Irwin Kim retweeted
My "gut" keeps worrying about a market downdraft. My weight-of-the-evidence indicators (for example, % of SPY stocks above 50dma crosses above 30% while the ETF close > 50-day average) keep giving favorable signals. What's a disciplined investor to do? @sentimentrader
Summarize the narrative change of these weeks perfectly
Why hasn't Iran escalated? Here's my theory.
Remember the 300 billion earmarked for "reconstruction" that Iran is supposed to be paid per the MOU?
The Iranians cannot agree to any "temporary" deal, certainly not one that comes with nuclear concessions up front. And their demands regarding the Strait are also a red line.
The US cannot accept anything that doesn't include those two items. It's made its position clear and it has thoroughly shredded the MOU from its own memory.
This means that the Iranians assume that there will not be a deal. The Iranians assume the US will try to force them to move rather than negotiate under their terms.
The US can use "economic siege" as part of that strategy. Some analysts seem to think it will be sufficient, although the timeframe for success is typically rather ambiguous (see the O'Hanlon op ed in the WSJ from this week, for example).
But this is hardly certain. And "it will work eventually" is hardly persuasive when the two main analogs that come to mind--cuba and north korea--haven't exactly been speedy. And time is very, very expensive.
Sure, its expensive for the Iranians. But how much is too much is a matter of perspective before it is a matter of accounting. Are they willing to pay that price? Probably!
The problem is that time is also very expensive for the americans. See my postscript on that below.
I think that the expense for the Americans is high enough that Trump doesn't have much to lose by escalating after the midterms.
The only meaningful escalation is an infrastructure war. They've already bombed the 15,000 most important military targets in the country, and detonating a performative airburst munition above a granite mountain isn't a strategically serious idea. All that's left is bridges and power plants. It's been that way since March, too.
The Iranians believe that, because a deal is impossible, the Americans will attack after the midterms. They expect the attack will target infrastructure.
AND THIS IS THE IMPORTANT PART: the expect the US to start it.
So, I ask again, remember that 300 billion earmarked in the MOU for reconstruction??
You are the Iranians and you think an infrastructure war is all but inevitable. However, you suspect that your enemies will ultimately blink first, as they have more to lose. If you can keep landing punches, it doesn't matter how many you take. Four weeks of this would wreak years of damage to regional economies. If you can keep swinging you can win.
And if you can win, you can win money.
Let the American bombs blow up all your old energy infrastructure. Then get Gulf money to build new infrastructure.
THE KEY OF THE STRATEGY IS THAT IRAN MUST DO EVERYTHING IT CAN TO PRESERVE ITS STATUS AS THE "DEFENDER" AGAINST "UNLAWFUL AGGRESSION." THAT IS WHAT MAKES THE STRATEGY FUNCTION. THAT IS WHAT CAN, WITH ENOUGH PAIN, MAKE THE CASE CLEAR ENOUGH TO WIN IRAN A DEAL SECURED BY UNGA AND UNSC.
And so, Iran won't start the infrastructure war. It can't. Best it can do is launder it through the Houthis (which is actually pretty damn useful).
Irwin Kim retweeted
This doesn’t happen unless there is noise coming from the repo desks.
There’s a major financial institution at the threshold of not being able to fund itself in The Street.
First check for NY Fed is to identify which other institution(s) have material counterparty exposure.
Not good.
我毫不怀疑十年二十年后AI的主要输入形式会变成视频，因为那时的存储行业已经扩产到存储芯片不值钱的地步，但在两三年之内要顶着存储价格做出这样的变革，这种场景的潜台词是视频输入输出AI带来的效率提升远高于当下的形式并且消费者愿意为这种形式付出相应的溢价
请发给你的韩国朋友。
AI能干的事情过去2年，
慢慢从一维数据到二维数据再到未来的三维数据进化，也就是文字，图片，三维视频。
这经历的时间也才短短2年多，24年到现在26年。而存储扩产在2028年，也就是2年后，AI会发展成什么样，这些傻逼预测分析师硅谷投资客都没有考虑进去，他们只是基于现在的AI状态来分析存储未来尤其是存储2028年的未来做出当下的判断，他们都是傻逼。
如果物理AI能突破，处理的资料都是视频资料，不是文字，存储数据内容从一维到三维，存储的需求量是天量。人们现在对存储的假设是基于现有AI水平来测算。
但这些傻逼都没有考虑进去2年后AI会发展成什么样，
现在AI视频已经可以以假乱真，
俩年后更是不敢想象会进化到啥程度，
如果AI能进化到理解现实物理世界，那它处理的数据全是视频数据，
存储需求的数量级可能是百倍于现在，
要坚信你的存储。
试想一下AI俩年后会发展成什么样，回想过去俩年AI的进化速度和对存储需求的增长速度，不要再扔掉你们宝贵的 #海力士 $SKHYX 了。
AI用量会在俩年后暴增数倍，想象你2年后使用频率，从现在一天3-5次会不会暴增到30-50次？
AI使用群体数量俩年后会不会从1B暴增到几B？（使用群体不限于人，包括汽车，包括手机，包括机器人，甚至智能体本身…）
AI现在人70%的使用是文字token，俩年后会不会进化到视频影像物理三维视频token输入输出？（这里存储数据量差别千倍万倍。）
这里面的需求差别2年后会是现在的多少倍？十倍还是百倍？
那俩年后的存储产量增量是几倍？翻倍都撑死了吧？这些傻逼预测师永远不会这么给你算，只会基于现在AI状态误导你。
拿出你们290万韩元梭哈的勇气，
190万韩元都不敢买，真去死了算了。
Same thoughts here
$GDX The gold miners are due for a decent sized dead cat bounce starting as soon as tomorrow given gold looks interim capitulatory overnight.
Watchlist!
