Globalfoundries is only a little over three years old, but is already the world's second largest pure-play foundry, with estimated annual sales of about $4.5 billion in 2012. All seems to be going well.
The company was initially formed by bringing together the manufacturing capabilities of a somewhat distressed processor manufacturing company Advanced Micro Devices, that was looking to go fabless, and the oil riches of the Gulf state of Abu Dhabi. Back in 2008, Abu Dhabi had a plan.
The billions of dollars Abu Dhabi has provided – an investment in adapting the emirate to a post oil-economy world – that has allowed Globalfoundries (Milpitas, Calif.) to make such rapid progress. But it now seems that the time of easy money – and we are talking of well in excess of $10 billion – could be coming to an end.
With talk of a future IPO of shares, Globalfoundries is entering a new phase of its existence and a reading of the runes suggests a significant change of heart in Abu Dhabi government circles. The announcement of a goal for Globalfoundries to be profitable by 2015 and a potential IPO suggests Globalfoundries' parent - the sovereign wealth investment vehicle Advanced Technology Investment Co. – wants to stop handing out cash and indeed is looking to reduce and, possibly, to eventually eliminate its stake in the chip maker.
At the same time, there was no explicit mention at the time we reported on the expected IPO that of a now-moot plan to build a wafer fab in Abu Dhabi. That was expected to cost another $6 billion to $8 billion. The apparent abandonment of the Abu Dhabi fab plan suggests that the numbers aren't adding up for the wealthy emirate. Without a Middle Eastern manufacturing and technology infrastructure as an end-goal, it is hard to see how Abu Dhabi would justify continued involvement in Globalfoundries.
Mike Noonen, Globalfoundries' executive vice president of worldwide marketing and sales, emphasized that profitability is the first step, and that step would allow Globalfoundries to take "control of our own destiny."
That could be another way of saying that, after 2015, Globalfoundries may have to fund future wafer fabs from profits, which will be no easy matter. Already, United Microelectronics Corp. (UMC), the previous No. 2 foundry, with annual revenues similar to Globalfoundries', is considering whether it can afford to keep up with the semiconductor race to smaller geometries.
If a company is strategic to a nation's or a region's interest, steps can be taken to reduce the cost of capital for that company, making spending decisions much easier.
But without a manufacturing foothold in Abu Dhabi, Globalfoundries looks set to be a global company that will have no one place to call home. That could be a significant disadvantage against formidable rivals like Taiwan Semiconductor Manufacturing Co., Samsung or Intel.
Kresearch/Peter! This logic has flaws. GloFo never was a foundry before acquiring Chartered at the end of 2009. (only had one customer, it's own parent company). Samsung is already doing substantial foundry business with major customers like Aple, Qualcomm, Xilinx, and etc..
I don't see a direct connection between funding for Global Foundries and building a fab in Abu Dhabi.
The Emirate recognizes that its economy must have a broader base. But an economic contribution to an economy doesn't require a facility in the economy. Money can be made anywhere in the world. The initial investment in Global Foundries was likely an effort to generate revenue and earn income. In such a case, one of the investor's questions will be "How much investment is required to make this a going concern, and when might we expect it to be profitable and producing a return on our investment?" It appears that the folks providing the funding have decided that it's time the investment started to provide a return. An IPO, if successful, does what any startup investor wants: it repays what they invested, and provides a return on top of that. I think we can assume Abu Dhabi's fund will retain a stake in Global Foundries, and anticipate a rise in the value of their holdings and a revenue stream from dividends if things go well.
Building a fab in Abu Dhabi itself is a different matter. Fabs are enormously expensive. Global Foundries exists as a "pure play" fab operator because of that. Fabs are so expensive that few companies can afford to build them. The soaring cost of building fabs is behind outfits like AMD wanting to go fabless, and concentrate on being designers of things others like Global Foundries will actually make.
Does it make economic sense to build a fab *in* Abu Dhabi? Who would the customers be? What would the fab make? What would the impact of having the fab halfway around the world from the manufacturers who would actually use what the fab made in products be?
The folks in Abu Dhabi may have looked at it, and decided they couldn't get a big enough return to justify the investment.
Meanwhile, the question is whether Global Foundries *can* meet it's profitability and IPO targets, but that would be true in any case.
I don't expect Glofo could achieve that goal in 2015 with 14nm FINFET in mass production based on past record. I bet Glofo set on sale in 2014 and SAMSUNG will acquire it and UMC, SMIC will be acquired by TSMC by then. In 2015, INTEL, TSMC and SAMSUNG became top three foundries.
Abu Dhabi isn't in the desert. It is one the United Arab Emirates on the coast of the Persian Gulf.
I suspect other consideration than the availability of water rank higher as to whether it makes sense to build a wafer fab there.
If water is needed it will be brought. Bringing people is harder.
In the year 2015 ....
there will be a reality check...
TI was always a driver in developing leading edge but they smelled the coffee...
TI getting rid of mobile business is telling -
Amazon (eventually) picking up TI's mobile business just shows a complete lack of understanding the manufacturing aspect of advanced logic chips.
here's what ASML has to say...
In regard to 20 nanometer nodeacegoa very little intention node, so you’ll see about from the 28 nanometer to the 20 nanometer node about a 1.7, 1.8 times the investment for the same number of wafer starts at the 28 nanometer node..
considering how much water a wafer fab needs it's silly to build a fab in the desert anyway ...
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.