AMD Spin-Off GlobalFoundries Gets First Non-AMD Customer

Vigile writes "Since the company was spun off in March, GlobalFoundries has struggled to answer how it will survive and compete against powers like TSMC and UMC in the global world of chip manufacturing. Part of that answer came today when they announced the company's first customer, excluding AMD. STMicroelectronics will be using GlobalFoundries' 40nm lower power process technology for future cell phone SoC designs in the second half of 2010. While one customer won't drive enough revenue to make the foundry completely independent, it is an important step in the right direction and could lead to other customers finally making the leap."

Re:Great

[cabjf]

To free AMD up to outsource to other foundries, to allow GlobalFoundries to take on outside customers, to prevent losses in one area from affecting total company health, to allow outside investment in either company where the investor would not be interested in the other area of business...

Re:Great

[maxume]

So why did they buy ATI, to get rid of CXOs and fuck around with the accounting books?

It isn't like a a multi-million dollar division is going to be run by that many fewer people.

Re:40 nm process...

[downix]

Really? *looks at his 45nm Phenom II chip humming along* You might want to double check your claim.

Re:40 nm process...

[level_headed_midwest]

You got a bunch of numbers screwed up. Most of AMD's current desktop and server CPUs are 45 nm (Phenom II, Athlon II, Opteron >2360/8360), while the laptop CPUs are 65 nm and will be going to 45 nm in a couple of months. AMD's GPUs are mostly 55 nm (Radeon HD 3xxx and 4xxx, except the HD 4770) with one 40 nm unit (HD 4770.) Most of Intel's CPUs are 45 nm, but all of the Celerons and some Pentium Dual Core and Core 2 Duo units are still 65 nm. Intel is going to the 32 nm process, but nothing 32 nm from them is going to ship for some time.

Re:AMD reminds me of 3Dfx....

[JorDan Clock]

But fabs are incredibly expensive. Fab 1 in Dresden has cost something like $6 billion dollars to date, from construction to current upgrades. Plus, you have to account for research and development costs of moving to smaller manufacturing sizes, which according to Intel was something like $600 million to $900 million to move from 45nm to 32nm. For a company like AMD that still has a lot of debt from the purchase of ATI, that's a lot of money. Add to that that the New York fab is estimated to cost $4.2 billion and it's easy to see why they spun this off. The productive coordination process isn't THAT much smoother to warrant all that overhead.