Blog Post

Intel’s Multi-Billion-Dollar Cost of Doing Business

Stay on Top of Enterprise Technology Trends

Get updates impacting your industry from our GigaOm Research Community
Join the Community!
Intel's planned development fab.

Intel’s (s intc) patriotic investment in American manufacturing news blast worked so well last year that the company apparently decided to make a fairly big to-do today over its planned capital expenditures again, showcasing its $6 billion to $8 billion in planned investments in manufacturing for the next 18 months or so. The news is great and includes a new development factory in Oregon, but it’s also not really game-changing news. Intel regularly makes investments in its fabs, and because fabs are such hugely complicated manufacturing plants churning out billions of expensive and precision-made chips, those capital expenditures aren’t cheap.

I’ve discussed how it takes a lot of money to keep making chips smaller and more powerful, with most fabs costing about $5-6 billion to build. Once built, some of the equipment inside is revamped every 18 months or so as part of making chips ever smaller. So sure, let’s cheer that Intel still has fabs in the U.S. and is keeping them moving down the process node cramming more and more transistors on a chip (Moore and Moore perhaps?), but let’s not kid ourselves that this is some big hoopla-worthy event. For Intel, it’s the cost of doing business.

Frankly, given the number of chips it needs to sell in order to support all those wafers churned out at new manufacturing nodes, and its competition against ARM (s armh), Intel’s investments in its fabs may end up hurting it. Here’s Intel CEO Paul Otellini talking to Fox Business News on the announcement:

Related GigaOm Pro research (sub req’d):

5 Responses to “Intel’s Multi-Billion-Dollar Cost of Doing Business”

  1. The author seems to hint that Intel is over-investing in own fabs and that they probably will not recover their investments because of competition from ARM.

    Fat chance.

    Firstly, there are no free fabs that Intel could rent out. So, someone has to do that investment. Is it possible for other companies to build fabs for Intel’s next generation chips – yes, but only if Intel invests in making that fab – with it’s people, technology(IP) and most likely, money. All this will take time – exactly what ARM would be hoping for. Also, a farming out a fab will allow competition to benefit from process IPs, and often these give semiconductor companies a head-start.

    Intel could themselves set up (exclusive) fabs in other parts of the world. If they see any business benefits in doing so, I’m sure they would start somewhere, sometime. Of course a dramatic move to a new fab could be disruptive, but a phased move to a new fab location is very plausible. This does require long-term planning and commitment, an eel in today’s corporate world.

  2. The article fails to mention is the financial burden that local and state governments share with Intel through bond issuance and tax incentives, not to mention local environmental impacts; it takes a lot of water to make all those chips, much of which comes from the local water table.

    Disclaimer: I live within sight of a major Intel complex with multiple fabs.

    While U.S. jobs and manufacturing are big pluses, increasingly host communities like mine are questioning the cost, often billions of dollars in forgone taxes and interest payments spread over multiple years, require to attract and keep such development.

  3. anonymous

    Give Intel some credit for keeping these investments in the USA unlike most companies nowadays. They could be making those investments elsewhere. “Cost of doing business” or not, $6-$8B is a lot of money and a lot of US jobs created.