How do you handle a disruptive technology when you’re the main target?
Intel announced that there will be nothing branded “Centrino Atoms” anymore. Centrino is Centrino and Atom is Atom and never the twain shall meet.
While we don’t pay serious attention to marketing names, leaving that for the shallow, this change actually sheds some light on what Intel is trying to do with its product line. In recent articles, we’ve pointed out bits and pieces of this strategy, now it’s time to wrap it all up in a single package.
Intel believes the CPU world is going to change dramatically over the next five-to-ten years, but isn’t too sure how and to what degree. It hopes the current CPU market (albeit one more notebookish) will stay more or less as is, with Atom-like processors being a whole new market. It fears Atom-like processors will end up gnawing everything else into niches. It expects something of a muddle between the two, but they don’t know how much of what will be in the mess.
So it’s going to put its fingers in every pie, see how things turn out, and be ready to stay flexible when it comes to product mix. They’ll continue the “normal” product development, and they’ll also have Atom and Larrabee initiatives. No doubt the three will crosspollinate and enrich each other over the course of time. Maybe they’ll all end up merging into the totally scalable CPU swarm we’ve talked about in the past, maybe they won’t. There will be a bias towards the heavy-duty stuff and against the tinyPCs, but in the end, the market will decide the mix.
If killer apps demanding killer power that even Grandma wants emerge, Intel will be happy to adjust their product mix and pricing to accommodate. However, since this is very unlikely, what we’re far more likely to see is a bifurcation between relatively few power users, some more semipower users, and many more casual users. Both groups will be serviced, but the latter will swap computing power for mobility, simplicity and a low price.
What is that going to mean in simple terms? In the long run, Grandma (and by that I mean most people) isn’t going to have a lite version of your PC anymore; she’s going to end up with something else. She’s going to pay less for it, and that means you’re going to pay more for yours, mostly because instead of Grandma helping to spread the cost of R&D and manufacturing for machines like yours with her purchase; it will be more like you’ll be paying for some of her R&D with your purchase, because that will be mostly your hand-me-down stuff.
What does it mean in the short-term? What it means is that Intel has been moving and is now completing a major shift in product mix to something much more complicated than the old Pentium/Celeron division. I see at least four classes of CPUs, but it’s really more a continuum of processor classes, with rather blurred edges, and what may start as the lower-end of one class may end up being the higher-end of the next lower class. But what we won’t see anymore is the top dog eventually becoming the low dog over the course of time.
What are these classes?
1) The professional elite class: These CPUs will be used in servers, in workstations, and by anybody else willing to spend a lot on a machine. In general, these machines will be bought by those who make serious money due to what they can do with computing power, so they’ll pay serious money for more performance. Issues like multicoring and irrelevant to them because they use software that has to take advantage of them.
Intel has always had Xeons around to handle part of this group; what we’ll see is an expansion of that class to include workstations and other high-enders. This class will get the cutting-edge technology first, and they’ll pay for it. Bloomfields are the first signs of this expansion. It’s not going to end up as a mainstream chip; it has its own elite little socket, and the rest will get something else. The only major group of CPUs that doesn’t have representatives in this group are notebook chips, but don’t worry, that will come, maybe with quadchips, maybe some time after that.
2) Cut-down cutting edge from last year/last generation: This could be called the new mainstream, but it’s probably better to call it the premium line. A year from now, the upper end of this class will be the somewhat cutdown versions of last year’s elite products (i.e. Lynnfields), the lower-end will come from the previous generation of products (the lower-end quads, upper-end dual Penryns). A year after that, a somewhat cut-down Sandy Bridge will take the upper-end, the Lynnfields will move down. Much the same will happen with notebook chips, except that there will be the usual lagtime of about nine months for those.
3) The New Celerons/OEM Specials: While I suspect the Celeron name will be retired fairly short, something else is bound to take its place. These are rather cut-down versions of mainstream chips, and in the past couple years we’ve seen essentially sub-Celeron chips (i.e. “Pentium Dual-Cores”) emerge. These are today’s true mainstream chips, these are the entry-level chips you get standard with your Dell or HP.
Up to this point, everything here is all very methodical and logical and gradually introduced, which is all very much Intel’s style. They are adaptations and modifications of the old Intel models, albeit one tweaked to maximizing profits. Yet there is something else which is revolutionary here:
4) Atom: This is the bottom of the barrel, and Intel’s great fear is that they’ll eventually have to make them good enough to become most of the barrel. That’s why Intel yanked that “Centrino Atom” label; they don’t want anybody convincing themselves out of buying a “real” notebook processor for less.
If Intel could get a wish granted, it would be to have all Atoms go into smartphones and never be seen or heard of anywhere else again, but the current Atoms can’t take on that role yet; they chew up too much power. So Intel put out a few just to get the ball rolling, and these Atoms are now proceding to cannibalize a good chunk of the hitherto very expensive ultraportable market. Sure, they can’t hold a candle to the more advanced models, but when it comes down to spending $500 rather than $2,500 for a two-pound PC, a lot of people don’t care, and even worse, people in the mainstream media won’t shut up about them. This leaves Intel very, very nervous, not so much for right now but a few years down the road.
As we said earlier, Intel wants Atom to be a supplement to their current sales, not a substitute for most of them. They’re introducing a potential disruptive technology, and they’re the most likely victim of it. Yes, the idea is to sell many, many times more cheap chips than they do expensive chips, but staking your future on selling billions of chips a year to markets that don’t exist yet is more than a little scary. Even more scary, if that’s what the future is, companies or groups of companies well-established in areas like cellphones might want to do their own CPUs, and prove to be a lot more formidable competition than AMD ever could be.
It’s a good time for Intel, but it’s also a very scary time, because they don’t know what’s going to happen next. They know their world is going to change, change a lot, change in ways they’ve never had to deal with before. So they hedge and cover their bets as best as they can, and try to make as much money in the current environment as they can because they might need it down the road.