Yesterday, when I deduced that Intel (Nasdaq: INTC) might end up fab-less, I must have hit a nerve. One of Intel's PR guys emailed me a terse "Not even close"; another Tweeted the same basic message to a few hundred of his closest friends; and then the phone rang to really set the record straight.
Why Intel loves manufacturing
Intel spokesman Chuck Mulloy cordially pointed out that his company is investing billions of dollars into manufacturing at the moment. Worldwide, there are 15 Intel chip-making plants, or "fabs." The company is spending $7 billion to establish the 32-nanometer production technology at fabs in Oregon, Arizona, and New Mexico – Intel’s largest capital investment ever in a single manufacturing technology.
And don't forget, dear Fool, that this relentless process improvement is one of Intel's biggest advantages versus competitor Advanced Micro Devices (NYSE: AMD). Intel tends to have the more advanced manufacturing technology at any given time. Hence, Intel's chips run cooler and faster than AMD's and are also cheaper to make, until AMD's manufacturing tech catches up. And by then, Intel is generally ready to take the next step. Intel is already planning and researching the 22 nm and 16 nm process nodes, future-proofing the company for years to come.
With all of these very material factors in mind, Intel would be crazy to spin off its manufacturing operations today, Chuck said. And I wholeheartedly agree.
Perhaps my previous article wasn't entirely clear on the time frame I was talking about. Intel clearly won't go fab-less this year or next, or even by 2012. But at some point, the manufacturing process will cease to be a material advantage. You can't make electrical leads any thinner than a single atom across, for example. And for every new generation, the upgrades become more expensive. According to market research company iSuppli, that 16 nm process could be the last upgrade we'll ever see under current manufacturing processes, simply because the manufacturing equipment beyond that point would be so expensive that "the value of their lifetime productivity can never justify it." We should hit that limit by 2015 or so.
So if Intel ever splits in two, it will be with that ultimate production technology in hand and fully matured. If your investment horizon is shorter than, say, 10 years, it won't matter at all. And like I said, Intel may simply be setting itself up to "pump unheard-of resources into Bryant's [manufacturing] division" over the next few years. That would be Intel's official position on the matter, anyhow.
The chip giant doth protest too much
Just don't be surprised if Intel changes its tune in a few years when the economic impossibility of pushing any further becomes all too real. And don't forget that there are precedents to the divide-and-conquer strategy way older than AMD's recent moves. For example, Tyco (NYSE: TYC) has partitioned into appropriate bite sizes years before splitting up into Tyco, Covidien (NYSE: COV), and Tyco Electronics (NYSE: TEL). And splitting up doesn't necessarily mean goodbye: EMC (NYSE: EMC) may have allowed VMware (NYSE: VMW) to leave the nest, but the company still owns 84% of its former virtualization division. A bifurcated Intel could still keep manufacturing on a tight leash, in other words.
© 2009 UCLICK, L.L.C