Neocloud Bloodbath: Updated Valuations, Assumptions, EV/GP/RG & thoughts for NBIS, IREN, CRWV, SPCX. — Transcript
Full transcript
- 0:00Hello everybody. So this video is
- 0:02entitled Neo Cloud blood bath,
- 0:06EV over GP over RG. And I'm going to
- 0:10go through the updated valuation that I
- 0:13have for Nebius, Iron, and CoreWeave,
- 0:16and also for the valuation for SpaceX
- 0:18because I value SpaceX as a Neo Cloud
- 0:21because
- 0:23the value of of the launch business,
- 0:25something like $4 billion in revenue or
- 0:27something like that. I think most of the
- 0:29value in SpaceX
- 0:30stands lies in the Neo Cloud business
- 0:32for now, at least for the next 12
- 0:34months. So that's why
- 0:35I use SpaceX as a comparison here.
- 0:38And I want to talk briefly first start
- 0:40with with the blood bath. I mean, it has
- 0:43been
- 0:44relentless. Like this is the fifth week
- 0:47in a row of consistent
- 0:50decline, heavy decline on the Neo
- 0:53Clouds.
- 0:55You can see Nebius
- 0:56as of after hours, we are and even
- 0:59before after hours, but we are at a one
- 1:01month low down, you know, 50 56 %, 57 %.
- 1:06You can see Iron. So Iron, if you
- 1:09actually look in after hours,
- 1:11it's also a one month flow on Iron.
- 1:14You know, different reports, different
- 1:16companies, both one month flow. The
- 1:18chart looks very much the same. You look
- 1:20at CoreWeave. CoreWeave, I think they're
- 1:22like 2 % of the lows, only 2 %.
- 1:26But same, if you had back the after
- 1:27hours price action down nearly 3 % in
- 1:30after hours. And you look at the whole
- 1:32sector, I guess we we are like 2 % away
- 1:35from an old time low for for the ETF
- 1:37that captures kind of all of these Neo
- 1:39Clouds, which is the Round Hill Neo
- 1:40Cloud ETF, which is kind of used to I
- 1:42use for for the sector.
- 1:43And so one of the one of the things, of
- 1:45course,
- 1:46that I that I think investors should ask
- 1:47themselves is
- 1:50did did their quarterly earnings even
- 1:52matter? And
- 1:54the internet has decided that Iron and
- 1:57CoreWeave had good earnings, you know,
- 2:00arguably plus quarters as I would say
- 2:02they had A+ quarters. I listened to both
- 2:04calls by the way. They were both
- 2:05excellent.
- 2:06And some folks didn't like the Nebius
- 2:09call. I I personally thought the Nebius
- 2:11call was perfectly fine, but I guess the
- 2:14internet absolutely hated on the Nebius
- 2:17call.
- 2:18And so one of the things of course that
- 2:19I think the internet should ask itself
- 2:21is did you hate on the Nebius call
- 2:22because you dropped halfway through the
- 2:24call because Nvidia was down 11% and and
- 2:28Jensen said they were resigning. And so
- 2:31is the dire sentiment on AI
- 2:34on Nvidia, is it the reason why you
- 2:37didn't like Nebius call? Because Jensen
- 2:39literally resigned in the middle of the
- 2:41Nebius call. People dropped off the call
- 2:44and we were going into a really really
- 2:45bad AI sentiment starting the call
- 2:48because Nebius just happened to report
- 2:50on the same day that that that that
- 2:52Nvidia was dumping and on the same day
- 2:53that Jensen was going away. So so that's
- 2:56that's I think the major reason as to
- 2:58why people didn't like the call. But
- 3:00some but of course price drives
- 3:02sentiment. Stock price drives sentiment
- 3:05and so people are trying to find a
- 3:06reason as to why the stock dropped. And
- 3:08by the way,
- 3:10Iron dropped more. And I but Iron
- 3:12reported on a day where sentiment was
- 3:14good. So people loved the call at Iron,
- 3:15but Iron dropped more than Nebius.
- 3:18But the point is the reason why people
- 3:20say they didn't like the Nebius call
- 3:23is because apparently some people were
- 3:25spooked by 40 to 45 billion in 2027. But
- 3:29if you go if you go deeper into the
- 3:30weeds,
- 3:31you can see that any any basic
- 3:34multiplication of their capacity by
- 3:37the the cost of filling that that
- 3:39capacity with Vera Rubins by the cost of
- 3:41for Vera Rubins would have actually got
- 3:43a new
- 3:45more than 45 billion actually actually
- 3:4750 to 55 billion
- 3:49is what this would have gotten you just
- 3:50taking that capacity, looking at the
- 3:52cost of Vera Rubins, multiplying
- 3:55Rubins expressed in megawatts by the
- 3:57capacity would have gotten to to 50
- 3:59billion. So this is this is actually
- 4:01uh a little lower of a guidance if you
- 4:04were assuming they were going to have
- 4:05the latest gen GPUs in there. But the
- 4:08analysts can't look up the cost. They
- 4:10can't be bothered to do that and that's
- 4:11the reason uh why they were seemingly
- 4:13disappointed because they were expecting
- 4:15like 35 billion or something like that.
- 4:17And then another reason why that people
- 4:18found to to dislike the the the Q3 is
- 4:21they say oh they're switching from ARR
- 4:23to revenue. And I don't like that. I
- 4:25don't like that they're switching from
- 4:26ARR to revenue. Um except there is a
- 4:29clear reason for that switch. The reason
- 4:33for the switch is because they expect to
- 4:35have more short-term contracts. The
- 4:38whole industry is going is going after
- 4:41short-term contracts. You may have seen
- 4:42that's that's SpaceX and and and Fropic
- 4:45are like resigning the contracts I think
- 4:46in 6 months or in 4 months or something.
- 4:49That was trending
- 4:50uh earlier today. Um so so so so of of
- 4:55course as the cost of compute is going
- 4:57up a lot of these companies are like
- 4:58well you know I want to do some super
- 5:00short-term stuff. I I I want to I want
- 5:02to I want to I want to rent my compute
- 5:04for the month. I want to rent it for 2
- 5:05months, 3 months.
- 5:07And so if you don't have long-term
- 5:085-year contracts if that's the new
- 5:10direction of the industry then ARR
- 5:12doesn't really make as much sense uh as
- 5:15using revenue and revenue perhaps could
- 5:18become you know even more volatile. And
- 5:20of course the market doesn't like
- 5:22volatility in general.
- 5:25But of course uh Iron is still using
- 5:27ARR, Coreweave are still using ARR, they
- 5:29dropped it just the same. So so the
- 5:32market just just doesn't like any growth
- 5:34story right now. The market is broken
- 5:36and and and it's probably better that
- 5:38these companies don't do um don't do
- 5:41don't try don't try to cater to the
- 5:44market and just just just look at the
- 5:46business and and and walk on the
- 5:47business. Um okay so before I go into
- 5:50the valuation there's there's there's a
- 5:52there's a few things.
- 5:54There's a few assumption
- 5:56assumptions. So it's there's a there's a
- 5:58challenge and the challenge of course is
- 6:00is figuring out the next 12 months
- 6:01revenue growth. That's the challenge.
- 6:03I'll get to that in a second. Uh and I
- 6:05think I handled it pretty
- 6:06conservatively, but I'll get to that in
- 6:07a second. That's on the next slide. On
- 6:09this slide, I want to talk about a few
- 6:11assumptions
- 6:12uh before going through my valuation. So
- 6:14first of all,
- 6:15for for for there's a lot of FUD going
- 6:17against these companies saying that, you
- 6:20know, they're they're going to go
- 6:21bankrupt. Uh they're they're going to
- 6:23die.
- 6:24Uh and that's the FUD out there. And and
- 6:26and and of course, if you listen to
- 6:28these calls, you know that for so many
- 6:30of these companies, the capex gets
- 6:33funded with prepayments.
- 6:35If you listen to the Iron call, you also
- 6:38know that the CEO of Iron was talking
- 6:41about this in depth. They're like, you
- 6:43know, if we can't
- 6:45finance or we see if we see we're going
- 6:47to, you know, get into a a financial
- 6:50crunch or a cash crunch, we're just
- 6:53going to develop the projects more
- 6:54slowly. Like very that was amazing on
- 6:57the call. It was like we're we're not
- 6:58going to purportedly drive the company
- 7:01into bankruptcy. If we see that we
- 7:03actually can't develop a project, we're
- 7:05just going to put them on hold until we
- 7:07get the money.
- 7:09Put it on hold until you get the money.
- 7:10I know that's a crazy concept, but some
- 7:12people think that these companies are
- 7:15are just just just going to go into
- 7:17bankruptcy just because they want to
- 7:19meet their capex numbers. No, if they
- 7:21see they can't finance the next data
- 7:24center, they just won't and won't drive
- 7:26a company into bankruptcy.
- 7:28But anyways, one of my assumptions in
- 7:30this model is that the capex will be
- 7:32funded by the prepayments mostly. And if
- 7:34you listen to the calls, you know you
- 7:36know that it's it's mostly prepayments.
- 7:38If you listen to the Nebula call, that's
- 7:39pretty impressive how much they're
- 7:41getting in prepayments. And then they
- 7:43have a lot of GPUs that are unencumbered
- 7:46that they can use to back the loans. And
- 7:50then I suppose
- 7:52if these stocks get back to cheap
- 7:55valuations as opposed to super cheap
- 7:58valuations, if they get back to the
- 8:00being simply cheap,
- 8:02I suppose there will be some dilution,
- 8:04but you know, I suspect not a lot of
- 8:06these companies are going to dilute at
- 8:08these levels, these nonsensical levels.
- 8:10That's the assumption number one,
- 8:11assuming the CAPEX get funded.
- 8:13Assumption number two, of course,
- 8:15is the belief in EBITDA
- 8:17and the belief that EBITDA matters. And
- 8:20so to believe in EBITDA,
- 8:23you have to likely be born after the
- 8:241980s
- 8:26probably, right? The EBITDA was
- 8:27popularized in the in in the 80s by
- 8:30I can't remember his name, but some some
- 8:32some telecom telco magnate
- 8:36you know, who was who was noticing that
- 8:38that you know, the lines the cell phone
- 8:41lines were lasting much longer than the
- 8:43depreciation schedule.
- 8:44And that's why they started using using
- 8:46EBITDA and and and and and of course,
- 8:48the only thing we have to understand is
- 8:49that the D is fake.
- 8:51Warren Buffett is I mean, I think Warren
- 8:52Buffett is very wrong on this. Warren
- 8:54Buffett is an advocate who thinks the D
- 8:55is real.
- 8:56I think the D is fake.
- 8:59And and of course, if you're a
- 9:00traditional Wall Street analyst, you
- 9:01think the D is real.
- 9:03You think you think it's a real expense,
- 9:05but depreciation
- 9:07is a fake expense. And so so the case in
- 9:09point of course in GPU is is is this
- 9:12this sentence right here. This was so
- 9:14important when Iron mentioned that.
- 9:16You have to understand that outdated
- 9:19H100s were just renewed by Iron
- 9:24at two times the former locked price in
- 9:28a long-term contract.
- 9:31You also have
- 9:33and before I get to the Transformers and
- 9:34and all of the other miscellaneous parts
- 9:36of the building, but
- 9:38H100s renewed at two times the former
- 9:40price.
- 9:41When were H100s When did H100s become a
- 9:46thing? Oh, March 2022. March 2022.
- 9:51What was the initial depreciation
- 9:53schedule on H100s?
- 9:56If the new clouds were doing four years,
- 9:59a lot of the hyperscalers were doing
- 10:01three years.
- 10:02A three-year
- 10:03or four-year depreciation schedule on
- 10:05the H100, which means that in 2025
- 10:10the H100 according to accounting
- 10:13the H 100 should be worth a zero,
- 10:16right? In March 2025, if you had a
- 10:19three-year depreciation schedule,
- 10:21the accounting the accounting tells you
- 10:22H 100 is worth zero.
- 10:25Now, fast forward
- 10:28almost a year and a half later, you find
- 10:30out that Iron is
- 10:32renewing the H 100s at two times the
- 10:35price that they used to cost in 2022.
- 10:38What we've seen in race of AI is that
- 10:41not only did GPUs not lose value,
- 10:44they actually appreciated in value.
- 10:48And so, if that is the case for an old,
- 10:50supposedly outdated GPU, ask yourself
- 10:54memory. What what is the case for
- 10:56memory? What is the case for some of
- 10:58these racks? What is the case for the
- 11:00transformers that they have in all of
- 11:03these buildings that have also very much
- 11:05very short depreciation schedules?
- 11:08And so so so when you and now they
- 11:11depreciate over five year or four year,
- 11:14but but you know, you you know, I I
- 11:17I think in two years they'll still be
- 11:20renting H 100s.
- 11:22The the life expectancy of a GPU may be
- 11:26very very much longer than two years or
- 11:30three years.
- 11:31And so and
- 11:32and so what if what if what if GPUs have
- 11:34a useful value of of 15 years
- 11:38and these companies depreciate them over
- 11:39four years,
- 11:41then it makes absolutely no sense to use
- 11:43a four-year depreciation schedule when
- 11:45you should really use a 15-year
- 11:47depreciation schedule. And if these
- 11:48companies started using a 15-year
- 11:50depreciation schedule on on on on on the
- 11:52CapEx, mostly on the GPUs, right, which
- 11:55are a bigger part of the CapEx, if they
- 11:56were to use 15-year depreciation
- 11:58schedule, all of a sudden they would
- 12:00show a profit.
- 12:02All of a sudden, a very nice profit,
- 12:04isn't it? But we depreciate in
- 12:06accelerated depreciation schedules
- 12:08because
- 12:10I guess that's how the accounting works.
- 12:12And and so and so that's why on this
- 12:13channel I have never liked earnings.
- 12:15Like, like it's a phantom expanse. Why
- 12:18why should I consider an H 100 a full
- 12:22loss in March of 2025 when a year and a
- 12:25half later they rent it for two and a
- 12:28half the price again, right? Anyways,
- 12:31this is and and I realize that if if if
- 12:34you're if you're a traditionally trained
- 12:36accountant, you're not going to like
- 12:38what I'm saying here. But but
- 12:40nonetheless like like like like these
- 12:43last longer than two to three years.
- 12:45These last much longer than two to three
- 12:47years. And I think that's why using
- 12:49EBITDA makes a lot of sense for most
- 12:53modern companies actually. For a lot of
- 12:55modern companies it makes a lot of sense
- 12:57to use EBITDA.
- 12:58So anyways,
- 13:00now that I've talked about that, let's
- 13:02look at my super sandbag valuations and
- 13:05let's talk let's get SpaceX out of the
- 13:08way. So so you know my SpaceX valuation
- 13:10is very back of a napkin because I
- 13:13ignore everything in this. I even
- 13:15ignored
- 13:16Starlink
- 13:18and I just look at
- 13:20the JP Morgan estimate
- 13:22for the implied growth rate for for the
- 13:26for the SpaceX compute
- 13:29as
- 13:30of a Q4 2026 and I extrapolated. So I I
- 13:33went back to Q3 2026 and I forexed it as
- 13:36an ARR
- 13:37and I used today
- 13:39today's revenue as as well expressed no
- 13:42trailing 12 month. I used it I I used it
- 13:46trailing 12 months. Anyways, the point
- 13:48here is
- 13:50I'm getting 800
- 13:52percent
- 13:54growth on their 800 percent growth on
- 13:57their is what I'm getting on SpaceX. So
- 13:59if I'm assuming 800 percent revenue
- 14:01growth and remember remember what Elon
- 14:02said, right? You remember the number
- 14:04that
- 14:05Elon Elon was saying for for for the
- 14:07growth rate some something absolutely
- 14:10you know mind-boggling.
- 14:12So anyways,
- 14:14800 percent rate
- 14:16you know he he was saying between 500
- 14:17billion and a trillion, if you remember
- 14:19that. 800% growth rate. My EV over GP of
- 14:22RG is 0.07.
- 14:24Rule of 40 is 856
- 14:27856.
- 14:29So
- 14:30SpaceX to me is dirt cheap based on the
- 14:32computer loan.
- 14:33Now,
- 14:34Nebius, I'm using I'm using Nebius
- 14:37Nebius number. So
- 14:39the the um
- 14:41and by the way, I I got my thingy. I I
- 14:43don't have the next number, right? But
- 14:44the but this is the trailing 12 months
- 14:46that's that's 58 billion. Uh that's of
- 14:48course the enterprise value right here.
- 14:49That's that's where I was confusing.
- 14:51Sorry, I'm tired about that. But
- 14:52anyways, okay.
- 14:54So Nebius right now, if you look at
- 14:55Nebius ARR right now, Nebius ARR right
- 14:58now is is is is 1.9 billion. That's the
- 15:01ARR right now for Nebius.
- 15:02For Nebius, I'm using uh their number
- 15:06for for um Q4 to 26 to Q3 to 27 and I'm
- 15:11using 10 billion knowing that they
- 15:13guided 12 to 13 billion. So you have to
- 15:15you you have to think for yourself if
- 15:16I'm being too conservative or not. I'm
- 15:18saying
- 15:19I'm saying the the the ARR figure
- 15:22um at at Q3 of 2027 the the revenue
- 15:25figure not sorry not the ARR. This is
- 15:27revenue. We switched to revenue. I agree
- 15:28it's confusing. But we switched to
- 15:30revenue. I'm saying from from cuz they
- 15:32guided a total revenue for 27, right?
- 15:34They guided between 12 and 13 billion in
- 15:36revenue for 2027 which is from Q1 to Q4.
- 15:40So I I took that number and I and I I I
- 15:43removed it one quarter and I say they're
- 15:45going to do 10 billion in revenue
- 15:47between Q4 of 2026 right upcoming this
- 15:50upcoming quarter and Q4 2027. I'm saying
- 15:53they're going to do 10 billion in
- 15:54revenue. If I use that number, that's a
- 15:56forward revenue growth for Nebius of
- 15:58416%
- 15:59from today trailing 12 month revenue of
- 16:021.9 billion and therefore that gives you
- 16:05an EV over GP over RG of 0.07.
- 16:09Nebius in the way I value stocks
- 16:11continues to be more expensive than Iron
- 16:14although ever so slightly. And lastly,
- 16:19this is where it gets complicated.
- 16:21Is I want to move to Iron and Coreweave.
- 16:24But if you look at Iron and Coreweave,
- 16:26they didn't give me any Q3 2027 guidance
- 16:31or you know, what what would be November
- 16:332027. Uh
- 16:35these these companies, you know, they're
- 16:37kind of flying blind, but they both gave
- 16:39us
- 16:41Q1
- 16:42or the equivalent of Q1
- 16:45guidance. So so February, you know,
- 16:48February ended numbers, they both guided
- 16:51to that. And and so if you look at each
- 16:54of these February ended guidance,
- 16:58you have Coreweave who's guiding 21
- 17:00billion ARR and right now they have 9.8
- 17:03billion ARR, so that'd be 114%
- 17:05okay. And then you have
- 17:08Iron who's guiding for a whopping 4
- 17:11billion ARR by the end of February and
- 17:15right now they're at 900 million ARR, so
- 17:17that would be 337%.
- 17:20And if I use these two numbers, which
- 17:23which is crazy, right? Because because
- 17:25because I should be using the November
- 17:272027 number, but I'm using the February
- 17:292027 number. So so what's going to be
- 17:32the growth rate between February and
- 17:33November?
- 17:34It's a question mark. I don't know. So
- 17:36so this is extremely extremely
- 17:38sandbagged and using using this extreme
- 17:41sandbagging, I'm still getting, you
- 17:43know, only assuming they're going to
- 17:44grow the next three months.
- 17:47Even just using that, I'm still getting
- 17:49Iron at a 0.05 on EV over GPU over RG
- 17:52and I'm getting Coreweave at a 0.1 on an
- 17:54EV over GPU over RG and that's six
- 17:57months before
- 17:59the revenue number achieved by Nebius or
- 18:01by Iron. And so you may say, hey, well,
- 18:03that's a little unfair to Nebius.
- 18:06But here's another thing to consider
- 18:08about Nebius, of course, is their
- 18:10Clickhouse stake, which
- 18:12I mean to me, based on what Clickhouse
- 18:15is trading in the in the private market,
- 18:17their stake could be worth 15 billion.
- 18:19So actually actually,
- 18:21um, uh,
- 18:22given that, you know, I, I'm, I'm also
- 18:24sandbagging Nebius because I'm not
- 18:27removing the ClickHouse stake from the
- 18:29enterprise value of Nebius. When, you
- 18:31know, if I was, I would be getting,
- 18:33getting, getting a number even, even,
- 18:35even lower here. But then again, I'm
- 18:37getting a number for iron and core
- 18:39weave, which is presumably higher on EV
- 18:41over GP, uh, or over RGB because I'm not
- 18:44using the forward two months number. Um,
- 18:48anyways, I, I, I, I, I hope you see why
- 18:50this is not that easy to, to estimate
- 18:53the value of these companies. Um, and
- 18:55you know, I, I agree with people, you
- 18:57know, I understand why, why Nebius is
- 18:59moving to revenue. Like I get it.
- 19:01But it's, it's, it doesn't make it easy
- 19:05when iron and core weave are still, are
- 19:06still using AR. And so, and so, so I
- 19:08think Nebius is forward looking on this.
- 19:10I think, I think they're leading the
- 19:11pack on this. Um,
- 19:13the point I'm trying to say is that
- 19:14everything's too cheap.
- 19:16Everything's dirt cheap.
- 19:18Now the problem is
- 19:20there's a lot of dirt cheap stocks right
- 19:23now. And there's a lot of stocks that
- 19:26don't trade at all on fundamentals. I
- 19:29have never seen so many cheap stocks.
- 19:33There's, it's insane how many cheap
- 19:36stocks there are out there.
- 19:38And you know, this is a market that has
- 19:40entirely forgotten about from a number,
- 19:42fundamentals and just trades on, on, on
- 19:44vibes, on, on X sentiment and and, and,
- 19:48and you know, it's just, it's just
- 19:50crazy. It's unfathomable how bad 2026
- 19:54has been.
- 19:55And it continues to be bad. Like if you,
- 19:57if you really think about it, I mean, we
- 19:59had, we had Nvidia, Nvidia Black Swan
- 20:03for any Nvidia shareholder, the biggest
- 20:05Black Swan of them all for any Nvidia
- 20:07shareholder is what happened early this
- 20:10week, right? And, and, and, and today,
- 20:13you know, five minutes before the drop,
- 20:15we can't even get a relief stocks dump
- 20:18and after our stock dump because of
- 20:21uncertainty about the ground halt in the
- 20:23United States and all planes not all
- 20:25planes not being able to fly because of
- 20:28some hack of the radar system or
- 20:30something like that. So so it's
- 20:32relentless. This market has been
- 20:34relentless. You know, I'm I'm kind of
- 20:36hoping the catalyst is the Armatic IPO.
- 20:39Maybe now there's a new catalyst on the
- 20:40horizon which is a Armatic IPO next
- 20:43Thursday. Maybe that's what we need. I
- 20:46really don't know. I really don't know.
- 20:47It's it's it hasn't been an easy market
- 20:49to navigate.
- 20:50So this was not investment advice. This
- 20:52is not financial advice. This is only
- 20:54entertainment. I'm hoping you were
- 20:55entertained. Please like, please
- 20:57subscribe, follow me on Patreon, follow
- 20:58me on X. Thank you for watching and have
- 21:01a wonderful wonderful day.
About this transcript
This page contains the full transcript of Neocloud Bloodbath: Updated Valuations, Assumptions, EV/GP/RG & thoughts for NBIS, IREN, CRWV, SPCX. by Beat The Denominator, generated from the public captions YouTube serves with the video. The transcript has 3,628 words across 565 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.
What you can do with it
Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.
Free YouTube transcript tool
YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.