NVIDIA CC at Citigroup conference

[dZeus]

But that's a discrete chip.
How is that related to a x86 license?
Neither AMD nor NV is a near-monopoly right now, so I'm not sure how this is comparable to a company preventing others from competing in an entire multi-billion-dollars industry.

sorry for digging up this old thread, but I read an interesting piece on the wikipedia page of Cyrix:

Wikipedia entry for Cyrix

and I quote:

By and large, Intel lost the Cyrix case. But the final settlement was out of court: Intel agreed that Cyrix had the right to produce their own x86 designs in any foundry that happened to already hold an Intel license. Both firms gained out of this: Cyrix could carry on having their CPUs made by Texas Instruments, SGS Thomson, or IBM, all holders of Intel cross-licenses; Intel avoided a potentially embarrassing loss.

Does this in any way imply that nV could produce x86 chips at any fab company that holds cross-license deals with Intel covering x86?

 

[_xxx_]

No. Cyrix had a x86 license, NV doesn't have it.

 

[dZeus]

No. Cyrix had a x86 license, NV doesn't have it.

well if you read what I posted, or click on the link, you'll see that Cyrix was involved in a lawsuit with Intel, _because_ they didn't have a license!

Unless I misinterpret this completely...

 

[_xxx_]

No, I mean the basic license to produce CPU's with the x86 IP, which nV doesn't have.

 

[INKster]

No, I mean the basic license to produce CPU's with the x86 IP, which nV doesn't have.

Really ?
Then what is the company name doing in this list ?

UMC had x86 tech, and ALI (later ULI) was created by UMC to do these and other designs (mostly chipsets).
Since Nvidia bought ULI, then they must own some form of x86. Maybe not the specific modern Intel bus license, but it's there.

 

[_xxx_]

That one is about chipsets, not the CPU. NV can produce chipsets for x86 and also bought some IP through I think stexar (or which company was it, anyone?), but they can't produce x86 CPU's for PC's. No license there.

 

[Arun]

Uhm...
- ULi actually had a license for making these embedded CPUs, but it was in no way transferable afaik. Once again, those were CPUs, not chipsets.
- Cyrix indeed had no license, but I am not sure whether that agreement with Intel was generic or specific. I suspect it was only valid for Cyrix, and not other companies.
- The problem with VIA was that they hoped to get the NatSemi license by buying Cyrix. NatSemi always had a license, even before buying Cyrix. However, Intel argued that the license was not transferable, and VIA couldn't benefit from it.

I don't think a x86 license is much of a problem for NVIDIA either way. If they need one, they just need to send in their lawyers and they'll get one. It's legally impossible to keep a company with the necessary resources out of such a large market... It's one thing to close your eyes on a monopoly, it's another completely to sponsor it.

 

[_xxx_]

Cyrix had the license through IBM, they were doing the CPU's together back then and just sticked the respective label onto them after the end of line test IBM and Cyrix Pentium equivalents were the very same CPU.

Also, you're talking about embedded, I'm talking PC. Those are different licenses. Of course nV could get one, but they haven't got it yet as of now.

 

[Arun]

No, that 'license' is what the lawsuit was all about. Intel agreed that Cyrix could manufacture x86 CPUs, as long as it was at IBM or another company with a license. Before that agreement, Intel argued that they could not. Cyrix themselves did NOT have a formal license, at least not before Intel agreed they could manufacture at IBM legally.

Yes, embedded and PC are possibly different licenses, I'm not sure. Either way, from a legal perspective, it should be easy to get one, as I said. That's not a real problem, IMO, as long as NVIDIA can prove they're serious about it. Which, they indeed, had no reason to do in the past - or even before they're a bit further ahead into their designs.

 

[_xxx_]

Dunno, I still can't see what sense it would make for nV to develop a x86 CPU. They can hardly produce anything that could beat Intel and AMD designs.

 

[Arun]

Dunno, I still can't see what sense it would make for nV to develop a x86 CPU. They can hardly produce anything that could beat Intel and AMD designs.

The point would presumably not be to create competitive designs against Intel and AMD's high-end. Instead, the target market would be <=$600 or so. Look at what you have today for that price at DELL: An Athlon X2 3800+ or a Pentium E2140. Around ~$350, you'll have a Conroe-based Celeron or a Sempron 3600+...

Of course, VIA's upcoming CN architecture could be very competitive in that market, and even more so if it was integrated in a single CPU-GPU-Southbridge chip. It will be especially interesting to learn about CN's die size.

 

[_xxx_]

I see no huge profits there, but that might be more likely. Though I think if, they'll do an embedded design rather than a PC CPU.

 

[Arun]

I see no huge profits there, but that might be more likely. Though I think if, they'll do an embedded design rather than a PC CPU.

No huge profits? That seems rather ridiculous to me, you just need to do some basic math to realize the potential profits are very substantial: desktops+laptops represent 200M+ units/year.

10% of that market with ASPs of $40 (for a single-chip CPU/GPU/Southbridge) and gross margins of 50% would correspond to $800M+ of revenue and $400M+ of gross profit/year. And that's before counting the potential extra revenue you could get in embedded/car navigation/etc...

The most obvious reason for entering that market, however, is that southbridges alone isn't a stable market going forward. In the next 3 to 5 years, you certainly should expect VIA and SiS to die or be acquired, IMO. Furthermore, all IGPs will be integrated into CPUs by then too, reducing the amount of potential differentiation for NVIDIA.

The net result of all this is that unless NVIDIA invests in CPUs, I find it rather unlikely that their MCP business will remain highly profitable in that timeframe. So the question really is whether to shred that business or invest in another one to complement it. And given that it's not like NVIDIA doesn't have enough money lying around nowadays...

 

[_xxx_]

No huge profits? That seems rather ridiculous to me, you just need to do some basic math to realize the potential profits are very substantial: desktops+laptops represent 200M+ units/year.

Not what I meant, I mean they wouldn't be able to effectively compete against the two big CPU hogs unless they produce a superior design (which would be a little wonder for a first-shot product). So either they'd have to compete on prices (which is a lost battle from the get-go) or have a CPU which will spank the similarily priced competition (which I simply doubt). Also, they'd need to widely penetrate the market first, which is another serious obstacle.