Thread 106572090

This is next nvidia thats all im gonna say. In next few years these guys will take semiconductor world by storm and you'll realize schizophrenic anon on /g/ might've been right.

>>106572090 (OP)
Use case?

>>106572167
AI, medical devices, smart sensors, robotics and essentially everything tech.

https://files.catbox.moe/994l39.pdf


Read this article and you'll realize what is coming

>>106572090 (OP)
Explain why in 5 sentences. And no, I will not read any links you post.

>>106572090 (OP)
Japs

Private?

>>106572414
yes
>>106572349
>AI is reaching bottleneck
>energy wise, hardware wise and software wise
>current computer architecture won't achieve AGI
>Memristor chips allow In memory compute
>1000x energy efficient and allows people to run AI on local computers
>Applications in Brain interfaces, health devices and drones/robotics

>>106572185
>pdf
not today, identity thief.

>>106572495
It's this article and i uploaded it cause it's behind a paywall

No stock... why do I care?

>>106572466
I've seen more of these vaporware techs than you know. If it's real the company gets bought and it's the bigger company that gets the main benefit, but more often than not it turns out to not be commercially viable and dies.

>>106572466
damn, I'm sold. selling everything to invest in this. gonna be rich af. thanks bro. With 3 asians in charge it can't possibly fail.

but for real tho it does seem somewhat legit. https://www.synopsys.com/cloud/tetramem-success.html

>>106572517
>>106572514
I know you think im bullshitting but plz read this article
>https://files.catbox.moe/994l39.pdf

They created a device that reads your brain signals, adapts overtime and no gpu/cpu at all, it was just memory compute. It currently only works with small models but i think you can see how this will pan out especially with compute on edge. Instead of having one big processor robots might have models in their legs, hips, spine and overall body creating much more fluid motion and less energy use. Once you read the paper and understand what lies ahead isn't this hype LLMs but smart, local and cheap models that can do things in your home, health, industry and many other uses where specific models are warranted. You won't need a bunch of ram, gpu, cloud access but can run it on 14nm chips.

>>106572579
I never said 'real', I said 'commercially viable'.

>>106572579
okay nevermind I did say real my bad

>>106572466
lol
lmao even

>>106572090 (OP)
Never heard of them but I just put my home on the market and will dump the proceeds of the sale and my entire life savings and 401K into their stocks. Thanks for the tip, anon!

I'll invite you all onto my yacht when it moons!

>>106572587
>>106572583
not a problem but it's commercially viable when Billions of dollars in VC actually get poured into it. Lot of people in AI are realizing that massive energy costs are unsustainable and scaling constantly will inevitably lead them bankrupt. These guys are literally set up in perfect spot to end up receiving billions of dollars in investments for their ideas since they're not only working on design but material side as well so they have it all ready plus they're connected to TSMC,Micron, IBM, global foundries and Synopsys. This is literal 2010 bitcoin tier where no one actually gave bitcoin time of day.
>>106572617
laugh but in 5 years you'll see faggot
>>106572633
it's private im not shilling any stock

>>106572638
saar if you want bag holders go to /biz/ saar

>>106572674
? there's nothing to sell

>>106572090 (OP)
>>>106570952
I'm seeing a pattern here. Analog circuitry is perfect for matmul. Will follow this and it will be interesting to see how this evolves.

>>106572506
Post the archive.org link.

>>106572185
>https://files.catbox.moe/994l39.pdf
problem with these fancy academic papers is that they never work out as promised. There are tons of papers like these but they either get no further funding after the initial presentation or they are hiding some obvious problem like scaling issue.
t. some dude who had to babysit a couple startups who got scammed by my prof into funding his research for biosensors

It's a private company. I cannot buy its stock. How do I become rich?

>>106573736
IBM has memristors circuits

>>106572466
Samsung is the leader in memristors

>>106573899
Become a venture capitalist and know people.

I don't trust that pdf chink info. I need Anton Petrov to explain it to me like you would a 5 year old retard child.

x86 cpus will always be smarter and faster than this meme shit

>>106574829
I will believe them when they release a real product. So far its the same scam as QC.

>>106572090 (OP)
Unless its a Chinese company, im not investing in shit lol.

All American tech companies are essentially on life support until they get bought out.

>>106572517
Ok nvm, their entire board is Chinese I think I might give this a try.

>>106576558
aannnnd their private company so I can't invest....

>>106576558
taiwanese* know the difference anon
>>106576566
invest into companies working along with them or what might ended up rising along with it.

>>106572185
>2025
>pdfs
God I hate this gay world. html/css may be bad, but at least it's "responsive"

>>106572185
>buzzwords
This is bearish.

>>106572090 (OP)
Is this your company? Do you work for them? Why are you shilling on 4chan?

Ima short sell your company, I say this because it’s American based therefore you will be forced to hire trannies which will make you non-competitive

>>106572466
There is no government support unlike China
Commies will just steal the idea and will do it 1000 times cheaper and faster without all the trannies