本质上是下游大厂的谈判能力强于内存厂。AI短期大量HBM订单，造成了产能从低端产品向高端产品的配置迁移，但AI CAPEX是有ROI的地心引力的，总会回到常规水平，届时需求会回归，且内存厂已经在扩产中。现在的股价已经超过DCF模型的定价

[crypted Labrador @Tom20200714]

@QuintenFrancois big bets that AI bubble crashed in 2027

[Brandon Carl @brandonjcarl]

There's a growing consensus that we're in a bubble now in semiconductors.

When does the bubble burst? I'll try to take a stab at that here.

Lots of people have lost their reputation, jobs and personal finances trying to time bubble tops.

In this I am going to use the Nasdaq 100 during the Tech Bubble and the SOX Semiconductor index for today. You can appropriately criticize that choice.

Measures suggest that the market isn't healthy, nor stable. The role of markets are to allocate capital, price risk and establish value. None of those things are happening in a health manner at the moment.

Why? Current hyperscalers have very large and mature businesses that are hard to separate from the rapidly growing ones. The semiconductor index is where the highest degree of index-level speculation is occurring.

I'll look at aggregate Return, Risk, Issuance, Funding and Policy. We have met most of the "necessary" conditions for a top. However, both Funding and Policy have the potential to prolong it.

PRICE MOVES

In my study of bubbles, I find that they operate similarly to @geoffreyamoore "Crossing the Chasm". This book is nominally about how products achieve dominance. However: bubbles are largely narrative-driven and I find the same principles apply.

• The Early Adopters see the trend before everyone else. Patience is required.
• The Early Majority catches a huge part of the run up. The money comes easy and fast.
• The Late Majority catches sizable gains, but with much higher risk.

During the Nasdaq Bubble
Anchor date: 1995, when "cyberspace" became popularized.
• First run = 200% over 3.5 years (37% pa)
• Second run = 100% over 1 year (100% pa)
• Third run = 100% over 5 months (428% pa)

Semiconductors Today
Anchor date: much harder but I'm interpolating between the Transformer (2017), GPT-2 (2019 when scaling laws began to show promise) and ChatGPT (late 2022)
• First run = 230% over 6 years (22% pa)
• Second run = 130% over 1 year (100% pa)
• Third run = 100% over 3 months (1500% pa)

You can argue with the base effects for the "First run" (understandably), but the recent runs I believe are much harder to dispute.

[crypted Labrador @Tom20200714]

@camelforce0508 the market cap has priced in the good position of storage chip in Ai products chain.

[crypted Labrador @Tom20200714]

@MilkRoadMacro @novogratz Presure is scheduled:
1) First unlock window，Around Aug. 2026, 20%-30%
2) Rolling staged unlock, Day 70 → Day 135 ,35%
3) Second major unlock, Around November 2026,28%.
4) Final 180-day full unlock
But OAI is more vulnerable than spacex。

[crypted Labrador @Tom20200714]

@unusual_whales 看来承受不住压力了，原油瞬间暴涨啊

[crypted Labrador @Tom20200714]

@miner_huang 我列的是一种可能而已，也许过于悲观，但在现有信息聚合下是有可能性的。大家集思广益🙏

[crypted Labrador @Tom20200714]

AI行情目前是牛皮行情，鸡肋的很。宽幅波动会持续到AI泡沫的估值被严肃修正。那么最可能破灭的时间呢？
1、2026年10月 → 2027年4月蜜月期。
2、承压期高危期，2027年4月-2028年4月

[crypted Labrador @Tom20200714]

不会是2000年式的瞬间崩塌， 因为没有债务爆仓的杠杆链条 ， 而是一个慢速漏气的过程 。 领头羊（微软、谷歌、亚马逊）能扛住 ， 但中层AI公司、AI基础设施公司会大量死亡， Nvidia的估值会腰斩但不会归零

[crypted Labrador @Tom20200714]

AI估值调整的关键因素：
第一，ROI数据已经是定时炸弹
2025~2026年企业调查显示95%的AI项目零可测量ROI，而2026年全年AI基础设施投入超过$7250亿，实际AI收入几百亿，同时token价格大战已经开始。这个缺口不能永远靠"未来想象力"填补。

[crypted Labrador @Tom20200714]

路径猜测：
2026年Q3~Q4： 大公司财报开始出现"capex继续增加但AI收入未达预期" 市场第一次真正动摇，Nasdaq小幅调整10~15%
2027年Q1： CFO开始大规模砍AI预算（ROI无法向董事会交代） ， OpenAI融资困难hyperscaler首次下调capex指引
2027年Q2~Q3：泡沫核心破裂窗口， BTC作为提款机被同步抛售

[crypted Labrador @Tom20200714]

唯一能打破这个逻辑的变量，BTC与Nasdaq的相关性什么时候解耦。
解耦的条件：主权国家大规模采购BTC作为储备（去美元化加速）；美元信用危机导致BTC避险属性重新激活。
目前这些条件都没有成熟的迹象。所以BTC的真正底部在AI泡沫破灭之时，时间窗口是2027年下半年到2028年

[crypted Labrador @Tom20200714]

如果真发生矿工周期失效的情况，那么今年9-10月份冲进去的哥们会很惨。所以，现在不仅要关注时间维度，还要关注绝对低价带来的绝对安全，更要关注熊底掠夺流动性的固定动作。宁愿错过，不能上错。本金在，永远不发愁什么时候翻番；本金没了，就gg下桌了。