The survey cited by "Chips and Change" may attribute tax advantage, supply of technical talent, quality of utilities, proximity to existing facilities and environmental permitting as the reasons for fab location, but if we go back to 1987, it was the Taiwanese government that took the risk (along with Philips) and formed TSMC ( http://www.fundinguniverse.com/company-histories/Taiwan-Semiconductor-Manufacturing-Company-Ltd-Company-History.html ). Taiwan created the market for contract semiconductor manufacturing... which enabled several fabless semiconductor companies in the US (not a bad deal for a lot of US employees). That made a lot of sense for Taiwan since the semiconductor industry didn't orginate in their country... they just had the nerve to buy into the industry and learned to compete.
Are employee salaries really a significant portion of semiconductor fab costs? I don't think so... I believe a semi fab's cost is driven principally by equipment cost/depreciation. In fact, the cost of lithography equipment represents 35% of an integrated circuit's cost (http://catrene.org/web/downloads/results_medea/2T304-LIQUID-result-outMEDEA%20(8-9-09).pdf ). The US is a leader in lithography equipment (Applied Materials, IBM) and Nikon (from Japan where labor also isn't so cheap) is another leader. If you compare Applied Material's and TSMC's annual revenue ($9.7B vs. $10.2B for 2008), equipping fabs with their high-tech machines isn't so bad and the US has a few jobs as a result of developing and producing all this fab equipment.
So this thread seems to have ventured into the philosophical area of whether the rest of the world should be allowed to grow their own economies by attracting (or subsidizing) manufacturing within their borders; or mistakes of US government. In many foreign countries, the government encourages their corporations to make long-term investments in high-tech. Case in point, OLED technology was invented in the US, but no US Company has had the courage to invest for ten years in maturing this technology for high volume low cost production. On the other hand, Korea and Japan have made huge investments in this technology for more than 10 years and it is about to pay off for them. In smart phones the display costs more than all the silicon... and there are more than 1.4B (that's billion) mobile phones produced each year!
So I believe the root of the problem in the US is the almost unanimous lack of corporate leadership (that includes middle managers, executives and boards)... I'm talking about true vision and the tenacity to stay the course. But these US corporate leaders are making decisions based on how they are being rewarded... bonuses and stock based on corporate performance this year, not five years or ten years from now. So is it a surprise that US companies nearly strangle long-term R&D so short-term profits look better? It is interesting to note that Samsung doesn't even want to be traded on any of the US stock exchanges... I wonder why that is?
Labor cost isn't such a big part of the story. A 2003 study sponsored by the SIA found that, while China's labor costs were much lower than the US, they were too small a share of manufacturing expenses to drive location decisions, and the salary differential has declined since then.
A 2004 survey-based study cited in 'Chips and Change' came up with the following primary reasons for fab location: Tax advantages, Supply of technical talent, Quality of utilities, Proximity to existing facilities, and Environmental permitting process and/or other regulations.
Does anyone for this idea compare the labor cost between Asian foundries and US fab? Average Asian fab operator (technican/ specialist whatever you call them, I meant those people loading/unloading wafers pod to the tools)make 700 US dollar a month, US fab average hourly rate is over 10 USD easily. You get the picture. This is beyond any incentives that can be make up, plus other labor consideration,etc. While the US fabless IC houses are still driving the cost down and making more profits than foundries. Asking around, would anyone willing to pay mcuh higher price of any electronic for the US made IC inside?
As for the consideration of losing IC fab technology leadership, no need to worry, US still have Intel/ IBM, at least Intel is way ahead of the Asian foundries in terms of IC process technology, and they are good at keeping information inside.
This kind of claims is good for patriotism, but proably not making a lot of sense in terms of economics.
To be fair those EPA hurdles are there to keep our air and water reasonably clean, and unless we want to have a wage war with China (which is insane) labor costs aren't going to be going down. IIRC, didn't our corporate taxes go down a bit during the Bush years? Part of the problem with that is we have an obscene budget deficit and an even more obscene public debt, cutbacks alone wont make up for it.
"The US badly needs to balance its trade deficit as it is the root cause of all its problems today"
I'm pretty sure our massive overspending had a lot to do with both the trade deficit and the current economic crisis. We have had savings rates that have been near or below zero for several years now. Who's fault is that?
The US badly needs to balance its trade deficit as it is the root cause of all its problems today. A trade deficit of $840 billion is equivalent to as many jobs off shored and if we multiply the deficit by cost of manufacturing the same goods in the US, it will come close to 50% of the $14.4 trillion UD GDP.
Warren Buffet has suggested that we use import certificates to put limits on the imports to balance this deficit. An import certificate which limits the imports from a country as a function of exports to that country will eliminate this trade deficit in one go without alienating the trade partners.
Shifting manufacturing to the US will become imperative in the presence of import certificates. So it might become essential for these foundries to set up manufacturing facilities in the US for the US market or else they simply wont be able to sell in the US.
The only catch with import certificates is that they cannot be applied to commodities like crude which cannot be manufactured locally.
Tax rates & labor costs are probably the biggest reason and they are big ones, but the EPA hurdles are also a factor. Manufacturing does include chemicals. Mark, I like your idea and we have leverage points you didn't mention if you progress the idea a little further. The issue skilled labor and engineering jobs are equally key. We need more than burger flippers in this country.
If a company made a decision today to build a fab in the US, it likely couldn't get built and be qualified for production soon enough to be online when the industry transitions to 22 nm. So we're talking about a 16 nm fab which isn't a whole lot of runway before we reach the 11 nm quantum tunneling barrier (oxide thicknesses also reach a monolayer thickness) in ~2015-2016. In addition, interconnects at the 11nm node will have pitches ~50nm and the diameter of a single copper atom is ~0.25nm... so interconnect resisitivity and electromigration are going to get extremely difficult to solve at and beyond the 11nm node. See http://en.wikipedia.org/wiki/11_nanometer for other info.
I believe the more important question is in what direction will fabs go beyond the 11nm node to keep marching to the beat of Moore's Law? Will the industry uniformly transition to spin-based devices, single-electron devices or molecular devices? If the industry goes in the direction of molecular devices, integrated circuits... and the fabs that make them will be completely different animals. Since I'm unaware of there being a great deal of concensus on where the industry is going to after the 11 nm node, it sounds like a risky venture to go put tax incentives in place to build fabs that may not be compatible with the direction the industry goes after 11 nm.
Regardless, it seems as though there is a great deal of uncertainty late in the decade we're about to enter that I would like to see congeal with industry consensus behind it before I would advocate any specific policy regarding future fabs in the US.
Mark makes a good point but Michael hits the nail on the head in his comment. It's not all about engineers. When fabs and factories leave the US the skilled labor supporting them, the companies providing services to them, the communities depending on revenue from them, etc. etc. all go away. Shipbuilding is a good example. Or steel. Tax policy is the way you influence this problem of course. Create favorable business conditions and the factories and fabs will locate there. (NY may have it right.) But the U.S. as a whole has this irrational fear of using tax policy to create jobs. "We can't be in the business of picking winners and losers" is the oft heard refrain. Bull.
US workers deserve a government that actively supports reasonable, well paying and plentiful jobs or we risk becoming a nation of the un- and underemployed. (Maybe we already are?) I don't know about you but I know many talented, skilled and educated folks - many formerly associated with the high technology sector - who are pounding the pavement and/or "consulting now. A sustained and enlightened industrial policy can fix this. If you need proof that industrial policy works witness what Japan, Korea, Taiwan and China have done for their economies in the last 40 years. Compare this remarkable progress to the "gains" of Americas working middle class over the same period. It's time to bring some jobs home however we choose to do it.
A Book For All Reasons Bernard Cole1 Comment Robert Oshana's recent book "Software Engineering for Embedded Systems (Newnes/Elsevier)," written and edited with Mark Kraeling, is a 'book for all reasons.' At almost 1,200 pages, it ...