NVIDIA: Record Q3 + $1.85B in cash

[dizietsma]

Graphics chips for the new Apple iConsole ?

 

[_xxx_]

Re. Ageia: I think the advent of pervasive multi-core CPUs have renders them largely irrelevant as far as their hardware is concerned.

Not only CPU's but also all the SLI/XF-physics thingies and such. Noone needs or will ever need a discrete physics hardware, I said it here multiple times in all the countless Ageia threads. They could have made it as a SW/API provider, but there they lag ages behind Havok & co as far as the acceptance goes.

Ageia = cannon fodder, period.

Arun: what's the licensing situation with VIA? Will their x86 license stay if they get bought? Because that is really the only valuable thing about them.

 

[Geeforcer]

Graphics chips for the new Apple iConsole ?

That I doubt but there are plent of "i..."'s and "...mac"s that might need graphic chips.. take a pick:

New iPod, new iPhone, new iMac, Powerbook, etc, etc.

 

[_xxx_]

I thought they already were delivering chips for the iPhone, Apple TV and iPod?

 

[Arun]

re Ageia: Of course it's a turd, otherwise you couldn't buy them for $50M or less, pfff!

They got $40-50M venture capital iirc, so my expectations is they'd just like their money back. If what I was saying is "zomg NV needs the hardware!" then they'd have to pay more than $50M. The justification is obviously not that.

The point is that GPU Physics Acceleration hasn't picked up *at all* (are there even any Havok FX licensees?! And now they're owned by Intel, ugh!) and that what the world truly needs is a free Physics API that includes (GPU) hardware physics acceleration, along with a good CPU path. There is no such thing today.

If someone bought Ageia at this point, it wouldn't be for the hardware, it would be for the API (to a certain extend) and for the engineers. The motivation would be to revamp the PhysX API and create a new acceleration path based on another form of hardware, such as via CUDA. AMD would obviously prefer if it was done via D3D or OGL, but that's precisely why NV might want to eventually buy Ageia.

Alternatively, if you've got the patience, you might as well wait for them to go bankrupt and buy them for less - but you run the risk that engineers might have left the ship already. In fact, I don't know if that's not already the case, heh.

 

[Arun]

I thought they already were delivering chips for the iPhone, Apple TV and iPod?

PortalPlayer used to deliver chips for the iPods, but they are no longer in any current Apple hardware. This is presumably the reason why they sold off to NV for such a low price: their revenue and income figures for 2007 would have been downright catastrophic.

As for the iPhone, same thing, despite rumours PortalPlayer had nothing in it (or they wouldn't have sold off, most likely) and NVIDIA had none of their own chips there either.

NVIDIA does have a GeForce 7300 in the Apple TV, however. NV is likely to have future vers of that too IMO, but it's not a huge market. iPhone and Apple products still using Intel IGPs is where NV's focus most likely is right now.

 

[_xxx_]

I think MS will be doing something on the physics front.

Thx for the info on Apple, I just had that wrong then

Any info on VIA's x86 license in case of acquisition?

EDIT: thinking back, I almost bought a pile of nV stock back in the Riva128ZX days shortly before TNT came out but I decided against it

P.S. can we please include this smiley here?

 

[Arun]

I think MS will be doing something on the physics front.

That'd be even better for both NV and AMD, but honestly, I'm not seeing a lot of evidence of that. So sure, if that happens Ageia is completely useless. But my predictions/suggestions obviously assume it won't happen.

Regarding VIA: Yes, their license would be cancelled, which kinda kills the point. However, I firmly believe it should be possible to negociate a new license, and there are legals reasons why Intel cannot simply refuse that. Copyright laws aren't there to enforce monopolies, so in practice the only thing they can do is add some conditions and get some compensation (one-time and/or per-unit).

And if the arguement is that Intel would rather prevent that at all cost or slow it down tremendously by going to court, then I'm not sure they'd even really want to do that if NVIDIA accepts to add some strict conditions to what they can do. For example, what about a license agreement to sell 'single-chip x86 solutions that include a southbridge, a GPU, and analog sound functionality on the same die.'

NV likely would never want to enter into any other market than that anyway (except to contonue to sell the products VIA already has, and they could just ask for a one-year permission to do that or whatever). So it does make a lot of sense from my perspective.

P.S.: Site feedback if you have any smiley requests, although it's probably not worth bothering if it's only one. If you have a larger set to suggest, though, do go ahead.

 

[_xxx_]

Oh, but don't forget about this:

http://www.theinquirer.net/en/inquirer/news/2006/06/21/microsoft-set-to-produce-its-own-physics-api

They indeed were/are offering jobs for coders experienced with physics:

http://members.microsoft.com/careers/search/details.aspx?JobID=6b94ac4f-0627-4851-8e6a-633186d96261

"Direct Physics", so go guess

 

[Arun]

Yes, but that was already discussed. What they were really doing is look into how to accelerate physics via SIMD processors such as GPUs, not the creation of a standalone API. See Demirug's posts in this thread: http://forum.beyond3d.com/showthread.php?t=31804

An industry standard, MS-backed or or just NV/AMD-backed, to accelerate physics via D3D10 would go a long, long way if it was free for anyone to use it. The current situation (or the old one, I don't know Intel's plans for that) with Havok FX is just a mess, and all companies involved were naive at best to think such a ridiculous locking-up of the technology could ever work out. Whoever made the Havok FX-related decisions at NV and ATI clearly had no proper understanding of the market.

 

[_xxx_]

Oh, I think they just pulled a fast one to be "there" since PhysX was the new cool at given moment

 

[Arun]

I want to rectify something about my earlier comments on the benefits of an 'acquisition spree'. First of all, looking at Clearspeed's financials, they're losing more money than I thought which really makes them too expensive for what they are. Secondly, Icera would be significantly more expensive than I thought, also not making them a viable acquisition target.

So overall, I'm down to VIA/Ageia/smaller companies. One smaller company I would definitely consider to be a very interesting acqusition target is Coresonic: http://www.coresonic.com/technology/index.html

Now, their tech might be interesting for handhelds, but I'm not convinced it's the best approach there until we get multi-purpose RFs to be mainstream (such as this thing). Meanwhile, where it could truly shine is chipsets.

Just imagine: integrated multi-purpose MACs in your chipset, with the motherboard manufacturer adding the RFs he wants to in order to differentiate his product (or laptop). That could be a significant advantage in the market, IMO.

 

[Maintank]

What does Via bring to the table that is worth 600 million?

 

[Arun]

What does Via bring to the table that is worth 600 million?

Mostly this: http://en.wikipedia.org/wiki/VIA_Isaiah and the associated engineers: http://www.centtech.com/company.htm

I would value the core at about $200M (given the existing R&D that went into it, and the fact it would allow NV to release a x86 product much earlier than any other alternative), and the associated engineers at about $100M.

The other groups still have about 1000+ engineers AFAIK. Some are in the USA, but the majority are in Asia. Given most ways that are often used to evaluate a company's value related to the number of engineers, it should be easy to see $600M isn't very expensive at all if they can get it at that cost (as explained before, I substracted VIA's cash position to their current market cap and added a nice premium).

 

[Razor1]

rofl...


Not sure, are you Doug? Strange, you were pretty damn sure last week when you told your clients to sell the stock since the revenue "peaked" the previous quarter.

Dime a dozen

 

[INKster]

Mostly this: http://en.wikipedia.org/wiki/VIA_Isaiah and the associated engineers: http://www.centtech.com/company.htm

I would value the core at about $200M (given the existing R&D that went into it, and the fact it would allow NV to release a x86 product much earlier than any other alternative), and the associated engineers at about $100M.

The other groups still have about 1000+ engineers AFAIK. Some are in the USA, but the majority are in Asia. Given most ways that are often used to evaluate a company's value related to the number of engineers, it should be easy to see $600M isn't very expensive at all if they can get it at that cost (as explained before, I substracted VIA's cash position to their current market cap and added a nice premium).

I still think it would be cheaper for them to just negotiate a x86 license with Intel and start core development from scratch than having to buy a bunch of peripheral crap and potentially hidden financial ordeals (a.k.a., "traps") just to reach a tiny part of the company, the key engineering teams who have, so far, not been able to demonstrate that they too can build a high performance core.
I doubt that Nvidia would be even interested in entering this market if not for the ultra-high-end segment, highly profitable niche first.
And even Intel would have little trouble going with that, since it would always limit Nvidia to a small section of the market (and they would always depend on 3rd party foundries anyway, so no significant volume growth would ever be in perspective).

As it is, VIA sounds less than thrilling with its surplus of unneeded "baggage" and money-leaking operations.
It's definitively not the same type of easy decision for JHH as were PortalPlayer and Uli, since those were an order of magnitude cheaper.

 

[FUDie]

Arun, you should bold the outlook for next quarter, because that's what's going to drive the stock back to 37+ tomorrow.

NVDA	33.36	-0.48

That's today's close. Were you one of the crazy one's buying in AH last night?

 

[Geeforcer]

NVDA    33.36    -0.48

That's today's close. Were you one of the crazy one's buying in AH last night?

Sadly (fortunately?) no. I cashed out at 38.50 not so long ago.

 

[FUDie]

Sadly (fortunately?) no. I cashed out at 38.50 not so long ago.

Good for you, but your prediction about "$37+" today was off by a wee bit. NVDA was up quite a bit in AH last night, but once the professionals got to work, it didn't last.

 

[Arun]

I doubt that Nvidia would be even interested in entering this market if not for the ultra-high-end segment, highly profitable niche first.

No, that's precisely the niche they do NOT want to enter.

As it is, VIA sounds less than thrilling with its surplus of unneeded "baggage" and money-leaking operations.

They've been shedding expenses in those progressively, so it could be worse, but I see your point.

It's definitively not the same type of easy decision for JHH as were PortalPlayer and Uli, since those were an order of magnitude cheaper.

PortalPlayer was hardly an order of magnitude cheaper, and they already realized they would be far from profitable in 2007. But yes, it *is* a harder decision, let alone because it's a more expensive one!