Thread 106623113

https://nvidianews.nvidia.com/news/nvidia-and-intel-to-develop-ai-infrastructure-and-personal-computing-products
https://newsroom.intel.com/artificial-intelligence/intel-and-nvidia-to-jointly-develop-ai-infrastructure-and-personal-computing-products

AYYMDEAD IS FINISHED & BANKRUPT

Wow. Looks like Intel found its client.

why the fuck would nvidia want to keep a dying company afloat.
I bet intel are getting peanuts out of this and its a panic measure

>>106623203
They are playing buddy with Intel in hopes they can license x86 in the future. AMD can do x86 but not Nvidia.

>>106623113 (OP)
mfw my 31 shares @ $39 still under water

ARMFAGS ON SUICIDE WATCH

>Intel shares are soaring (30%) in premarket trading
>Data Centers: Intel will build NVIDIA-custom x86 CPUs, which NVIDIA will integrate into its AI infrastructure platforms.
>Personal Computing: Intel will produce new x86 SoCs incorporating NVIDIA RTX GPU chiplets, powering next-gen PCs with combined CPU-GPU integration.

>>106623113 (OP)
>Intel to design and manufacture custom data center and client CPUs with NVIDIA NVLink
>CPUs with NVLink
Isn't NVLink for GPUs? The fuck does this even mean. Also didn't NVIDIA kill off all NVLink support for their consumer GPUs? Why include this on client CPUs

>>106623203
>dying company
It was never dying just look at the fucking market shares, Intel is a giant. It was just stagnant and you're basically fucking retarded because you bought into a very basic FUD. Intel was at $20 per share, way below book value for a YEAR and you didn't take that free money because you literally believed one of the biggest superconductor outfits on the planet is "bankrupt" with 70% of the global server market and 80% of the mobile CPU market

Stop investing, man, you're too stupid for it.

>>106623211
>license x86
You mean the ISA that nobody uses anymore? NVIDIA using x86 would be a major step backwards. AMD holds the license to x86_64/amd64 which is infinitely more relevant.

>>106623113 (OP)
Obvious Trump setup this shitty deal in order to save intcel

Just give NVIDIA an x86 license, we need to drain the swamp of x86 duopoly.

If we had a sane admin in DC they would force intel make the X86 license public in exchange for the $10B bail out.

This is a shitty deal for Nvidia and they'll never get these $5B back.

>>106623531
>Obvious Trump setup this shitty deal in order to save intcel
>
>Just give NVIDIA an x86 license, we need to drain the swamp of x86 duopoly.
Shitel is pissrael's steaming feces of a crown jewel in tech they will never let it go under it is why gobmit bought 10% of it. The absolute state of the tech industry is due to jew hubris.

>>106623442
x86 = x64

>>106623575
x86 is not x64 and Intel doesn't own the license to give out rights to x86_64 (aka x64, AMD64, Intel64), AMD does. The only ISA Intel owns that's 64-bit is IA-64 which is dead.

>>106623442
Stop being a pedantic faggot

Well, RIP Intel graphics then, even integrated, hope we will enjoy the "duopoly"

No seriously how does AMD cope?

>>106623404
>Why include this on client CPUs
To kill it again obviously

https://www.tomshardware.com/pc-components/cpus/nvidia-and-intel-announce-jointly-developed-intel-x86-rtx-socs-for-pcs-with-nvidia-graphics-also-custom-nvidia-data-center-x86-processors-nvidia-buys-usd5-billion-in-intel-stock-in-seismic-deal
>The products include x86 Intel CPUs tightly fused with an Nvidia RTX graphics chiplet for the consumer gaming PC market, named the ‘Intel x86 RTX SOCs.’ Nvidia will also have Intel build custom x86 data center CPUs for its AI products for hyperscale and enterprise customers. Additionally, Nvidia will buy $5 billion in Intel common stock at $23.28 per share, representing a roughly 5% ownership stake in Intel. (Intel stock is now up 33% in premarket trading.)
>For the PC market, the Intel x86 RTX SoC chips will come with an x86 CPU chiplet tightly connected with an Nvidia RTX GPU chiplet via the NVLink interface. This type of processor will have both CPU and GPU units merged into one compact chip package that externally looks much like a standard CPU, rivaling AMD’s competing APU products.
How is this legal? NVIDIA is straight up giving Intel CPUs a faster link to GPUs. HOW IS THIS LEGAL?

>In either case, Nvidia has been mulling using Intel Foundry since 2022, has fabbed test chips there, and participates in the U.S. Defense Dept.'s RAMP-C project with Intel. The DoD project involves Nvidia already making chips on Intel's 18A process node, so it wouldn't be a total surprise.
NVIDIA has been secretly working with the government on Intel 18A without telling anyone WTF.

>>106623203
>why the fuck would nvidia want to keep a dying company afloat.
The order came from the top

>>106623113 (OP)
holy based.
noshittia and intelaviv linked up!
need it or keep it?

like every modern major partnership, it's all vaporware.
don't worry about it.

What does this mean for AMD?

>>106623442
>>106623710
x64 is just an extension retard, it was built upon the existing x86 architecture, do you even understand what cross licensing is?

>>106624049
It means AMD will continue to be shit on in the GPU market (no surprise) and Intel is almost certainly going to abandon Xe in favor of bundling RTX now on higher end chips and likely ditch their dGPUs entirely.

Ross and Rachel ahh companies

WTF this is actually crazy

You clowns really should have just let Nvidia buy ARM. Then they'd be stuck on ARM junk. You've now given them x86 keys it is fucking over.

>>106623113 (OP)
>china bans nvidia ai chips
>suddenly partners with intel
Hmm

>>106624049
nothing.
this is a nothingburger.

what are the chances this kills nvidia vs the chances nvidia saves intel?

>Shares in Arm, Nvidia’s current partner on CPUs, fell 6 per cent in early trading on Thursday. Last month, Japan-based SoftBank, Arm’s majority owner, said it would purchase $2bn worth of Intel shares.

Arm must be begging Apple to buy them at this point

>For data centers, Intel will build NVIDIA-custom x86 CPUs that NVIDIA will integrate into its AI infrastructure platforms and offer to the market.
okay, seems more like nvidia's arm chips are a failure than anything else.
>For personal computing, Intel will build and offer to the market x86 system-on-chips (SOCs) that integrate NVIDIA RTX GPU chiplets. These new x86 RTX SOCs will power a wide range of PCs that demand integration of world-class CPUs and GPUs.
ouch. rip arc.

>>106624139
0%, unless the AI bubble catastrophically bursts. Intel needs the bailout far more than Nvidia needs Intel or x86-64 products. It's a boon to Nvidia's datacenter product stacks / margins to have x86 available but they're still going to sell basically everything they can make regardless. Nvidia can't buy Intel outright because a change in ownership would invalidate the license agreement with AMD that allows Intel to legally produce x86-64 processors. So Nvidia will likely ride Intel's license for whatever it's worth and maybe Intel can save its CPU business along the way, but they're definitely going to be shafted on their own margins because Nvidia's the one driving that partnership.

>>106624177
>nvidia's arm chips are a failure than anything
yeah it's been known for ages they suck. They've wanted this x86 licence for decades it's insane this deal is happening

>>106623113 (OP)
Anyone who is not a retard knows that this AI is strictly for user behaviour surveillance.

AMDBROS DID WE GET TOO COCKY???

So US goverment bought 9.9% stake in intel; $8.9B USD in late August at $20.47/share.

That same 9.9% stake is now worth ~$13.5B today, ~$31.28/share right now.

Not a bad $4.5B gain in less than 30 days.

>>106624058
AMD's cross license to Intel of the 64 bit extensions does not give them the right to license it out to another party.
It's may be a moot point because the parents are expired and AMD would have a very weak IP case against Intel licensing it to Nvidia
But it certainly doesn't matter because Nvidia doesn't want x86

>>106624406
Nvidia doesn't NEED x86 when they can just get intel to design and fab whatever they need.

>>106623404
NVLink is just an interconnect. Nvidia superchips also use it to connect the custom nvidia CPU to the GPUs and also to connect superchips to superchips.

>>106624406
>But it certainly doesn't matter because Nvidia doesn't want x86
nta but something of interesting that no one seems to bring up. Nvidia actually has a x86 license, and has had it for quite some time after acquiring ULI (aka ALi, Acer labs inc). Back when Nvidia was first producing their Tegra CPU which is a ISA agnostic chip, they actually showed it off running x86 code, but eventually just went the ARM path with it. Nvidia /could/ enter the x86 space if it ever wanted to.

>>106624139
nvidia is going to buy Intel lmao.

God I can't wait for this AI bubble to pop already. Fuck the investooooorrrrrr bros for making the US economy even more retarded.

>>106623113 (OP)
>AYYMDEAD IS FINISHED & BANKRUPT
Minus the whole still being the only company that makes CPU/APU's for the consumer market that anyone actually gives a fuck about. Even Normie Niggers know that your run of the mill Ryzen 5 iGPU will out perform any Core 5 or meme ARM chip at most tasks only being beaten slightly by ARM in terms of battery life. Qualcomm gave it a shot to compete with AMD and Apple, so far their efforts have been futile, mostly because of relying on gAIy shit hype powered by Google Samsung and ByteDance slop that Zoomers don't give a shit about because they've already got iPhones and professionals could care even less for those gimmicks. Hopefully Qualcomm gets their shit together and makes something that doesn't suck, be neat to see some genuine competition.

>>106623203
>why the fuck would nvidia want to keep a dying company afloat.
Work around to avoid monopoly laws that would force Nvidia to be broken up into several companies. Same reason Google pays FireFox and Apple/Safari to '''compete''' against them.

>>106624177
>okay, seems more like nvidia's arm chips are a failure than anything else.
This is just cope. Arm's datacenter presence has been growing rapidly and Nvidia's superPOD shit is definitely a big part of the reason. Market research firms suggested Arm currently holds 25% of the datacenter market, up 10 points from last year, almost certainly because of AI growth.

>>106623113 (OP)
This doesn't make Intel better, it makes Nvidia worse.

>>106624457
You're wrong and misunderstanding the license ALi had.

ALi had a CHIPSET license, they could produce x86 chipsets for intel CPUs. They were not licensed to produce/develop x86 CPUs.

Beyond that, those licenses are almost always non-transferable so when Nvidia bought them, Nvidia didn't get the x86 chipset license.

And finally, the x86 code Nvidia showed off was just running on a translation layer, like how apple ARM silicon runs x86 software.

>>106624511
I was mistaken, but ALi did produce a x86 CPU... kinda
https://www.cpushack.com/ULi.html
The embedded CPU they produced had a integrated Intel 386sx core, so not one of their own making.
That said, Nvidia did still show off Tegra once running x86 code so it was at least a consideration of Nvidia at one point, but granted that was about 20 years ago. That ship long sailed and I doubt they have any real consideration to step away from ARM.

>>106624993
Yes, but again, it was an x86 proof of concept on an ARM processor, not x86 code running natively on x86.

Nvidia doesn't have an x86 license, and even if they wanted to, would not legally be allowed to produce an x86 CPU.

We're officially moving into the APU era, local AI models running on system ram + GPU chiplets. If you aren't working on an application using local AI you're doing it wrong

>>106625041
Tegra was sold as an ISA agnostic chip originally, it could run x86, MIPS, and ARM, all defined in microcode. Kinda like the Transmeta CPUs which themselves were VLIW based.
But I do doubt they operate the same way anymore, my guess is that modern Tegras are just outright ARM chips proper for the sake of efficiency.

>>106623113 (OP)
Just bought a Lunar Lake laptop, feels good.

>>106625081
My guy, is it THAT hard to admit you're wrong?

> Tegra was sold as an ISA agnostic chip originally, it could run x86, MIPS, and ARM, all defined in microcode
Nope, it has ALWAYS been ARM. Specifically the first Tegra chip was ARM11 based.

>Kinda like the Transmeta CPUs which themselves were VLIW based.
Nope, Transmeta was doing on-the-fly code translation called dynamic binary translation. It was never natively running x86. It WAS emulated at a hardware level instead of software, but it still wasn't natively running x86.

So to reitirate, Nvidia's Tegra chips were standard Arm-based SoCs, and any x86 functionality was achieved through a software layer, which does not require an x86 license.

The claim that Tegra could "run x86, MIPS, and ARM, all defined in microcode" is a fantasy. It fundamentally misunderstands how computer architectures and microcode work.

The demos of Tegra running x86 code were a technical proof of concept of their software-based emulation capabilities, not an indication of a hardware-level feature.

>>106623203
>dying
ARM will never be the norm, fag

>>106623113 (OP)
ah yes the free market

>>106623113 (OP)
i can't wait for shit that work only on nvintel, what a joke

>>106623415
>investing
the only who shouldn't invest is you, intel has be 20 bucks range for more than half a decade.

>intel doesn't innovate CPU
>no one buys intel cause they are worse than AMD and they kill themselves.
>only intel uses their foundry.
>fire engineer ceo that new intel was behind for axe man of short term gains.

at this point the best thing that happens is nvidia shadow controls intel and make them make a competitive cpu. else all this is just shit intel cpu architecture with a better connectivity to NVIDIA.

>>106625679
>intel has be 20 bucks range for more than half a decade

They were in the $40-60 range less than 5 years ago moron.

>>106625704
im sorry the covid feels longer anon doesn't change it's a shit company.

>>106625766
No need to be mad you didn't buy the dip.

I'm up 56% by the way.

>>106625775
that bad lol, come back when your at 200.

>>106625281
Bro, I DID admit I was wrong. Nvidia still made the _thing_, but they were not licensed to actually sell it, it was merely a tech demo. you were right, fucking christ anon, get over yourself.

>>106626034
I mean, you didn't though, you doubled down trying to say well ACHTUALLY nvidia DID have an agnostic ISA, which as I proved again, was a bullshit lie you made up.

unfortunately this is just another trump deal for israel. nvidia should just buy intel and make apus that can defeat macbooks already.

>>106625853
I've got a defense stock up almost 300% for the year, my single highest performing stock in my portfolio.

>>106626162
Anon, I continued talking about it since I found it interesting, NOT because I was trying to one up you. I even provided a link proving _you_ were right about the licensing thing.
Not everyone is trying to converse with you in bad faith anon, again, get over yourself.

>>106626193
Wow, so you knew you were wrong originally, conceded and then STILL kept parroting even MORE wrong information.

Honestly you look even more retarded now.

just last week openAI announced they are making their own AI chips with Broadcom

>>106626193
he didn't get over himself

>>106626223
No, I didn't, I thought they had a lisense but it turned out the CPU ALi produced was just a sub-licensing from intel to integrate their core, not one to actually make their own.
How many times do you need me to say you were right?????
Also, here is a 2014 article referencing Nvidia's prior x86 effort with Tegra since you seem to believe that it never happened.
https://www.extremetech.com/computing/174023-tegra-k1-64-bit-denver-core-analysis-are-nvidias-x86-efforts-hidden-within
Again, what they did was a tech demo, nothing more.
BTW, work on your reading comprehension you fucking retard.

>>106626269
Again AFTER that you said
>Tegra was sold as an ISA agnostic chip originally
which is blatantly, categorically incorrect, and you should've KNOWN that, from the previous posts, yet you posted it anyway like a fucking moron.

>>106626307
ESL?
"Selling" also means "marketing"/"presenting", which is what I was originally meaning, since this was BEFORE it was an actual finished product and was just being shown off to tech websites for news at the time. Not LITERALLY selling the product....

>>106626358
Even then, it was NEVER marketed as "ISA agnostic"

It has ALWAYS, even in development, been acknowledged to be an ARM ISA chip, it was never marketed as "ISA agnostic".

You're just taking their x86 demo and extrapolating that out beyond anything Nvidia themselves claimed, none of the Nvidia marketing at the time made the claims you're currently making for Tegra.

>>106626415
Reference to the chip's ISA agnostic binary translation chip from Nvidia
https://pcper.com/2014/08/nvidia-reveals-64-bit-denver-cpu-core-details-headed-to-new-tegra-k1-powered-devices-later-this-year/
>NVIDIA is using a custom ISA and on-the-fly binary translation to convert ARMv8 instructions to microcode before execution.
Original x86 ambitions, there are many articles from 2010-2014 regarding this, but here is one
>https://web.archive.org/web/20250524124637/https://www.semiaccurate.com/2010/08/17/details-emerge-about-nvidias-x86-cpu/

>invest in company
>kill GPU division
>take over after
>obtain licenses
woooooooow

>>106623113 (OP)
>SOC
So soldered RAM and possibly SSDs? Who the fuck wants this other than leather jacket man???

>>106623113 (OP)
I own a lot of nvda shares. This is good. I dont care about intel, tech or anything other than my green number going up.

>>106624488
>Market research firms suggested Arm currently holds 25%
I'm not sure about that. If you include hyperscalers deployed server racks only, then maybe; but most computing still happens in x86. In the company I work, we use mostly x86 AWS instaces. We did move some DB instaces to Graviton CPUs, but they are garbage. They tend to stall a lot and we had to increase instance size just to match the old instances (still cheaper than x86). The only reason why you would pick ARM is price, anything else is not worth it. With AMD/Intel pricing war that is happening right now, ARM has lost a lot of traction.

>>106625679
>>no one buys intel cause they are worse than AMD and they kill themselves.
what part of 60% overall (desktop 48.3%, laptop 73.5% and data center 67%) market share do you not understand?

>>106623113 (OP)
x86 license open source when
FUCK THIS GAY EARTH

>>106626844
60% market share and they are still im the shitter lol
They are only not dead yet because of OEMs like Dell

>>106624049
This won't improve Kiketel on the CPU side. It might unironically mean that future consoles might make a deal with intel and nvidia though.

>>106626808
>I own a lot of nvda shares
how many?

>>106624391
>Nothing has been produced
>Not bad gain

>>106623930
Why wouldn't it he legal? It's their tech.

AMD bought ATI 19 years ago and they were never able to fully integrate both companies together. Lisa Su should have been fired 2 year ago when it was obvious she missed the AI boat completely. She's an awful CEO, Rory Read was the one who hired Jim Keller back at AMD and created the road map up to Zen 2. Zen 5 is barely different from Zen 2, it's the engineers at TSMC who are making the real work, not AMD's.

>>106623930
>NVIDIA has been secretly working with the government on Intel 18A without telling anyone WTF.
It wasn't a secret.

>>106623203
same reason why google keeps firefox alive

>>106623211
Intel can't license out x86 due to the cross-licensing agreement with AMD. Also, if Nvidia were to buy Intel, the cross-license would be voided.

>>106623113 (OP)
Amd will give Nvidia an x86 license in exchange for a CUDA license. The golden age will come for pc consumers.

>>106623415
>70% global server market
I don't think this has been true for a while and with how things are going they are going to end up at like 10% in a decade

>>106623211
They literally can't
They made the licence non transferrable because they thought AMD would go bankrupt and bought by someone else

>>106626844
>60% overall (desktop 48.3%, laptop 73.5% and data center 67%)
What were those numbers last year? And five years before?

>JENSEN YOU CAN'T LET INTELAVIV DIE
It is so tiring.
It is all so fucking tiring.

>>106623203
Political cover, same as Softbank $2Bn
They likely have a bond or covenant for which they need the cash, they just got a 16Bn infusion, follow the money

>>106623113 (OP)
According to MLID INTC have a marketshare apocalypse coming in the form of laptops, the only place they have appreciable market share.
They have been in a bad way for a while, and 30 miles up the road is TSMC with a new US fab. IMO is it unlikely INTC will ever build a new fab, because of the cost. They are getting rid of people and products, this is all but a bailout.

https://discord.com/channels/951039693565333535/952629802542694461/1418684850348621904

>>106623203
Intel arc GPUs were positioned to disrupt the GPU market, and especially the availability of vram chips. Jensen is just being a smart mf like he's been for his entire career. Genuinely impressive guy.

>>106623531
You can't do that, you would also have to pay off AMD.

>>106623930
IBM Power8 and Power9 also had integrated NVLink.

>>106638183
Isnt his main competitor the Chinese startup GPU chips currently being released? Currently they are crap, but in a few years they will be more than good enough to run inference and then training.

Well this will probably kill intel gpus, whatever component that division may need will be allocated to Learher Jacket man

>>106639986
>whatever component that division may need will be allocated to Learher Jacket man
God I hate leather jacket man (pic not related).
RIP Arc