IBM Out of Chip Manufacturing Business and General Execution Thread

[Kaotik]

http://www-03.ibm.com/press/us/en/pressrelease/45110.wss

There's IBM's own press release

IBM will reflect a pre-tax charge of $4.7 billion in its financial results for the third quarter of 2014, which includes an asset impairment, estimated costs to sell the IBM microelectronics business, and cash consideration to GLOBALFOUNDRIES. Cash consideration of $1.5 billion is expected to be paid to GLOBALFOUNDRIES by IBM over the next three years. The cash consideration will be adjusted by the amount of working capital which is estimated to be $200 million.

 

[3dilettante]

IBM really didn't want the fabs anymore, but such a very motivated seller means there's a very good chance that it's not something worth paying money for.

It doesn't seem like GF was interested in the fabs and their heavy costs as much as the engineers and IP. The amount IBM paid is likely heavily based on the burden the manufacturing operations represent, on top of IBM contracting manufacturing from the division it has separated from.

Those more familiar with the fab industry would probably have more insight, but it reminds me of the GF spin-off where ATIC paid AMD for a struggling silicon manufacturing arm.
Hector Ruiz wrote a book that discussed about how he suckered Abu Dhabi into bidding high for GF, perhaps they learned a lesson about buying factories that can't seem to sell enough product, and how few other bidders there likely would be.
Whether there still isn't a catch is a different question.

Granted, GF as an organization has a far more intimate understanding of the business than a sovereign wealth fund would have had back then.

 

[Putas]

So is this a sign that IBM giving up on their Power CPUs?

Not at all, GF have to manufacture them for 10 years.

 

[Erinyes]

IBM really didn't want the fabs anymore, but such a very motivated seller means there's a very good chance that it's not something worth paying money for.

It doesn't seem like GF was interested in the fabs and their heavy costs as much as the engineers and IP. The amount IBM paid is likely heavily based on the burden the manufacturing operations represent, on top of IBM contracting manufacturing from the division it has separated from.

Those more familiar with the fab industry would probably have more insight, but it reminds me of the GF spin-off where ATIC paid AMD for a struggling silicon manufacturing arm.
Hector Ruiz wrote a book that discussed about how he suckered Abu Dhabi into bidding high for GF, perhaps they learned a lesson about buying factories that can't seem to sell enough product, and how few other bidders there likely would be.
Whether there still isn't a catch is a different question.

Granted, GF as an organization has a far more intimate understanding of the business than a sovereign wealth fund would have had back then.

Yea..according to reports..the division lost $700m last year. Even if they paid GF $1.5 billion, or rather $1.3 billion if we exclude the $200 million working capital, that will be lost in 2 years if they continue at the present rate. Do GF really think they can turn it around that fast?

 

[lanek]

Yea..according to reports..the division lost $700m last year. Even if they paid GF $1.5 billion, or rather $1.3 billion if we exclude the $200 million working capital, that will be lost in 2 years if they continue at the present rate. Do GF really think they can turn it around that fast?

Well, i dont think GF will run the fabs as IBM was doing ( so by product their own Power processors only ).

Basically, for the number of chips produced ( and sold ), the cost of the Fabs was way too high. Its why today, you nearly have only Samsung and Intel who are product their chips in their own Fabs now.

 

[Silent_Buddha]

IIRC, IBM didn't only produce their own chips. They also made chips for others similar to TSMC.

Regards,
SB

 

[lanek]

IIRC, IBM didn't only produce their own chips. They also made chips for others similar to TSMC.

Regards,
SB

Some, this was extremely limited... nothing like GF or TSMC .... their clients base and profit are, well, the money they do with that speak for itself.

 

[Kaotik]

IIRC, IBM didn't only produce their own chips. They also made chips for others similar to TSMC.

Regards,
SB

Last time I remember IBM doing chips for anyone else was NVIDIA in the GF FX (or was it GF6?) times

 

[Erinyes]

Well, i dont think GF will run the fabs as IBM was doing ( so by product their own Power processors only ).

Basically, for the number of chips produced ( and sold ), the cost of the Fabs was way too high. Its why today, you nearly have only Samsung and Intel who are product their chips in their own Fabs now.

But its not as if GF's fabs are running full and they really need the extra capacity do they? If IBM's fabs are in an even worse position, where are they going to magically find the customers to fill capacity?

True but also note that Samsung probably produces more chips for Apple than for themselves. And even Intel is doing a bit of foundry work for the likes of Altera AFAIK.

Last time I remember IBM doing chips for anyone else was NVIDIA in the GF FX (or was it GF6?) times

Yep..GF FX is the only one I remember IBM doing. And they switched back to TSMC pretty soon.

 

[Putas]

It should not be so hard to find somebody in need of 22nm process capable of making chips as complex as Power 8.
AFAIK IBM manufactured 130nm Geforce 6 line + NV36.

 

[Erinyes]

It should not be so hard to find somebody in need of 22nm process capable of making chips as complex as Power 8.
AFAIK IBM manufactured 130nm Geforce 6 line + NV36.

Well if it isn't that hard..surely IBM would have found them by now? Just to clarify, I'm just very curious about this whole deal. Obviously there must be some solid reasons for GF to go for it, I'm just trying to figure out what they are.

 

[I.S.T.]

Last time I remember IBM doing chips for anyone else was NVIDIA in the GF FX (or was it GF6?) times

They manufactured CELL and the Xenon for a while.

 

[AlNom]

They manufactured CELL and the Xenon for a while.

& Nintendo*. I suppose all their orders would be quite a bit less lately.

*maybe not particularly lucrative considering die sizes.

 

[Kaotik]

Those three were more or less "their own chips" though, even if MS was part of Xenon development, Nintendo part of their CPUs development and Sony & Toshiba part of Cell

 

[willardjuice]

Those three were more or less "their own chips" though, even if MS was part of Xenon development, Nintendo part of their CPUs development and Sony & Toshiba part of Cell

They make 32nm cisco chips too. IBM does a lot of low volume (but highly custom) niche stuff in addition to their power chips. It's important to remember fishkill in general is a small fab that has to cover a broad range of nodes (22nm-1x0nm, eventually 14nm finfet). IBM was never going to match tsmc, gf, etc. They targeted different markets (highly specialized vs high volume). It's just the highly specialized fab market pays out far less than it used to.

 

[UniversalTruth]

And now there is the deal:

GLOBALFOUNDRIES Obtains U.S. Government Clearance for IBM Microelectronics Business Acquisition

http://globalfoundries.com/newsroom...epage&utm_medium=brief&utm_campaign=IBM150629

 

[Megadrive1988]

IBM unveils world’s first 7nm chip
With a silicon-germanium channel and EUV lithography, IBM crosses the 10nm barrier.

IBM, working with GlobalFoundries, Samsung, SUNY, and various equipment suppliers, has produced the world's first 7nm chip with functional transistors. While it should be stressed that commercial 7nm chips remain at least two years away, this test chip from IBM and its partners is extremely significant for three reasons: it's a working sub-10nm chip (this is pretty significant in itself); it's the first commercially viable sub-10nm FinFET logic chip that uses silicon-germanium as the channel material; and it appears to be the first commercially viable design produced with extreme ultraviolet (EUV) lithography.

First, the facts and figures. This is a 7nm test chip, built at the IBM/SUNY (State University of New York) Polytechnic 300mm research facility in Albany, NY. The transistors are of the FinFET variety, with one significant difference over commercialised FinFETs: the channel of the transistor is a silicon-germanium (SiGe) alloy, rather than just silicon. To reach such tiny geometries, self-aligned quadruple patterning (SAQR) and EUV lithography is used.

http://arstechnica.com/gadgets/2015/07/ibm-unveils-industrys-first-7nm-chip-moving-beyond-silicon/