Developing Investors with @fink_money | “Anything can happen and it probably will” | Everything’s Gambling | Risk Ain’t a Dirty Word
Spain
Joined July 2016
- Tweets40.1K
- Following2.1K
- Followers8.1K
- Likes45.9K
Pinned Tweet
👇👇👇
This Thursday at 12pm, we’re running a free group call, mainly for British traders and investors.
We’ll be hammering home a few key points about the upcoming market and hedging political insanity.
Like this post and I’ll drop you the link.
Replay will be available but if you don’t show up live I won’t be able to answer your questions 😉
Watching a few accounts greedily buying every dip in beaten down low quality AI names
I’m sure it’ll work out fantastically for them
If Grok crashes today, it's my fault.
Gave it two paragraphs of a McElligott note with instructions to "simplify this"
Tim retweeted
Quick 3 mins on a small part of how I look at volatility for both risk management and execution
Lots more to it but perhaps this fills in a blank
Tim retweeted
'irrational selling'
There is nothing irrational about selling.
The market is the market and not reality.
Over the long term the biz case matters but the short term, you're at the mercy of market dynamics/mechanics which are not irrational, they just 'are'.
Tim retweeted
We don't actually make any promises that you'll make loads of money.
We promise that you won't act like a total c*nt which will enable you to make loads of money.
Structure & clarity > false hope of riches.
Tim retweeted
Barlcays CEO on the Bank levy increase.
Savers will end up paying it by reduced rates.
Which might actually be a good thing since people might look to equities for a greater return instead, so I am withholding judgment.
Tim retweeted
$OPEN shareholders, this one's for you.
The market doesn't understand or you don't understand?
xitter.cf/VolaTim/status/2113277…
The bear case for neoclouds is ridiculously obvious, and not new information.
TLDR: Neoclouds are leveraged GPU landlords, borrowing expensively to buy chips that lose value fast.
They rent those chips to a handful of customers, many of whom are also rivals with access to far cheaper capital (like hyperscalers and SpaceX), that will compete for the same business while building out their own self-sufficiency
Neos have been able to print cash as 'temporary capacity buffers' while compute demand massively outstripped supply.
But market is forward-looking and the future doesn't look so bright.
- Returns sit below their cost of capital, even on generous GPU lifespans
- Debt & debt servicing costs are piling up: CoreWeave's interest takes 42% of its EBITDA
- GPUs lose value fast, and hourly rental rates can halve with each new chip generation (chip value is important if you're using those chips for asset-backed financing)
- A few customers make up most of the revenue, and those customers are mainly loss-making frontier labs and massive competitors like...
- Hyperscalers and SpaceX, with big balance sheets and access to cheaper capital
- Required capex runs above revenue, so future growth depends on ever more debt or dilution = death spiral
$CRWV $NBIS
The bear case for neoclouds is ridiculously obvious, and not new information.
TLDR: Neoclouds are leveraged GPU landlords, borrowing expensively to buy chips that lose value fast.
They rent those chips to a handful of customers, many of whom are also rivals with access to far cheaper capital (like hyperscalers and SpaceX), that will compete for the same business while building out their own self-sufficiency
Neos have been able to print cash as 'temporary capacity buffers' while compute demand massively outstripped supply.
But market is forward-looking and the future doesn't look so bright.
- Returns sit below their cost of capital, even on generous GPU lifespans
- Debt & debt servicing costs are piling up: CoreWeave's interest takes 42% of its EBITDA
- GPUs lose value fast, and hourly rental rates can halve with each new chip generation (chip value is important if you're using those chips for asset-backed financing)
- A few customers make up most of the revenue, and those customers are mainly loss-making frontier labs and massive competitors like...
- Hyperscalers and SpaceX, with big balance sheets and access to cheaper capital
- Required capex runs above revenue, so future growth depends on ever more debt or dilution = death spiral
$CRWV $NBIS
What's really mind-boggling is that there are so many credible bearish theses one could make against neoclouds as a group.. Off the top of my head:
- Business is extremely capital intensive.. And they're competing against companies (hyperscalers) with the biggest balance sheets and operating cash flows in history (e.g. a lower cost of capital).
- As things stand today, much of aggregate neocloud revenue and backlog is concentrated across a handful of their competitors (hyperscalers) and frontier AI labs. Not ideal when your competitors represent a big chunk of your revenue and could opt against renewing those contracts.
- Frontier labs also clearly want to vertically integrate which would reduce the need for them to buy from neoclouds
- Most neoclouds are extremely reliant upon one single supplier for the infrastructure their entire business runs on.. Supplier concentration is no bueno.
- The "new contracts are being done at $40MM/MW" headlines suggests neoclouds are over-earning.. New, well capitalized entrants will drive down returns (cc: Blackstone + Google JV, SpaceX, etc.).
Yet for some reason people like OP resort to just making shit up. Frame their claims as facts when in reality the claims are entirely false. Pure gaslighting. Almost have to think they're being paid to do so because they can't be that stupid?
