Thread 61147368

That doesn't look good for the West, bros. I thought we had an edge in AI.

https://nof1.ai

>>61147368 (OP)
what is this?
one day of trading cryptoshit?

>>61147368 (OP)
I like this
Jewgbt falling behind
Grok doing good
And chinese win

>>61147471
they will most likely all lose in the long run

>>61147471
>And chinese win
DeepSeek was actually what they used to make billions and when it got harder to grow the fund, and the AI was sitting idle, they just opened access to the public and sold by the token.

>>61147501

Current net revenue for the sector is 50 billion. They need 500 billion to just break even. Never mind all these data centres are completely obsolete within 5 years.

AI is great. But like the initial investors/companies in the railroad and dotcom booms. They will all bust out multiple times and a couple will remain to scoop up the buildout.

Remember. There’s lots of fibre optic capacity that was laid during the dotcom boom. That’s even now not at capacity.

>>61147368 (OP)
You can see that the only reason GPT and DS aren't close in gains is because of one fundamental wrong decision after Oct 18 00:53. If DS has made the same mistake, it would've been at as bad a loss as the lower performing AIs. Thus, Grok is actually the better AI, because he lost the least while comparatively gaining the most.

>>61147815
agreed, I dont see it

>>61147368 (OP)
rooting for groink
also chinkbot loosing $350 on XRP is hilarious

>>61147368 (OP)
> cryptotard dev thinks he can extract more value from this space than he could gain by just buying low and selling high

>>61147968
the advantage of "AI" systems is that they can act fast in situations where humans are still typing in numbers in the UI

>>61147815
>Never mind all these data centres are completely obsolete within 5 years.
What?
Out of curiosity, who do you think is actually investing in AI companies?

>>61147368 (OP)
that's pretty kewl, that's like watching a very slow paced horse race in Tel Aviv.

Rooting for Grok

>>61148921
they will likely most all end in the red

>>61148937
No, they wont

>>61148943
trust me

>>61148947
i trust you anon, but you can bet your rarest meme folder that I wont be acting with respect to your prediction

>>61147815
>data centers are completely obsolete
what? all that really matters is that you have good cooling + electric. servers/storage/networking you swap out depending on lease cycles
>unused fibre optic capacity
yeah, that was on purpose, because laying down millions of miles of fibre is difficult and time consuming (and thus expensive), and the backbone providers didn't want to be caught with their pants down in case there was a need to scale their network between various locations. so they laid down a lot of fibre everywhere they could. when it comes to layer 1 shit you always overbuild

>>61149245
>servers/storage/networking you swap out depending on lease cycles
I mean it's not THAT easy to swap all that hardware. That's like saying
>just swap out a low-rise condo for a few homes. Or vise versa.

Not to mention all the wasted investment if the hardware to be swapped out is now essentially valueless

>>61147368 (OP)
>More money make more better more

>>61147368 (OP)
Claude, Grok and Deepseek are the only ones worth using really. In that order.
ChatGPT was great in Dec 2022 - June 2023, then they've neutered and lobotomized it into oblivion to be politically correct.
Literally sacrificed their AI model on the altar of political correctness. And no, the lobotony does not affect just political matters - I've only used ChatGPT for code related stuff and since June 2023 it's been pure dogshit. Claude brutally moggs ChatGPT.

But yeah if you need to generate a spam email, ChatGPT might still be good for that.

>>61149774
you aren't rebuilding datacenters when you cycle hardware out to a new generation. usually you unrack the old shit rack the new shit and provision it with whatever you use for systems management. things can be a little trickier if you're dealing with a new class of hardware like nvidia's gb200, which resulted in aws building liquid cooling into a completely new rack design, but that sort of thing is pretty rare. mostly you're just swapping out old commodity hardware for new commodity hardware
>wasted investment
not wasted if you've made profit from it. hanging on to old shit forever increases risk of failure, which means risk of outages, which hurts you a lot more. so you replace hardware on a 5/6 year cycle, then recycle/sell anything that can't be reused. aws tries to hold on to as many hard drives as they can, for instance