Intel increasingly letting customers lead the way

When you're the world's largest chipmaker, it's hard to turn on a dime. It can be even harder to admit when you've overreached.

A shift has taken place at Intel over the last year or so. Once known for dictating the direction of the PC market, Intel is increasingly letting its customers carve their own path. With that subtle yet important change, the PC industry is moving past its Model T era and entering a new world of style and design, where a simple black or gray box won't do.

The most recent and telling example of this shift is the special version of the Core 2 Duo that Intel built at the request of Apple, one of its smallest yet most influential customers. Intel accelerated the development of a much smaller chip packaging technology and lowered the chip's power consumption, just so Apple could build the MacBook Air.

Intel has done exclusive chips for special customers in the past, but since Sean Maloney took the helm of Intel's sales and marketing in 2006, the chipmaker is much more willing to find a way to meet the needs and ideas of its PC customers--and to recognize that father doesn't always know best.

These days, Intel is cleverly taking a page from Advanced Micro Devices, the smaller chipmaker that has managed to work the phrase "customer-centric" into just about every press release and public appearance since 2003.

Intel salespeople are now encouraged to spend more time out of the office, talking to PC customers and designers and taking their ideas and concerns back to the mothership. Those customers have rapidly matured; instead of one basic desktop and laptop design for all, they are starting to realize that different people want different things.

They are customizing products for various segments and even different geographies, and introducing new designs like Gateway's all-in-one or Apple's MacBook Air. That reality makes it harder to dictate a top-down vision for the PC market, since there are now so many different types of customer needs to try to satisfy.

Intel's product-planning priorities have changed as a result. The Nehalem generation of processors, due in the second half of this year, will be one of Intel's most complicated launches ever because of the huge variety among different chips.

Some will have integrated memory controllers. Some will have point-to-point interconnects. You'll see dual-core, quad-core, and perhaps more in the server market. Some will be hot-and-heavy powerhouses, while others will be cool and nimble notebook chips. In short, Intel is going to have perhaps its widest variety ever of so-called SKUs (stock-keeping units) to offer its customers, allowing them to choose among several chip versions to find the one that best fits their goals.

Just five years ago, the industry worked in a different way. There was little differentiation among PC companies like Dell, Hewlett-Packard, and Gateway; everybody was cranking out bulky black desktops that sounded like jetliners or bulky notebooks with desktop processors that got two hours of battery life playing Minesweeper.

The market became all about price competition, and Dell rose to the top on the strength of its world-class manufacturing and assembly operation in central Texas. In their haste to keep up with Dell, everyone in the PC industry hustled to become lean and mean, and basic research and development into design techniques became a near afterthought.

That is, until 2005, when Apple announced its decision to switch to Intel's processors. Suddenly, the rest of the PC industry was on notice: they could no longer dismiss Apple as a sideshow act. Now their products were going to be held up against Apple's to an even greater degree, and that scared the hell out of several PC companies that had little to no experience in cutting-edge design.

Perhaps in return for enjoying a profit margin four times greater than that of its customers, Intel had been the one pushing the leading edge of PC design prior to 2005. It called for standards that would help reduce the costs of building PCs. It cajoled PC makers into adopting more interesting designs with its own set of concept products. And it recognized the growing importance of mobile computing with Centrino, a one-stop shopping experience for PC companies looking to build smaller, thinner notebooks.

That success emboldened Intel. PC companies, trying to recover from the collapse in business spending following the dot-com crash, were eager to cut costs and let Intel and Microsoft spend the money researching new ways to use PCs to grow the size of the market. Intel felt that with its reach, it had a better understanding of the PC market than any one vendor, and the smarts needed to take the industry where it needed to go.

Even if they don't want something like the MacBook Air, people now want stylish PCs, not cookie-cutter boxes.

(Credit: Corrine Schulze/CNET Networks)

Unfortunately, that only went so far. The platform strategy introduced by Intel CEO Paul Otellini, where Intel provides a pre-ordained set of components to PC makers, eventually showed its limits when it comes to predicting consumer tastes and styles in a changing market.

It wasn't that the strategy itself was flawed: PC makers definitely wanted the combo meal deal, where they could get the processor, chipset, and networking components fully assembled and tested. Even AMD, which for years criticized the idea of restricting "choice," jumped on board with the platform strategy after acquiring ATI Technologies.

It's just that at some point, Intel's ideas stopped resonating with its PC customers, as well as their customers. The Viiv digital home strategy has been a notable failure, as the general public has shown little interest thus far in putting a Windows PC at the center of their entertainment lives. At one point, Dell even forgot that it was supposed to be promoting the brand concept.

Earlier on, AMD became a serious player in this business because it listened to server customers who wanted no part of Intel's Itanium route to 64-bit processing. It came out with Opteron, a much cheaper yet still-powerful chip that preserved backward compatibility with software written for the x86 instruction set, and the rest is history.

Intel has changed a lot in the past decade. The company was left shaken and humbled by its clear misread of the market in the middle of the decade, and has vowed not to let it happen again. The trick now will be avoiding the same trap AMD fell into in 2006 and 2007.

Barcelona, AMD's first quad-core server processor, was the product of repeated customer requests for an integrated quad-core design, according to AMD executives. Faithful to its customers, AMD set off on building that chip, only to run into problems of nightmare proportions as it realized how "complicated" (in the words of CEO Hector Ruiz) that design would be to complete. Barcelona will ship at least a year later than expected as a result, and Intel cornered the quad-core server market by taking a less elegant but easier and quicker route to market.

Companies have to make leaps of faith from time to time. Just look at Apple; despite all the risk involved in switching to Intel's chips, it needed a lower-power chip if it wanted to stay relevant in a computer market where people were demanding laptops. Two years later, that has worked out pretty well.

But sometimes father does know best. Every now and then, you have to tell a customer who comes to you with a request that "no, it can't be done. And here's why." The bet-the-farm strategy can wind up leaving both you and the customer in the lurch if something goes awry.

Can Intel walk the fine line between management consultant and flexible supplier? It will be quite the balancing act for one of the world's largest tech companies, and one that historically at least, is used to pushing rather than pulling.

[System48w]

Apple told Intel what to do?

Hardly, the chip package had been shown at the fall IDF and the chip itself already existed. The plan though was to wait until the 45nm shrink to throw it on the new small package, since Penryn would better fit, lower power consumption, the ultra portable devices the package would allow for. Apple asked, can we get this now and Intel said for the right price. The price, is evident in the price of the MBA. Intel's current success with the Core arch was done by simply designing a really great laptop processor that could achieve top performance and low power consumption and then throwing into a desktop where that's not too much of a concern. Intel's future products were not somehow done at the behest of specific customers but rather by what Intel see's(dictating) the future looking like.

[BigGuns149]

Apple is one the smallest customers? OEMs control Intel's decisions?

While Apple isn't the largest buyer of Intel chips other from HP and Dell who buys more Intel products than Apple? The last I checked Apple was third in market share. Since Apple not only uses their CPUs, but also their boards as well Apple is probably a pretty good size account for Intel. A lot of other vendors frequently use other vendors for the motherboards and in some cases don't even use an Intel chipset.

The real reason Intel largely doesn't make their decisions upon what OEMs request is because they don't make up a significant enough market share. Nobody has a majority of even the 1st tier(top 5) OEM market. Once you throw the 2nd tier and the VARs and channel partners into the mix and you can see that even HP and Dell Intel's top customers individually don't wield that much influence over the market. Intel will make sure that they get as many chips as they want and try to get them in their hands quickly so that they can ship out new models to customers and resellers quickly, but that is about as much additional clout as Intel is willing to give. Intel has been so successful because they have been able to keep enough OEMs Intel exclusive as possible. If you seem too chummy with one OEM you might alienate OEMs into AMDs camp.

As a previous poster noted Intel didn't really bend over backwards for Apple because they already had the socket design already. As indicated by perspective talks with Fujitsu and Lenovo into the same chip it appears Apple wasn't the only company that Intel saw as a potential customer for this chip. Apple was simply the first to ship a product with the new CPU. As the largest Intel exclusive vendor Apple tends to sometimes get things first or at least ahead of other non-exclusive Intel vendors, but I don't think Apple or anyone else has as much influence over Intel's decisions as the author of this story thinks.

As I've been told by several Intel employees most of their decisions are more based upon demands of the software industry, Microsoft specifically. That is why Intel's consumer boards, which is largely going to be running 32bit versions of Windows don't use EFI yet. When M$ either adds EFI booting support to 32bit or they put a more firm timetable for completing the the transition towards 64bit Windows for general consumers Intel will move to EFI, since 64bit Vista versions support EFI 2.0, Intel will release consumer boards with EFI on them. Presumably Windows 7, whenever that comes out may go 64bit exclusive because by there will be few people still running older 32bit CPUs any longer, but we won't know until we get much closer to Windows 7.