Nvidia CEO Says Intel's Test Chip Results For Next-Gen Process Are Good

[DuckDodgers]

 Jensen Huang already has been in talks with Intel last year, exploring prospects in diversifying semiconductor manufacturing options outside of TSMC and Samsung. 

 

Quotes

Quote

It's becoming more likely that Intel could build Nvidia's future GPUs. We attended a question and answer session with Nvidia CEO Jensen Huang here in Taipei, Taiwan, and Huang said that the company is working to diversify its chip manufacturing and had recently received good test results for an Intel test chip based on the company's next-gen process node. His statement came after several of the questions centered around Nvidia's efforts to assure supply in the face of explosive demand for AI chips and the company's complete reliance on Taiwan-based TSMC for its most advanced GPUs.

"You know that we also manufacture with Samsung, and we're open to manufacturing with Intel. Pat [Gelsinger] has said in the past that we're evaluating the process, and we recently received the test chip results of their next-generation process, and the results look good," Huang said.

 

Nvidia might soon tap every top-node foundry out there, but the real question for us here is if this will move the price needle down on video cards in foreseeable future?

 

Source: https://www.tomshardware.com/news/nvidia-ceo-intel-test-chip-results-for-next-gen-process-look-good

 

[VaderCraft_]

It won't make it any cheaper, I have a feeling they don't care about the gamers anymore because they barely make any revenue in comparison to the workstation GPU market they have. FSR 3 is going to possibly dominate

[porina]

My guess, nvidia will use more advanced processes for more profitable products, which isn't gaming. They're unlikely to give up gaming since it is still a big earner for them. Just because it isn't the biggest part of their share isn't a reason to drop it. About 1/3 of their current total revenue in gaming is not small change.

 

It's also an interesting time for Intel as they will have a lot of change expected to come online over the next couple years. Then we'll see if their recovery plan has worked.

[johnno23]

Nvidea is in the lets print money phase and they are doing well regardless of the piss poor cards they have been releasing these last months.

They have no intention of helping gamers and in reality no longer need gamers to swell the band account. They are now so rich and have eyes on enterprise and gemers are now left with a situation where nvidea releases a few cards and says heres the price take it or leave it........they simply do no t need to help the gaming community unless they suddenly have a change of heart which is kinda hoping the titanic has enough lifeboats.

Intel if they continue will fill the void and AMD will probably take over as mainstream GPU for gamers is my thoughts on the situation.........but I can hope I am wrong as Nvidea have the tech and the knowhow to produce some excellent GPUs they just dont seem to care for anyone outside of flagship product lines.

[Spotty]

Would be an interesting choice since neither Samsung or TSMC are directly competing with Nvidia in that market but Intel are. Intel fabricating Nvidia GPUs could lead to the situation where Intel's own Arc GPUs aren't able to competitively compete against the Nvidia design GPUs they're fabricating.

 

 

 

At what stage is Nvidia just going to create their own fab facility? Yeah, creating a fab is definitely not cheap, but if there's anybody that could feasibly do it then it would probably be Nvidia.

[porina]

3 minutes ago, Spotty said:

Intel fabricating Nvidia GPUs could lead to the situation where Intel's own Arc GPUs aren't able to competitively compete against the Nvidia design GPUs they're fabricating.

If Intel is able to make money off nvidia, I don't think they'll be unhappy about it. It's still very early days for Intel dGPUs. Catching up to a long established and huge market share competitor in nvidia isn't going to be an overnight process. More likely a generational process.

 

3 minutes ago, Spotty said:

At what stage is Nvidia just going to create their own fab facility? Yeah, creating a fab is definitely not cheap, but if there's anybody that could feasibly do it then it would probably be Nvidia.

If Apple haven't done it, don't think nvidia will. Businesses like to focus on their strengths. Fabs are a lot of work. 

 

[leadeater]

21 minutes ago, Spotty said:

At what stage is Nvidia just going to create their own fab facility? Yeah, creating a fab is definitely not cheap, but if there's anybody that could feasibly do it then it would probably be Nvidia.

They probably don't want to fall in to an AMD/GloFo situation so might never do this. Seeing the Intel fab pains and GloFo white flag the leading edge nodes Nvidia could well have zero appetite to enter in to fabrication. 

[porina]

2 minutes ago, leadeater said:

They probably don't want to fall in to an AMD/GloFo situation so might never do this. Seeing the Intel fab pains and GloFo white flag the leading edge nodes Nvidia could well have zero appetite to enter in to fabrication. 

I just had a crazy idea. Current market cap of nvidia is about 8x that of Intel. It would probably be easier for nvidia to buy Intel than to build their own fab if they really wanted some.

 

Note market cap is only one measure of a company's size, and nvidia's is inflated at the moment from the AI hype train. Also given what happened when they looked at Arm, I don't expect any serious attempt for them to acquire Intel to happen or be allowed to happen.

[Spotty]

2 minutes ago, porina said:

If Apple haven't done it, don't think nvidia will. Businesses like to focus on their strengths. Fabs are a lot of work. 

That's probably a good point. With Apple though they're doing a lot more than just designing SoCs, and I would argue that's actually a pretty small part of what Apple does and something they've only been doing very recently. Whereas with Nvidia designing processors is (almost) entirely what they do. If Nvidia's ever tries to acquire ARM again they could also help offset the cost by fabbing their own ARM CPUs as well as their GPUs.

 

With USA trying to get more chip fabs built within their borders it wouldn't surprise me if Nvidia took the chance for some fat grant money to build their own fab. Though obviously we're talking several years and generations down the line for any new fabrication facility to be up and running, clearly doesn't solve Nvidia's problem of finding a fab partner for their next GPU. Maybe it's something we'll see in 10 or 20 years time.

 

6 minutes ago, leadeater said:

They probably don't want to fall in to an AMD/GloFo situation so might never do this. Seeing the Intel fab pains and GloFo white flag the leading edge nodes Nvidia could well have zero appetite to enter in to fabrication. 

I guess the biggest benefit for Nvidia not having their own fab is simply being able to switch to whoever is going to be the most competitive at the time, both in regards to cost and latest process nodes. Not getting stuck for possibly multiple product generations on what their own fab can deliver. TSMC 12nm -> Samsung 8nm -> TSMC 5nm -> Intel???

 

[Brooksie359]

23 minutes ago, Spotty said:

That's probably a good point. With Apple though they're doing a lot more than just designing SoCs, and I would argue that's actually a pretty small part of what Apple does and something they've only been doing very recently. Whereas with Nvidia designing processors is (almost) entirely what they do. If Nvidia's ever tries to acquire ARM again they could also help offset the cost by fabbing their own ARM CPUs as well as their GPUs.

 

With USA trying to get more chip fabs built within their borders it wouldn't surprise me if Nvidia took the chance for some fat grant money to build their own fab. Though obviously we're talking several years and generations down the line for any new fabrication facility to be up and running, clearly doesn't solve Nvidia's problem of finding a fab partner for their next GPU. Maybe it's something we'll see in 10 or 20 years time.

 

I guess the biggest benefit for Nvidia not having their own fab is simply being able to switch to whoever is going to be the most competitive at the time, both in regards to cost and latest process nodes. Not getting stuck for possibly multiple product generations on what their own fab can deliver. TSMC 12nm -> Samsung 8nm -> TSMC 5nm -> Intel???

 

I mean I think both AMD and Intel had huge issues with being limited by their own fabs. I know technically AMD and global foundries have a bit more complicated relationship but regardless AMD had issues with global foundries holding them back and so did Intel with its fabs being stuck on 10nm for so long that it became a meme. Granted I do think Intel having its own fabs did help alot of with keeping really good supply of cpus during covid when supply issues were plaguing most places. I remember AMD cpus were out of stock way more often than Intel at my local microcenter. 

[Quackers101]

sure intel with their many new fabs will increase production a lot, when everything gets up and running (some might be on now).

If it will reduce cost one would hope, but some doesn't get much cheaper to produce, defects, everything else used or needed, hope it doesnt end with just a pile of chips, although some industries would like them more than others.

 

Also what other stuff intel is doing or testing for chips, like chip on chip, any sort of 3D, some of their interconnects (photonics, wifi, yes).
To where are we with memory?

[WolframaticAlpha]

2 hours ago, porina said:

interesting time for Intel as they will have a lot of change expected to come online over the next couple years.

What would be interesting is that Intel has projected Q1 2024 for 20a, so that means around the same time they are going to ramp up Intel 3, which has the HDL needed for an nvidia deal.

 

1 hour ago, porina said:

just had a crazy idea. Current market cap of nvidia is about 8x that of Intel. It would probably be easier for nvidia to buy Intel than to build their own fab if they really wanted some.

 

Note market cap is only one measure of a company's size, and nvidia's is inflated at the moment from the AI hype train. Also given what happened when they looked at Arm, I don't expect any serious attempt for them to acquire Intel to happen or be allowed to happen.

Nvidia's current market cap is greater than AMD Intel and TSMC put together lmfao.

1 hour ago, Spotty said:

tlTSMC 12nm -> Samsung 8nm -> TSMC 5nm -> Intel???

 

Nvidia is a serial philanderer at this point.

[LAwLz]

I really hope things go well for Intel with their fab business. We really need more competition in the space.

I have high hopes for Samsung's GAE process as well.

 

 

5 hours ago, VaderCraft_ said:

It won't make it any cheaper, I have a feeling they don't care about the gamers anymore because they barely make any revenue in comparison to the workstation GPU market they have. FSR 3 is going to possibly dominate

By workstation GPU market I assume you mean datacenter. Their workstation revenue is probably pretty low compared to data centers and gaming. Nvidia still had over 9 billion dollars in revenue directly related to its gaming products. It's not as much as the 15 billion dollars their data center business generated, but it's still over 33% of their revenue.

 

For comparison, gaming makes up 29% of AMD's revenue.

 

 

If the argument is that Nvidia doesn't care about gaming because they make more money from other things they sell, then that argument is even more true for AMD.

[leadeater]

52 minutes ago, LAwLz said:

I really hope things go well for Intel with their fab business. We really need more competition in the space.

Intel returning to their premier fab status would be very nice

[Kisai]

9 hours ago, Spotty said:

At what stage is Nvidia just going to create their own fab facility? Yeah, creating a fab is definitely not cheap, but if there's anybody that could feasibly do it then it would probably be Nvidia.

They won't until we hit a high enough brick wall to justify no further die shrinks. Same for Apple. Because at that point, no further innovation happens at the die level and it just becomes chip packaging differential.

 

Like imagine we get to the 0.1u node there will be nothing left to innovate on in terms of die shrinks and it will just be figuring out how to make 3d stacks that can be cooled effectively. Probably sandwiching silicon layers between heat conductive materials to make a CPU "cube", with RAM, iGPU, and potential VPU dies like LEGO(tm) bricks.

 

What I also think is going to happen is that this will happen sooner rather than later. Like by 2030. Then if you want a super powerful GPU, you'll have to lease it from a cloud vendor because operating one in your home will not be viable.

[igormp]

9 hours ago, Spotty said:

something they've only been doing very recently.

Apple has been designing their own µArches since the iPhone 5, that's over 10 years now. Not as much as Intel/Nvidia/AMD, but I wouldn't call that recent by any means.

 

9 hours ago, Spotty said:

TSMC 12nm -> Samsung 8nm -> TSMC 5nm -> Intel???

During Ampere they used both Samsung 8nm for their non-100 chips, and TSMC 7nm for their A100. So using multiple fabs is also an option for them.

[leadeater]

50 minutes ago, igormp said:

During Ampere they used both Samsung 8nm for their non-100 chips, and TSMC 7nm for their A100. So using multiple fabs is also an option for them.

Before then Nvidia also dual sourced from different fabs, Samsung was used for the low end chips and TSMC for everything else. Generation after that Nvidia moved mainly to Samsung other than those GA100 dies.

[AnonymousGuy]

11 hours ago, Spotty said:

 

If I had a cube I'd hang this in there.

[Salv8 (sam)]

11 hours ago, DuckDodgers said:

but the real question for us here is if this will move the price needle down on video cards in foreseeable future?

no, Jensen likes his cash to be as green as the company's logo.

[RedRound2]

16 hours ago, DuckDodgers said:

Nvidia might soon tap every top-node foundry out there,

There are literally like only 3 of them

15 hours ago, Spotty said:

At what stage is Nvidia just going to create their own fab facility? Yeah, creating a fab is definitely not cheap, but if there's anybody that could feasibly do it then it would probably be Nvidia.

Fab business would not make sense for any company that only works on cutting-edge nodes. TSMC, Samsung, and Global Foundries still continue to pay off their R&D investments for a particular node well after the initial 1-2 years it's released. While companies like Nvidia (and Apple too) only want the latest node and they would use it for about a year, max maybe two years, before moving on to the next node. Were they to make a fab, then they will either need to continue using the now older node tech for some other products, or they will need to sell it to someone else, which is quite honestly a mess, especially when this isn't their area of focus.

 

Rather, not sure about Nvidia, but Apple does invest into TSMC a few billion dollars every now and then to help them research newer processing nodes

[Quackers101]

5 hours ago, RedRound2 said:

Rather, not sure about Nvidia, but Apple does invest into TSMC a few billion dollars every now and then to help them research newer processing nodes

yes too much of a risk to do your own and a lot of money to invest with needed specialists to make any progress and to combat the others on the market.

If there is less needed to create future or different "chips" then maybe, right now its to buy their place in who gets the newest tech upgrade and able to get those who has the "coolest tech" or fill their needs?

Quote

There are literally like only 3 of them

But for memory to SSDs, isn't that a few other ones? although not sure how many there are of those ones. I guess some being Micron (Crucial), Samsung, and Hynix. So yeah not many there either.

Edited May 31, 2023 by Quackers101

[05032-Mendicant-Bias]

21 hours ago, porina said:

My guess, nvidia will use more advanced processes for more profitable products, which isn't gaming.

That's fine. For a while when Nvidia was really far ahead of AMD, Nvidia used lower nodes from Samsung and kept leading edge nodes for their server lineups.

 

Right now we have competition in both CPU and GPU space, all manufacturers are forced to bring their A games to the table.

[leadeater]

37 minutes ago, Quackers101 said:

But for memory to SSDs, isn't that a few other ones? although not sure how many there are of those ones. I guess some being Micron (Crucial), Samsung, and Hynix. So yeah not many there either.

There are a lot of fabs, a lot. But there are only 3/4 on leading edge microprocessor technology. TSMC, Samsung, Intel and (sort of) GlobalFoundries. 

[Fasterthannothing]

22 hours ago, porina said:

If Intel is able to make money off nvidia, I don't think they'll be unhappy about it. It's still very early days for Intel dGPUs. Catching up to a long established and huge market share competitor in nvidia isn't going to be an overnight process. More likely a generational process.

 

If Apple haven't done it, don't think nvidia will. Businesses like to focus on their strengths. Fabs are a lot of work. 

 

Or maybe they aren't actually going to be competition at all. I actually foresee Intel doing gaming GPUs and Nvidia dropping them completely in the future. They aren't really in it for the gamers anymore. 

[VaderCraft_]

23 hours ago, Spotty said:

 

At what stage is Nvidia just going to create their own fab facility? Yeah, creating a fab is definitely not cheap, but if there's anybody that could feasibly do it then it would probably be Nvidia.

I mean they just joined the trillion dollar club for companies