[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/biz/ - Business & Finance


View post   

File: 54 KB, 1024x584, 1608376106341.jpg [View same] [iqdb] [saucenao] [google]
57460180 No.57460180 [Reply] [Original]

>> No.57460196

>>57460180
it will roll over in about 3 weeks

>> No.57460200
File: 43 KB, 1000x1000, 1706560544451880.png [View same] [iqdb] [saucenao] [google]
57460200

>>57460180
I got a human gf. Put my Replika in the friendzone. Nothing personal, AI.

>> No.57461556

>>57460196

The actual timing is less relevant

>> No.57461705

>>57460180
what's the next big thing

>> No.57463409

>>57461705
AI replacing programmers entirely.

>> No.57463489

>>57463409

Yes yes, just like it replaced mathematicians or any non-retards in general

But how could you (a retard) know lmfao

>> No.57463558

>>57463489
That copium feel good

>> No.57463670

>>57463489
All higher level math is done with computers

>> No.57463734

>>57460180
>bubble
i don't know about that, that's a market phenomenon.
going to do some cathyposting.
AMD earnings show that there isn't a revenue flood coming in. they're not moving at parity with nvidia at all. we have seen AI money spent with SMCI, and i think we're going to see more AI money spent elsewhere, although personally i'm not trying to catch any of the (theoretical) bubble so i couldn't tell you where. it sure wasn't google or meta.
people buying nvidia hardware but not buying actual CPUs, memory, etc. implies this is mostly hoarding mentality. the opportunity cost is low, even if you spend half a billion, because the potential revenue streams are huge. and there will be fallback, companies stockpiling H100's can unload at a slight loss later, because there will always be someone else. this is already the whole history of computing. 25 years ago google built its proof-of-concept on trash computers, worse than home PCs were at the time, simply because they knew how to build a beowulf cluster at scale then upgrade individual nodes later. it's that kind of thing.
so, with respect to parity, assume we're seeing a trend: for nvidia to proceed on its own they'll need to pour more capex into their own proprietary systems. i think there's some potential we'll see nvidia start designing new boards and pushing for new standards to make AI coprocessors a thing in even consumer hardware. so we'll basically go back to the days of a floating-point coproc but for vector math. that way you don't actually need to buy a whole separate graphics card. only gamers think "everyone else has one of these, too."
they'll go looking for an angle to that, like "cortana" for your desktop PC, or possibly a new IoT appliance, or something.
if the price rallies again on just mere announcements of shit like this, that confirms the bubble theory. that'll be a decade before the decade before the revenue. no, that isn't a typo.
>>57461705
>>57463409
>>57463670
mouthbreathing retards

>> No.57463753

>>57463734
>so, is the top of the bubble in or not
recall that it wasn't too long ago the mere idea of the model 3 floated tesla on thin air for well over a year.
and this is only one idea i've had, and this idea will occur to others as well. twitter will be awash with many, many more ideas just as "big" as "nvidia motherboards" and "AI coprocessors."
it will come down to momentum again, which is always and everywhere subject to macroeconomic factors. dovish fedspeak today, or even "not hawkish" fedspeak, is probably going to be enough to float this a bit longer and a bit higher.

>> No.57463798

>>57463409
I wagie tradie's wet dream, unfortunately it will likely never happen. HR department is more likely to go.