LONDON – IBM appears set to gradually back away from semiconductor manufacturing and to rely for its leading-edge silicon on Samsung and GlobalFoundries as foundry suppliers. Both Samsung and GlobalFoundries are set to have wafer fabs for foundry operations in the United States, which could help cement relations.
IBM, GlobalFoundries and Samsung are set to co-host the Common Platform Alliance Technology Forum at the Santa Clara Convention Center in California on January 18, 2011, but whereas GlobalFoundries and Samsung are set to spend $12 billion in 2011, IBM's spending is likely to be less than $500 million according to data from an analyst at market research firm Gartner.
For years IBM has been a beacon of research in semiconductor technology and had carried that work into manufacturing with its own processes and wafer fabs. As well as including aggressive process miniaturization its pioneering work has included high-frequency RF on CMOS and silicon-on-insulator. Such was its leading position in research that it was able to drive the Common Platform Alliance as a means of sharing semiconductor process research costs.
But an analysis of the capital expenditure plans show that IBM is gradually allowing itself to exit from leading-edge manufacturing at high volume. IBM appears to have joined the broad class of semiconductor companies that will never build a major wafer fab again.
An examination of capital expenditure data from Gartner reveals that the last time IBM spent more than $1 billion on semiconductor capital expenditure in one year was 2004, when it was the 11th biggest spender.
IBM is not in the top 20 of semiconductor capital expenditure in 2010 and nor is it expected to be in the top 20 in 2011, according to Bob Johnson, research vice president with Gartner.
There are going to be 10 or 11 companies in the billion-dollar capex club in 2011, according to Johnson's latest figures. The top 20, with their spending in millions of dollars, is currently set to be:
This has been the stratergy adobpted by many semiconductor biggies. TI sold some of its fab (they retained analog fabs) and went for TSMC. If this helps companies to cut down expenses, nothing wrong in this.
I am tired of everybody saying that fabs are consolidating because capital equipment prices are skyrocketing. Has anyone thought that they might be confusing cause-and-effect at least a little bit. For the capital equipment makers, the cost of producing better-and-better performing systems has not gone down. Yet there are increasingly fewer customers out there buying fewer tools because they want to save money by, "consolidating". I hate to break it to these guys but consolidation only works if you are the only one doing it. When everyone does it, prices skyrocket because there are fewer tool sales to spread your fixed costs across. This does not even take into account R&D and additional costs from the loss of capability and experience as engineers and technicians flee the capex industry.
No matter how much they try to streamline to reduce their costs, the more the hidden costs will, "skyrocket", no matter how much they whine (don't hold your breath for capex manufacturers to pick up 450). They will end up with more bottlenecks and, "premium", pricing for in-time orders. Intel knows this and will continue making money from the foolishness of others.
Good article. I heard IBM had discussions with Global Foundries this summer about selling their Fishkill fab to them. I also heard Global Foundries decided to pass on the opportunity since IBM was asking for too much money.
The research is the source of IP and therefore the differentiator in the marketplace. With fab costs skyrocketing the need to leverage the fab price tag over many chip streams and customers it becomes a better business model. As long as there is sufficient capacity and 2nd/3rd sourcing options this will continue to be the way to keep costs down and enabling resources to be concentrated on generation of money streams.
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.