Zephyr fanboy. Bagholding memory. Not investment advice
Tejas
Joined November 2019
- Tweets2.6K
- Following648
- Followers18.9K
- Likes24.6K
As you can see, $30M wasn't enough 😂😂
strictrope retweeted
Neoclouds have to go to spot pricing for them to make sense I think like the All In guys said. Borrowing at higher rates vs SpaceX, deploying slower than SpaceX, and taking in lower revenue than SpaceX per watt doesn't feel smart. You can at least fix one of the three.
The additional bonus for SpaceX is that they have their own models they can serve with Grok/Grok Bot, so in the case where they overbuild, they can still serve their models/repoint compute for training bigger and better models.
Neoclouds probably become acquisition targets for SpaceX in a couple years to restructure their debt at this rate, especially if SpaceX continues to soak up more compute and more capital.
strictrope retweeted
so $SPCX obviously blew the doors out with those earnings
especially if Elon gets anywhere close to his goal of bringing 20GW online by next year
the question now seems to be what this means for every other neocloud
SpaceX is up today, the rest are flat to red.
On one hand, $SPCX proved how important the neocloud business model is and how much demand is out there.
On the other hand, $SPCX was so good that now people are doubting the ability for other neoclouds to easily finance their operations, how quickly those companies can bring compute online compared to Elon, and how much debt/dilution SpaceX themselves will have to take on to finance that buildout which then begs the question around how much dilution the smaller players will have to take on. SpaceX is also profitable whereas the others are years from profitability.
My take is more that $SPCX has validated the entire concept of neoclouds which obviously have been under stress given the $CRWV bonds getting hit over the past two weeks, but the earnings were also so good that there might be more questions now about how well the others can execute.
The broader theme seemed to be legitimized from those earnings: the compute constraints exist and have not eased up in any meaningful way.
Compute will get commoditized, but hardware will be forever. Airlines vs airplane manufacturers
Ken Griffin of Citadel questions the economics of the AI infrastructure boom:
›What happens when GPU supply exceeds demand & pricing goes into free fall?
›Are today's semiconductor margins sustainable? Are investors overestimating the useful life of GPUs?
“Look, I think this market is actually experiencing people doing research.
For the couple of years, if you were on the thematic program of we're gonna spend hundreds of billions or a trillion dollars in data centers, we're gonna build compute at a scale never seen before, which companies in that story are choke points in supply chains that will have abnormally high margins? And who will be the big winners in this transformation?
Will it be the Microsofts & AWSs who will be the gatekeepers to the foundational models? Like, who's going to earn the economic rents?
But in some sense, like, 'I'm just gonna buy them all. I'm gonna buy the hyperscalers, I'm gonna buy the chip, I'm gonna buy businesses that make things I've never even heard of 5 years ago in Korea and Taiwan & in Japan.'
And now we're reaching a point where people are becoming more discerning.
We do realize that we are seeing extremely high margins in DRAM & MLCCs in areas where there are real supply constraints, but people are now starting to look out 2 years, 3 years. How long is this going to last?
And in particular, I don't wanna be doom & gloom, but GPUs, like airlines, have a very low marginal cost to run.
The incremental seat on an airline that's not full, if you can sell that ticket for $20, you're making money in that seat.
And that's one of the reasons that the airline industry is such a nightmare, is because the moment you have overcapacity, pricing goes into free fall.
We saw this actually with the telecom build-out back in 2000. We built more fiber than we had demand for, & the price of telecommunications just went straight down.
If we reach a point where the provisioning of GPUs exceeds demand for GPUs, marginal pricing will actually go into free fall.
And that really upends all the investment assumptions that people are working with today.
That's issue number one.
And issue number 2 is:
I think people are taking a step back & saying, 'What is the useful life of a GPU?'
Right here, right now, there's people that are earning extraordinary pricing on GPUs that are 5 or 6 years old.
And so there's a growing set of concerns that I can't give the same capitalization rate or the same revenue models to year 7, 8, 9 of a chip in the future that maybe I was using 12 months ago.
So I think we're really grappling with just how far will this demand carry us in the marketplace.”
I’m saying that overtime as more providers spend on hardware to bring more compute online, neoclouds become riskier because their business only works when they have positive leverage (NOI > debt service). As more compute comes online (higher supply), that spread between net operating income and debt shrinks. It’s really just real estate.
What I’m not sure of is how long it will take for that to happen given the demand but lots of people are building compute to capture value so directionally I’m confident. While this is happening tho, pick and shovel providers swim in cash.
strictrope retweeted
SpaceX is effectively the best neocloud right now...
SpaceX insists on short term leases only, so as long as compute keeps getting scarcer and the compute shortage curve keeps sloping up and to the right, SpaceX can keep raising prices.
On top of that, SpaceX has the best infrastructure buildout capability in the industry and its own compute demand, so it can absorb excess demand at any time. That is why it commands a premium of 2x, and as much as 3x, over competitors like IREN. ($60 billion per GW)
To track Nvidia's revenue, SpaceX is now the company to watch most closely. Until SpaceX ASICs come out of Terafab in 2030, it will devour more Nvidia chips, and more ravenously, than anyone else in the world. Wouldn't SpaceX easily absorb the slowdown in Nvidia chip demand caused by Google and other hyperscalers developing their own ASICs?
Feel for the guy, but this is why I don't use margin or sell options. Anything where you can potentially owe money if the market moves against you is scary to me. When you buy calls or puts at least, your risk is contained to 100% loss of your premium.
Having to ask my wife for money to pay bills. For a few months now, she has had a separate bank account because I drained ours to cover margin calls but I’ve now drained my own bank account to the point that I don’t have enough to cover bills. So now I have to see if she can transfer money to my account to pay the bills about to hit because these markets aren’t recovering enough for me to get out of margin call and put money back. Exhausting stuff. Everyone is getting wiped.
👀👀
Taking the stage at @TomorrowXSummit: Stan Druckenmiller (@standuquesne), Chief Executive Officer of Duquesne Family Office, and Iliana Bouzali (@ilianabouzali), Global Head of Derivatives at Morgan Stanley (@MorganStanley) $MS.
November 17-18 at the Moody Center in Austin. Hosted with @GavinSBaker and @atreidesmgmt, @antoniogracias and Valor Equity Partners, @rbiscardi and @iconnections_io.
Registration link below.
Hyperscalers aren't spending trillions of dollars building compute infrastructure just for LLMs. If you're bearish Nvidia and/or memory cos, you should probably research what those other model types are.
The bubble @bubbleboi is trying to pop
My quant is a dog
Bottom line
ERCOT’s additions plus recips, converted miners and smaller sites should more than cover the ~21 GW of new US power I think is actually needed. Even the full ~40 GW gap looks far more manageable than the market fears. So power is a solvable constraint, even if we don’t assume regulations ease after midterms or people rely on outside sources more.
The neocloud business model is the oldest and most time tested model in history. You buy assets, and you rent them out. Economics will normalize once supply and demand find equilibrium, but power constraints, accelerating compute demand, and existing shortage won't allow that to happen for some time imo. Cost of capital will come down over time with more supporting data. Just seems like a novel business rn, but its not.