News NVIDIA and Intel to Develop AI Infrastructure and Personal Computing Products

[branch_suggestion]

I don't think that this will run afoul of any anti-trust overview.

Eh, I think there is a small chance that Intel/NV combined products may be banned from sale in China at the very least.
EU has some good reason to follow.

 

[LightningZ71]

The current ban is focused on dedicated AI chips. This is going to be trying to skirt the edge of that one.

 

[Joe NYC]

This is why i think this is for Halo or High end products not cheap ass products

Most people think that Intel will not abandon their low end iGPU efforts, which may include low end monolithic chips

 

[Joe NYC]

The current ban is focused on dedicated AI chips. This is going to be trying to skirt the edge of that one.

It has not been announced as an official ban. It may just be, for now, part of the posturing / negotiations.

 

[Antey]

This could also be related to the fact that ARM is entering the PC/laptop market (and for real this time). Intel, NVIDIA, and AMD are at risk. New players like Qualcomm, Samsung, and MediaTek will all enter the PC market and offer their C1 + Mali/Adreno iGPU-based SoCs. NVIDIA is about to release its N1 and N1X SoCs… but I feel they want to eat the whole pie. ARM brings a SoC paradigm: NVIDIA can’t just put an RTX GPU in laptops with Qualcomm or Exynos SoCs, and Intel and AMD might focus on APUs to counter that reality. This could be NVIDIA’s way to remain part of the x86 ecosystem in the near future.

 

[Josh128]

Big news. Wow!

It will be interesting to see how this plays out in the long run, but the details of the collaboration say a few things up front:
1) This is purely a collaboration on products, not to fab Nvidia GPUs on IFS (although I’m guessing the final packaging may be at Intel). You take an Intel chiplet with integrated NVLink and join it with an Nvidia GPU with NVLink, not unlike GB10.

2) This gives Nvidia another avenue to push AI PC to the market outside of the late to market GB10. I’m guessing Jensen is not happy at how their WOA effort has stagnated.

3) These products will take years before they hit the market, and $5B is a paltry amount over the span of a few years. It’s not hard to imagine that the cash infusion is mainly to curry favor with the Trump administration given the administration’s now desire to prop up Intel (the US Government owning Intel shares sure does make the interests intertwined).

4) Will Intel give up on their own GPU ambitions? Does Nvidia think their GPU tech is so desirable that the consumer would rather choose an Intel CPU with an attached Nvidia GPU over an Intel CPU+GPU SoC? I don’t see massive volumes of this collaborative product, which leads me to feel like this deal is mostly performative.

100%.

Huang is playing the political game here, 5B is basically nothing for Nvidia, but it does give Intel another quarter or so to turn things around, as they are as of the most recent earnings, losing almost $4B per quarter.

Nvidia gets presence and visibility in x86 APUs and potentially some ROI if Intel doesnt completely shat the process node bed. It would be very funny if NV decide to fab their chiplet on TSMC and pair it with Intel fabbed CPU chiplet. I think Huang is jealous of the optics of the Strix-Halo chip and wants to show it up, even if its a similar low volume product.

Intel certainly wont give up on their own IGPU ambitions, I think this will be for "premium" Strix Halo style APUs only.

 

[adroc_thurston]

Wait what's gonna happen with Intel gfx stuff? GPGPU too.

 

[511]

Wait what's gonna happen with Intel gfx stuff? GPGPU too.

with this dGPU may be canned but none knows about iGPU and Analysts are too dumb to ask this so only time will tell

 

[branch_suggestion]

Wait what's gonna happen with Intel gfx stuff?

iGPU seems to be fine.

GPGPU too.

That is a lot murkier, I guess there is nothing stopping Intel from keeping it alive since NV wouldn't see it as a threat.

 

[adroc_thurston]

That is a lot murkier, I guess there is nothing stopping Intel from keeping it alive since NV wouldn't see it as a threat.

that would be top kek.

 

[Win2012R2]

Under normal circumstances regulators won't allow it just on the basis that Nvidia is already totally dominant in GPUs, plus clear abusive behavior - "buy our Mellanox stuff or get to the back of the queue for GPUs"

 

[StinkyPinky]

I wonder if Nvidia want access to Intel's fabs. They have a bunch of money if they can get them up to scratch they won't need to rely so much on TSMC.

 

[branch_suggestion]

Intel is playing all sides to try to win, NV is throwing out Tsar Bomba levels of FUD to feed the hypebeast.
Based on the track record of similar moves in the past, will probably end up as a nothingburger.

 

[adroc_thurston]

Based on the track record of similar moves in the past, will probably end up as a nothingburger.

no this is actually relevant and is indicative of both the state of Intel gfx roadmap (bad), and Tegra (exploded again. but when hasn't it?)

 

[gdansk]

No impact on N1X and N2X apparently.

 

[adroc_thurston]

No impact on N1X and N2X apparently.

those were completely stillborn anyway.

 

[branch_suggestion]

no this is actually relevant and is indicative of both the state of Intel gfx roadmap (bad), and Tegra (exploded again. but when hasn't it?)

Oh yes I think this is indeed bullish for AMD et al.
But in terms of how this collab will impact the marketplace in a few years time?
Impossible to know, but I think it won't change that much, EPYC gets virtually zero NV adjacent sales as is and future DGX/HGX boxes will probably have to be Intel, and their rackscale stuff can also be Intel now.
It is largely edge case things as I'm sure NV will try to maintain their ARM stuffs as being the de facto option, but very bullish for x86 overall, especially in PC.

 

[Schmide]

I think this says a bit about the arm HEDT AI development system. You do not see any reviews of the the DGX Spark and what you do read says it's not ready for prime time. On the other side strix halo seems to be a darling of the at home AI development community. Why? Software! You can spin up your choice of distros, make it your daily driver, with pytorch and the like on the side. This is a void where no arm system is going to compete, for now, and for higher end systems you need $10k video cards and 5k worth of PC behind them.

It generally seems except for long run training, the cpu side needs to be robust for setup and a mid range x86 cpu is the sweet spot.

 

[Saylick]

No impact on N1X and N2X apparently.

Not surprised. Nvidia are just trying every avenue they can to proliferate Nvidia GPUs to the market. They have so much cash flowing in, they can afford to invest pretty much everywhere. And for the cash they cannot invest, stock buybacks baby.

 

[adroc_thurston]

It is largely edge case things as I'm sure NV will try to maintain their ARM stuffs as being the de facto option, but very bullish for x86 overall, especially in PC.

Yeah NV went from "ARM is all you need, we're gonna literally buy the goddamn company" to this.
whatever that is (total complete surrender to x86?).

 

[adroc_thurston]

Not surprised. Nvidia are just trying every avenue they can to proliferate Nvidia GPUs to the market. They have so much cash flowing in, they can afford to invest pretty much everywhere. And for the cash they cannot invest, stock buybacks baby.

roadmap changes don't happen overnight.

 

[Saylick]

roadmap changes don't happen overnight.

But the Nvidia roadmap to invest their mountain high pile of cash everywhere does.

 

[gdansk]

Now AMD must fight the three-headed hydra of US federal government and Intel and Nvidia.

Maybe next quarter Intel will find a way to expand the competition into the four horsemen.

 

[Tup3x]

So is this "investment" or "takeover"? I'm not really sure what to think about this...

 

[Joe NYC]

Agreement on products was reached just 5 days age. So nothing is in the pipeline. No work on anything mentioned was even started:

https://twitter.com/x/status/1968670577461276819