Prediction: In a year, NVIDIA buys the combined AMD/ATI

[Jakob]

Umm. Given that the market has had a week to digest this now, AMD/ATI is a $14B company, not a $9.2B one. Any market perceived "overpricing" of ATI has already been taken out of AMD's hide.

Not sure I understand that. AMD's price already reflects the planned purchase. When tge deal closes, AMD will pay cash and stock for all of ATI. Do you expect the market cap of AMD to suddenly jump to $14B from $9B when it closes? I don't think so... The purchase is already priced in to AMD, no?

 

[Arun]

Not sure I understand that. AMD's price already reflects the planned purchase. When tge deal closes, AMD will pay cash and stock for all of ATI. Do you expect the market cap of AMD to suddenly jump to $14B from $9B when it closes? I don't think so... The purchase is already priced in to AMD, no?

The number will increase proportionally to the amount of shares created in order to buy ATI. Since most of the deal is cash, it won't jump by 5B, more like 1.5B. That's because the share emission is already counted in the valuation by investors, but the actual shares are not on the market yet.

Uttar

 

[Geo]

Hrmm. It's going to be more complicated than that I think. The other option would be to consider AMD "today" is a 4B company --do you really expect that is reasonable either?

There are two tickers out there today that add up to $14B in market capitalization, they are becoming one. I don't think I'd be willing to say what the market thinks that combination is worth until there is onely one ticker for a little bit, and people have received the money for their ATI stock.

What portion of ATI cashout stock holders are going to reinvest in AMD? I don't know, and neither does anyone else.

Edit: For instance, would you agree that the issuance of the new stock has been priced in? Probably, in the sense that those who don't want to be part of that have already left the building. Is the actual issuance of that stock included in the market cap today? I don't see how it could be, as it hasn't been issued.

 

[Jakob]

Ah, thanks Uttar, that makes sense. So the current valuation of AMD will be increased when the new shares are issued. But since the deal is 80% cash, the increase will be way less than $5B. Here's the specifics from AMD: "Based upon the closing price of AMD common stock on July 21, 2006 of $18.26 per share, the consideration for each outstanding share of ATI common stock would be $20.47, comprised of $16.40 of cash and 0.2229 shares of AMD common stock." [1]

Doing some math, ATYT has about 250M shares, based on $5B / $20.11. That will result in 250M*0.2229 == 56M new AMD shares. AMD already has $9.2B / $19.06 == 484M shares. Adding 56M shares gives a market cap of $19.06 * 540M == $10.3B, or just a 12% increase in the current valuation of AMD. Certainly something, but not all that significant, if my math is right.

1. http://www.amd.com/us-en/assets/content_type/DownloadableAssets/AMD_ATI_Fact_Sheet.pdf

 

[Geo]

Should we start a pool? "Pick AMD's market cap 30 days after ATI deal closure?"

 

[Jakob]

Should we start a pool? "Pick AMD's market cap 30 days after ATI deal closure?"

Heh. Well, to PYMWYMI, you should go long on AMD if you think it is going to be higher than $10.3B. I'm betting on the other side.

 

[Geo]

I have way too much fun talking about this stuff to mess it up by having a personal pecuniary interest in stock price movements of a given IHV.

 

[Rolf N]

Ah thx for the correction Acert93 :smile: . Well given that ATi is definitly going unified, I don't see them going with any shader structure other then 1x1 per pipeline, anything else would just defeat the purpose of unification.

example:

Lets say a program calls for 5 vertex and 5 geomotry shaders, <...>

Stop right there.
A program just can't require any number of "shaders", unless I'm grossly misunderstanding you.

A program can have a number of instructions but that's still not the same size of problem you're describing. Are you drawing a comparison with the classic number of texture units per pipe, perhaps? If yes, I don't think it's very relevant. Average shader program lengths already exceed and are still growing much faster than the number of textures per pass.

The point of avoiding architectures with more than 1 "TMU per pipe" is that you're using transistors that will sit idle in a statistical 50% of the time, and you could use them to speed up the trivial (1 texture or no texture) cases which are quite frequent.
The same reasoning doesn't seem to apply for shader ALU structures.

 

[Razor1]

Stop right there.
A program just can't require any number of "shaders", unless I'm grossly misunderstanding you.

A program can have a number of instructions but that's still not the same size of problem you're describing. Are you drawing a comparison with the classic number of texture units per pipe, perhaps? If yes, I don't think it's very relevant. Average shader program lengths already exceed and are still growing much faster than the number of textures per pass.

The point of avoiding architectures with more than 1 "TMU per pipe" is that you're using transistors that will sit idle in a statistical 50% of the time, and you could use them to speed up the trivial (1 texture or no texture) cases which are quite frequent.
The same reasoning doesn't seem to apply for shader ALU structures.

Didn't mean require but no matter what will be using x number of shader units.

Unless each ALU has its own dedicated components to make them be able to do vertex/geometry/pixel shader there is no way to have a unified structure that could truely be efficient in a 2x1 or 3x1 or what ever scenario. If a shader array is being used for a certain type of shader, that array even if it has extra ALU's free will not be able to do another type of shader, I think, I'm not 100% sure about this, but I don't see how it will be possible since the different type of shaders will be calculated at different times in the rendering of an image. (by type I mean vertex/geometry/pixel)
typo

 

[Brimstone]

If nVidia were to go shopping, my guess they'll buy Creative (Creative Labs of sound blaster fame).

 

[_xxx_]

If nVidia were to go shopping, my guess they'll buy Creative (Creative Labs of sound blaster fame).

Dunno, I thought Creative was bigger than nV (though I have no data).

 

[Sc4freak]

In other news, Microsoft buys Intel, AMD/ATI and Nvidia. Problem solved. (Wouldn't be surprised if Microsoft had enough money to do so, as well).

 

[phenix]

If AMD/ATI really fails then they surely will be sold to somebody. But not in 1 year. Something this big can happen only in the next 5 years I think. And Nvidia cannot be the company who buys them. It would be either Microsoft or IBM. IMO

 

[Jakob]

Dunno, I thought Creative was bigger than nV (though I have no data).

http://finance.yahoo.com/q?s=CREAF

market cap: $430M

 

[Jakob]

http://finance.yahoo.com/q?s=CREAF

market cap: $430M

Yeah, this seems small to me too. They bought 3d labs for $100M, and they have 5,000 mostly engineering employees. Sure they have challenges, (ipod, integrated killing sound blaster) but only a paltry $430M? Seems too cheap. Not sure if that valuation is for the ENTIRE company, or what. Seems they also trade on the Singapore stock exchange.

 

[Geo]

Given they are a Singapore company, I'd be wondering what that Nasdaq ticker covers.

 

[Rolf N]

If nVidia were to go shopping, my guess they'll buy Creative (Creative Labs of sound blaster fame).

That would be sweet. This might be a way for all of Creative Labs' customers to get proper drivers for the hardware for once.

 

[Dave Baumann]

Not sure, but it was my understanding that Creative had the backing of a rather large IP house behind it - perhaps they are majority sharholders of it. However, I'm not sure how buying Creative would give much in the way of strategic benefits.

 

[INKster]

Well, Creative DOES own 3DLabs, and 3DLabs was one of the most active OpenGL ARB members when they were deciding what should OpenGL 2.0 be like.

I gather their IP in the 3D graphics field still has something worth buying it for...?

 

[_xxx_]

I think that 3DLabs IP is worth nothing to nV, they do have more capable HW after all. The only interesting thing about Creative is the SB brand and the X-Fi, nothing more. I don't see why nV would be interested in that, especially given that they have enough competence regarding sound HW in house already.