Intel to speed development while cutting spending

IN ANOTHER LIFE, a certain young analyst used to spend hours and hours doing research on technology firms such as Intel, Cisco, and Nortel. Then the tech bubble burst, and he now spends his days going back to school and writing on those same firms for a certain website.

Things have changed for the certain young analyst, and things have certainly changed for those tech firms.

Take Intel, for example, also known as Chipzilla for all you new readers.

Capital spending as a whole for 2002 is expected to be approximately $4.7 billion, lower than a previous forecast of between $5.0 billion and $5.2 billion, and the original plan of around $7 billion even earlier.

This has had a huge impact on firms such as Applied Materials, which makes the equipment that makes all those fancy chips we all take for granted. Every time a firm changes a process (going to 130 nanometers, switching to 300mm wafers), good old AMAT makes a bundle. But when things are slow, they are slow.

The majority of spending reductions is being realised by cost savings within ongoing construction projects. In addition, the company is reducing its fourth-quarter equipment spending by re-using certain equipment from older process technology.

Consider that a brand new fabrication plant costs $2 billion USD just to set up, such as the recently announced Fab 11X in Rio Rancho, New Mexico. It will use 300 mm wafers and the 130 nm process, in a building 1 million square feet in size with 200,000 square feet of clean room space.

Please recall that we reported about how Intel was backing off from its heralded Colorado Springs purchase earlier this year.

Intel has been accused of failing to pay bills in the past before in order to ensure sufficient cash flow, as noted here, and here.

Chipzilla, which was faced with 35 liens worth nearly $60 million, has apparently settled the matter, in which construction companies felt aggrieved that their bills hadn't been paid.

Our glorious Mike Mageek has recently written about Intel's latest roadmaps. Intel let loose some interesting facts on October 22nd.

According to current plans, Intel will launch Pentium 4 Prescott processors in the second half of next year. They will feature Hyper-Threading II technology/marchitechture -- interesting that Hyper-Threading I hasn't even been officially launched yet. Perhaps that is the official name of the Prescott New Instructions, which may still be called SSE-3.

Of course it will be made with Intel's advanced 90 nanometer process, quite a feat considering that TSMC and UMC are still having trouble with the 130 nm process.

Someone also mentioned that the new Pentium 4 easily reaches 4 GHz, and may in fact scale past the 5 GHz mark easily.

Intel says it wants to speed up the transition to the 90 nanometer manufacturing process. By the end of the second quarter of next year, five per cent of all processors made by Intel are supposed to be on the 90 nanometer process.

By the end of the Q3 2003, that figure jumps up to 20 per cent and 40 per cent by the end of the year.

What is really amazing is that Intel wants to produce a full 90 per cent of its CPUs using 90 nanometer technology by the end of the first half of 2004.

After that, the craziness starts again by moving to the 64 nm process, which is supposed to use EUV technology that we aren't even sure will scale yet.

Of course, we can't forget wafer size as well. By the end of Q2 2003, 25 per cent of production will be on 300 mm wafers. An astounding 75 per cent CPUs produced in January of 2004 will be on 300mm wafers.

With revenue in the fourth quarter "only" coming in between $6.5 billion and $6.9 billion, it is easy to see why Intel is cutting costs. But one can't help but wonder how will they accelerate their development while cutting spending? One way to do it is to shut down a couple of fabs for a while, similar to what happened in Israel, and what will happen in Ireland (They all start with "I"?). And of course, retooling for new technologies also means shutting down a fab for a year.

Recall Craig Barrett's keynote speech at IDF Spring 2002, which was heartily noted by us. "We have to innovate our way out of this recession with new technology, new products. If there was ever a time where we needed to move faster down the technology curve, this is it."

Apparently Intel moves faster by cutting capital spending.