把话说清楚

🤖 INFJ
💻 全面拥抱Ai
🤝 找同频的人一起交流

[把话说清楚 @wantlvm]

美国要把俄罗斯天然气重新卖回欧洲
而且是走乌克兰管道

据《华尔街日报》，这是库什纳和威特科夫9月份提方案：

俄气交给一家美国中间商来卖收入放进托管账户，莫斯科动不了

乌克兰收过境费，换一个"能源停火"欧洲不干，人家定目标是2027年前彻底摆脱俄气克里姆林宫也说不太可能

但现实是欧洲气价已经是美国11倍！

储气率不到75%，冬天马上就到俄罗斯卖气、乌克兰收钱、美国赚差价最后买单的会是谁？

[把话说清楚 @wantlvm]

上周四纳指单日跌了近5%
25天蒸发4万亿美元
10月AI股创下有记录以来最大月跌幅

今天就反弹了
纳指100 +1.3%，标普 +1.0%
美股一天收回7000多亿市值
原油跌6.5%，黄金涨2%

抄底资金进场，美伊和谈的消息又冒了出来
市场的情绪，翻脸比翻书还快

这是修复，还是下跌途中的喘口气？
我觉得还得看这周的中期选举和财报

[把话说清楚 @wantlvm]

欧洲超市的冰柜，被黑客攻击了。

法国10月就开始出事，大概700家店中招。到11月2号，比利时、荷兰、瑞典、爱尔兰、爱沙尼亚都陆续冒出来了，比利时约75家，荷兰至少17家企业。

黑客不偷钱、不偷数据，专门让你冷柜停转。

以前觉得网络战离生活很远，现在发现它可以直接把你家楼下超市的冻肉变成常温肉。

[Rand Group @randgroup]

🇪🇺 BREAKING: Hackers are shutting down Europe's supermarket fridges. Brussels is drafting a 400-page directive on fridge feelings

[把话说清楚 @wantlvm]

美国10月制造业PMI终值54.9。

低于初值的56.0，也低于9月的55.9。

还在扩张区间，但回落得很明显。明天中期选举，市场情绪大概会更敏感一点。

[把话说清楚 @wantlvm]

招股书里还有个更夸张的数字：Anthropic已经承诺要花掉8954亿美元。

去年底这个数字大概才1050亿，现在直接飙到接近万亿级别。

谷歌还在它前面一点，Meta已经被它超过了。AI军备竞赛，账本上已经写死了。

[zerohedge @zerohedge]

Anthropic has $895.4BN of committed spend (per latest S-1). That’s up from about $105bn at the end of 2025, and the only hyperscaler with more is GOOGL ($1,059bn). META is at $866BN.

[把话说清楚 @wantlvm]

Anthropic更新了招股书，三季报数字挺吓人。

营收169.6亿美元，去年同期才13.8亿；运营支出175亿，手头现金还有956亿。

营收暴涨，烧钱也暴涨。这大概就是现在AI公司的常态。
https://t.co/S4f5nIstTD

[把话说清楚 @wantlvm]

甲骨文说要再融200亿美元，股票和债券一起发，钱拿去砸AI基建。

消息出来股价一度跌超5%，离今年高点已经砍掉一半多了。

AI故事还在讲，可市场已经开始问：这钱砸进去，什么时候能赚回来。

[The Kobeissi Letter @KobeissiLetter]

BREAKING: Oracle stock, $ORCL, falls over -5% after announcing plans to raise $20 billion through equity and debt issuances to fund its AI buildout.

The stock is now down -54% from this year's high.

[把话说清楚 @wantlvm]

@gnn1948944 大部分人都是辛辛苦苦3个月

[把话说清楚 @wantlvm]

SpaceXAI的Colossus 2还没完全上线，算力就已经相当于111万块H100，是目前全球最大的AI数据中心之一。

国内最大的单集群大概也就十几万H100当量的水平。总量上中国算力规模在追，但单点规模和先进芯片的差距还在。

有人已经按百万卡的规模在堆算力了，国内还在靠国产芯片和集群堆叠往上赶。差距不是缩小，是在拉开。

[Edward @EdwardTWinz]

Before its next clusters even come online, SpaceXAI's Colossus 2 in Memphis is already the world's most powerful AI data center.

Roughly 1.11 million H100 equivalents. That's computing power, not a chip count: what 1.11 million Nvidia H100 chips would deliver. The real chips: Nvidia Blackwell B200 and B300.

Ahead of OpenAI's Stargate. Ahead of EVERY big tech campus on Epoch AI's list of 96 sites.

This is where it stands:

1.11 million → Colossus 2, Memphis (SpaceXAI)
1.02 million → Stargate, Abilene (OpenAI/Oracle)
818,000 → New Carlisle, Indiana (Amazon)
769,000 → Fairwater, Atlanta (Microsoft)
637,000 → Pryor, Oklahoma (Google)
584,000 → Nusajaya, Malaysia (DayOne)
536,000 → Prometheus, Ohio (Meta)
481,000 → Kempas, Malaysia (DayOne)

By mid 2027, Colossus is expected to make that gap WIDER, with more Blackwell clusters coming online.

Epoch AI's Q1 2027 projection:

1.82 million → H100 equivalents
$58.0 billion → cost, up from $35.8 billion so far

In October 2025, it was at 278,000. Today: 1.11 million.

That's 4X. Not in 10 years. Not in 5. In about 12 months.

Follow me for more news like this.

[把话说清楚 @wantlvm]

一辆自动驾驶车里装了4块H100，光算力就20万美元。

不是买不起，是这个成本根本没法规模化。

车端芯片和内存的价格不降下来，自动驾驶再怎么吹，也只能停在少数高价车上。

[Jun Song @jun_song]

More context :

A single Waymo car uses 4 H100 GPUs just to drive itself. That is about $200k worth of compute inside one car.

Right now, the Nvidia Orin self-driving modules used in Mercedes pack, and Tesla FSD use 64GB of LPDDR5X RAM.

Most people think better tech means we will need less memory over time. The exact opposite is happening.

Tesla's upcoming AI5 chip will jump to 96GB, and the AI6 chip is going straight to 144GB.

Even your smart fridge and smart TV need RAM now.
Soon, literally every physical object around us will have memory built into it.
And as robotics takes off, those memory numbers are going to get much bigger.

There is no way around the hardware bottleneck:
If AI robots run on the cloud, you need massive data centers full of memory.
If AI robots run on-device, you have to pack huge amounts of RAM directly into the machine.

Either way, the world is going to need an insane amount of memory.

This is not a problem we can fix easily.

[把话说清楚 @wantlvm]

工作对一个人最大的消耗，可能不是身体累，而是让你逐渐失去对生活的兴趣。

以前下班以后还想看电影、打游戏、出去吃饭，周末也会计划去哪里玩。

后来每天只想躺着，连喜欢的东西都懒得碰。

最可怕的是，你明明是在努力赚钱改善生活，最后却连生活本身都快顾不上了。

[把话说清楚 @wantlvm]

什么是起死回生?

起床就要死了

回床上就活了

[把话说清楚 @wantlvm]

@Actionmbns 浇朋友来

[把话说清楚 @wantlvm]

@LXiangchen71169 恭喜一半了！

[把话说清楚 @wantlvm]

@fengyij12654413 浇朋友！

[把话说清楚 @wantlvm]

@Masionww 浇朋友噢

[把话说清楚 @wantlvm]

@Duiady 浇朋友！

[把话说清楚 @wantlvm]

@guan_jiang99446 就一片😭

[把话说清楚 @wantlvm]

一碗牛肉面但是没牛肉

[把话说清楚 @wantlvm]

@EVAtk7b 手搓1000条直接要了老命了

[把话说清楚 @wantlvm]

@Justin1219666 狠狠避雷