Jim Chanos

@FransBakker9812 @techstockpros (2) And of course, that’s just the GPU’s! Now factor in the D&A of the additional PP&E needed to upgrade/greenfield the facilities for AI, as well as those capital costs. The financial models for these Bitcoin miners-to-AI-DC companies leave a lot to be desired, to be kind.

“I’ll take ‘Things Not Seen At Bottoms’ for $600, Alex.”

Ok, Team $IREN, what is your best estimate of the total cost of these 23K GPU’s? https://t.co/VpYnXBDgpV

Jensen’s estimate that a 1GW AI factory will cost $20-30B BEFORE GPU costs, is well above what many AI data center companies currently are telling investors their costs will be.

The last sentence of the $NVDA/OpenAI press release. https://t.co/5NWAJgVKDp

[Narrator: $BMNR of course raised this cash at a big discount, as the warrants have a large implied value themselves. Which is why their stock is down on this announcement.]

The Bitcoin Treasury Theater of the Absurd

Important to remember this is still baseball season, Wisconsin. #Yikes

@SilbergleitJr @gnoble79 $OPEN’s gross margin (“spread”) is comparable economically to an agency commission. And as you know, residential RE commissions are now under pressure themselves. And counting on rates going down means you have a rate-sensitive, leveraged broker-dealer that’s not growing now.

@SilbergleitJr @gnoble79 With all due respect, that’s not how the residential real estate market works. And online agency businesses have been trying to “disrupt it” for almost thirty years now. The product is not standardized and the other frictional costs (title, transfer taxes, etc) can’t be cut.

@gnoble79 That’s because “house flipping” as a principal is the ultimate adverse self-selection business. Every home is literally unique, and the seller is ALWAYS the party with an asymmetric information advantage. Buyers offset this advantage only over time (and w/TLC). $OPEN does not.

Wow, hard to believe.

I’m sorry @BadgerFootball, but if you’re punting from the Maryland 40 down three scores in the 4Q, you’ve given up. (Maryland got the ball on its 1 and promptly went 99 yds for a game-ending TD)

“We have always been at war with Eastasia.”

Energy Traders Cite Gains, But Some Math Is Missing | A shoutout to my good friend (and former analyst) ⁦@JonathanWeil⁩, for tomorrow’s 25th anniversary of his column in the Texas ⁦@WSJ⁩ that started our journey with Enron. https://t.co/U7nk4sb9RO

Good thing xAI was not trying to raise capital at a $200B valuation this quarter, according to its CEO. https://t.co/XYIBLByx3H

This is a pretty serious allegation stated as a fact. Would love to see the evidence of this, as most of the mNAV bears have been pretty transparent on their reasoning. $MSTR (2) And just so we’re clear here, in 2025 so far $MSTR has issued 32M new shares while the short interest has actually DROPPED 600K shares over the same period. Someone tell Saylor that the phone call may be coming from inside the house! https://t.co/paZl0x9pDf

Meanwhile, back at DOGE…

Luckily it was still Biden’s Market at the April lows!

Oh no $ENE

More financial gibberish from Saylor. So $MSTR may sell options on Bitcoin (or sell Bitcoin itself) to finance the dividends on securities he issued to buy Bitcoin?! Lol, ok. (Who falls for this stuff?)

Losers will get four tickets.

Keep in mind that most of these thematic baskets have betas of 1.5-2.0 (vs S&P). The “Most Short” basket (we know it well) is at 1.6, for example.

$WING, yikes! https://t.co/DXqswp6i8O

The casino is packed, and it’s getting a bit late, but the booze is still flowing…(Just double your money ten times in a row.)

Great thread on the hidden risks in the neocloud land grab deals with the hyperscalers (Nebius and Microsoft here). Again, are these companies competitors, suppliers and/or landlords?

@sleepingdogo He might be. It’s the only thing that remotely makes sense.

I’m sorry, but can someone point me to a company that is receiving 10x “yields” (whatever that means) on their capex spend on $NVDA chips…?! (2) Because the reality, of course, is exactly the opposite of his claim. AI-hosting and hyperscalers are having to invest $4-6 for every $1 of revenue that derives from that capex. He must know this. $NVDA

“…it might say that”?!

Legacy colo data center companies have almost $20M per MW in invested capital (and that’s WITHOUT GPU’s/servers), or $20B per GW. So just where are all these new players in this space going to get tens of billions in $, to buildout their 1-2 GW locations?! $DLR $EQIX