What’s frustrating to me is the market is asking a question about financing, for the AI trae. It has nothing to do with usage or the models themselves. So it doesn’t matter if the models are improving or usage is going up. If financing is still a concern, and the return on capital is still a concern, you’re just talking past each other.
You're telling the market that the amusement park is packed, but they are selling tickets for a lost. The market is questioning the business model yet you are pointing still at ticket sells.
That’s what’s frustrating about seeing people bring up OpenRouter stats or another model release. Okay, great. Usage is going up. How does that address what the market is worried about? What does it cost to finance all this capacity? What return does that capacity earn? And how long do shareholders have to wait for the cash?
The technology can keep improving while the economics for the people financing it get worse. Demand doesn’t have to collapse for these stocks to go down. If financing gets more expensive, projects become harder to fund, it’s a duration issue, so please stop posting P/E for semi names, i beg.
And you have to pay attention to the reaction, right? If these narratives keep coming out and the stocks aren’t catching a sustained bid, maybe the market is looking in the other direction. You’re making a point, but it isn’t answering the question the market is asking.
This has been going on in one form or the other all month, wish i shorted harder when i saw all this cope.
I'm seeing more fear and less cope on the TL.
I'm not eager to long AI names in meaningful size still. But i guess the days you don't want to buy is the day you should so maybe inverse me.
But, you know, take care of yourself. Do not try to make everything back in one day or one trade or whatever. It's really a mental game. And if you don't have your mental health, you don't have anything , stay safe.
Sold most probably selling the rest of the shorter stuff after this post.
Any attempt to rally the name i fade it tbh.
Thanks for playing
LOD
$SPCX
Taking the trade will be somewhat difficult sense its a cult stock. bounced off 150.86 todays current low.
but i just don't see how rallies are not faded on tis name. if you are a short term trade enjoyer could be something to add to your playbook.
or wait for a pop and buy some spreads+time
I know you don't short cult stocks but hear me out.
All things being equal, if cost per gigawatt stays flat say 50B/GW
$SPCX needs 500B to hit the 10GW target in 2027.
Now, the issue is where do they get this funding from, right? It probably has to be a mixture of existing cash, cash flow, equity raises, debt.
recently the market has killed the neo clouds due to cost of financing concerns.
And on top of that, we have the batches of shares become sellable
today 1.3B (28%) and Monday 328MM shares.
i just dont see how this stock doesn't go back to 110-120 over the next month min.
this feels like an $ORCL
I know you don't short cult stocks but hear me out.
All things being equal, if cost per gigawatt stays flat say 50B/GW
$SPCX needs 500B to hit the 10GW target in 2027.
Now, the issue is where do they get this funding from, right? It probably has to be a mixture of existing cash, cash flow, equity raises, debt.
recently the market has killed the neo clouds due to cost of financing concerns.
And on top of that, we have the batches of shares become sellable
today 1.3B (28%) and Monday 328MM shares.
i just dont see how this stock doesn't go back to 110-120 over the next month min.
this feels like an $ORCL
Rough action, people are using every rally to sell, distribution across many names, if you are holding anything toxic gotta take profits quickly in this env. talking about stocks here, the index looks fine as usual lol
Anyways for the ones who are new. My wife spent a fair bit of money in Korea and Japan.
So we need to lock in for the amax bill
sorry i could not expand on this here is the reason why i think it will take a day or two, we also saw this earlier this year with muse, so yes i'm making a bottom call here.
So what @zephyr_z9 is basically saying is we can inference 80-90% GM on inference, so if that is true that can undercut the current narrative 'neos can't finance they going to 0 ergo sell today'
So i'm borrowing heavily from the work of Tversky /Daniel Kahneman price anchoring.
it will take a day or two to proliferate throughout the market 'anchoring and insufficient adjustment'
People hold onto the belief that neos can't finance and will only update priors partway when a new narrative comes into the picture.
basically not enough people have this in their mental model today, and it will take some time
anyways i'm long $NBIS caught some nasty fills today
If true, might take the market a day to figure it out.
Then i would guess today is the local bottom for neos.
Inference Gross Margins - 87.76% to 96.94%
Assumptions:
97% cached input, 2% uncached input and 1% output tokens
Prefill: 80k uncached input tokens/s/GPU.
Decode: 16k output tokens/s/GPU (250 tokens/s/user)
Token mix: 970M cached input + 20M uncached input + 10M output per billion total tokens.
It takes around 875 GPU seconds (250 for prefill & 625 for decode) to process 20M uncached input + 10M output tokens.
Assuming GPU rental cost is around $3-$6/hr, the GPU system cost per billion tokens is around $0.73-$1.46
1B tokens generate $11.91 in revenue at off-peak prices and $23.82 at peak prices
At off-peak prices, inference margins are between 87.76%-93.88% depending on the GPU rental cost ($3-$6/hr)
At peak prices, inference margins are between 93.88%-96.94% depending on the GPU rental cost ($3-$6/hr)
I hope this helps.
federalreserve.gov/monetaryp…
Do not allow yourself to get too high nor too low. It's been a pretty difficult tape this month, but I want to try to articulate my thoughts, but I'm kind of having a hard time.
I'll try to put something out this evening, but right now, something is telling me the vibes are off. Even after the beating in AI names last week, I still have this voice saying danger and I just can't shake it.