Jim Chanos

@razazaidi58 @AJManaseer The $NBIS Faithful don’t seem to want to answer my very basic question, despite management’s highly promotional conference call highlighting new contracts at very lucrative rates. How many connected power MW’s will Nebius have at YE 2026, to support its $7-9B ARR guidance? @razazaidi58 @AJManaseer The Company still believes it will be 800 to 1,000 MW of connected power at YE 2026, so I’ll let the rest of you figure out the appropriate Revenue/MW figure for December. (Hint: It isn’t even remotely close to $20-25M per MW, annualized) $NBIS https://t.co/c9PL8R1XWC

Paying for compute with capital raised, you say?

@Lazarus_Capital (2) Furthermore, they are unprofitable (and seemingly will remain so for years) at a point in the cycle when they should be extracting monopoly rents. If not now, when…?! $CRWV $IREN $NBIS

Nobody Tell CoreWeave It's Not a Business $CRWV $IREN $NBIS https://t.co/77kojD7vVo (2) The author makes the very salient point that the neocloud moats are both dug and filled by $NVDA, who can breach or bypass those same moats whenever it chooses. Whether by allocation policies, price hikes, etc. The neoclouds are financial conduits, not technology companies.

So $HOOD customers were down (5.2)% in July, net of new deposits. The S&P 500 was flat. https://t.co/8UcMm5aI89

And of course, $CRWV 2027-28 Adjusted EPS estimates DROPPED after the print, particularly 2028. Stock now at 90x Adjusted 2028 EPS. https://t.co/ajAuKOE4ye

@RHouseResearch (2) And of course despite all the price-hiking happy talk, the biggest increase in 3Q guidance relative to expectations was…capex. $CRWV https://t.co/sj44Iii18C

So $CAVA, which trades at over 100x flattish EPS, just guided to a huge slowdown in 2H SSS. The 1Q was 9.7%, and this 2Q was 9.0%…but the entire year is now guided to 4.5% to 6.5%…?! That’s 0-4% for the 2H of the year. https://t.co/JcnpSraTrn

It’s not out of the realm of possibility that the next time these guys are all at the same table explaining AI financing is before Congress in 2031… https://t.co/fY8zGqk0Jk

@MelvinInvests There’s just one little problem with this super-bullish Goldman model for $NBIS: Note the $20B difference between EBITDA and EBIT in 2030. That implies a capital base of $120-150B. $9B in EBIT will not cover their WACC in 2030…with 76% EBITDA margins. See the problem? https://t.co/tzSAhHnFt5

@firstadopter @SemiAnalysis_ You really believe that SpaceX is going to build 25% of all estimated GW build next year(40 GW)? For $500B? By the end of the year? $SPCX @firstadopter @SemiAnalysis_ (2) Particularly given that the Company itself just said that its capex will be flattish in the 2H of 2026. $SPCX https://t.co/JzazBcQMwF

Our favorite solar science project, $RUN, burned another $(620)M in cash in the 2Q, and guided down 2H expectations. 2Q EBITDA again did not cover interest expense. Complete joke.

@biotechpain And what “fully autonomous” cars would those be?

@DumbMoneyCapitl Because those end-users don’t earn ridiculously high returns as Elon’s claim indicates.

@FaraoniEmanuele If there is really demand at that price you would have renters willing to pay that price to $NVDA, not $SPCX.

@FaraoniEmanuele What are you talking about? The leased return in Year One alone is greater than a sale, given NVDA would depreciate the chips over the term, not put them in COGS. This would be the easiest business decision ever.

@corban_urbanic @firstadopter $CVNA SG&A at 9.5% of revenues is higher than CarMax’s 7.9%.

So if that’s the case, why won’t $NVDA simply lease the chips out themselves…?! Lease them for 3-5 years, then sell them at residual value. (2) $NVDA would earn 200-400% margin at those rates, over 3-5 years, versus 75% on a sale of the GPU.

@sunmoonbuffalo @firstadopter CarMax didn’t know that? Or offer financing?

@firstadopter Now ask yourself just how are they able to make huge gross profits per vehicle if they are routinely paying such huge premiums over the next bid in a highly competitive/efficient market? $CVNA

@MB_Hogan So $SPCX is going to build 20% of the expected 40 GW in total new AI data center builds next year…?! For $250 to $350 billion? @MB_Hogan (2) All while claiming they are going to keep AI capex “stable” over the next few quarters. How does that work…?! #SayAnything $SPCX

@YousifElmusharf @elonmusk And certain hyperscaler CEO’s as well.

@OnodaCapital They are. And the companies are, too.

@wholemars Is that why $TSLA earnings estimates for 2027 and beyond keep dropping every year? #Insane

@omegaaaaa4 The problem is, “all at once” will involve every auto OEM, with ubiquitous FSD as a cheap option/standard equipment.

At $100/month this is a $1.3B increase in $TSLA FSD revenues…over five years. After ten years on the market.

Now that FSD is proving to be less than a $2B annual business for Tesla after ten years, and Optimus has been pushed out(again), I see we are back to Robotaxi’s being a 10% of global GDP “opportunity”. The NeverEnding Story. $TSLA https://t.co/5fM1gmzWNf

$RUN

A classic from Hank Blaustein, circa GFC… https://t.co/GbKPGyDbhi

@icecube2000000 (2) And remember they are actually not selling that many vehicles to the general public, either. $RIVN https://t.co/tJsbUqFwJt