Intel's Barrett leaves chipmaking legacy, shortfall

Intel Chairman Craig Barrett

(Credit: Intel)

Chances are pretty good that once Intel Chairman Craig Barrett formally retires in May, he'll head straight for Montana just as the snows are melting and the rivers are high.

The avid fisherman, who served as Intel's fourth chief executive officer until three years ago, seems equally at ease discussing how to create some of the most complex structures known to man and the best way to tie a fly. The company announced Friday that Barrett will step down as chairman of the board of the world's largest chipmaker, three years after turning over the CEO post to current Intel head Paul Otellini.

He leaves a mixed legacy behind: under his watch, Intel cemented its reputation as the world's leading chip-manufacturing organization. It operates a network of sophisticated "fabs" around the world that are the envy of the industry and of which Barrett, who is credited for perfecting Intel's approach to chip manufacturing in the 1970s and '80s, is immensely proud.

But despite allocating billions of dollars toward R&D and acquisitions, Intel has failed to translate its dominant position atop the PC and server industries into a significant presence in any other market, such as consumer electronics or communications. That could give the company big problems as PCs and servers turn into slow-growing commodities, which is exactly the situation it sought to avoid in attempting to expand its reach.

Barrett leaves Intel at a time when the company is in as bad a shape as it has been in a long time, mostly thanks to the state of the worldwide economy. The last time things were this bad at Intel, however, was the time when Barrett made perhaps his most important contribution to the company as CEO.

During the dot-com bust years, Intel was very worried about its chip manufacturing plants. Such plants, known as fabs in the industry, are ridiculously expensive to build and maintain, and therefore have to be run at very high capacity to generate a return. With PC demand falling faster than the Nasdaq, there was considerable sentiment both outside and inside Intel that the company should think about pulling back on plans to invest billions in new fabs and new manufacturing technologies.

Barrett disagreed, and his was a very important vote. His decision to damn the torpedoes and plunge money in the manufacturing operation left Intel sitting pretty once PC demand recovered, with enough capacity and manufacturing expertise to ensure that PC makers would have to come to Intel for the majority of their chips.

While Intel embarked on that strategy, however, it took its eye off its core business in a very important way. A day of reckoning would soon be upon its Netburst architecture, the soul of the Pentium 4 and Xeon processors that made Intel the vast majority of its money.

Simply put, Netburst ran too hot. The architecture was designed to allow Intel to crank up the clock speed of its chips to new heights, which the company's marketing people believed was the only metric attached to processor performance that regular people understood.

But the faster you run a chip, the more heat it produces. Intel once spoke of taking the Pentium 4 architecture to 10GHz, but concerns about power consumption and heat dissipation ensured the chip would top out under 4GHz. Faced with this reality, to divide the workload across two cores that ran at slower speeds and consumed less power.

Barrett is credited with developing Intel's "copy-exactly" manufacturing strategy and investing billions in new plants.

(Credit: Intel)

Advanced Micro Devices, which had anticipated this problem earlier than Intel, was able to capitalize on that misstep with its Opteron processors, transforming the company from an enterprise afterthought to a serious supplier of silicon to the Fortune 500.

AMD's rise also forced Intel to dramatically scale back its hopes for its Itanium processor, which it once felt would bring the computing world to 64-bit processing but that now occupies a small niche in the high-end server market. Server customers had little interest in rewriting their software to run on Itanium, and preferred AMD's simpler approach that was compatible with existing software.

Intel recovered nicely with a series of products based on and AMD's amazing ability to snatch defeat from the jaws of victory. It was helped by another key decision started under Barrett's leadership.

Barrett moved Intel more into the business of chipset, graphics, and Wi-Fi chip production. PC companies loved the fact that they could now come to Intel for all the chips they needed to build their products, which also allowed them to reduce the amount of money they spent testing and verifying that various third-party components would come together. Centrino was the most visible part of this strategy, a brand that came to represent "wireless notebook" for a good portion of the world.

Still, Intel enters 2009 having gained little from its efforts to diversify its business.

Under Barrett, Intel has heavily promoted a wide-area wireless networking standard known as WiMax, thought of in its simplest terms as Wi-Fi on steroids. So far, however, the wireless industry has not been as receptive to the idea as Intel had once hoped. Sprint has dithered over its plans for a WiMax network amid troubles of its own, and Intel had to write down $950 million of its investment in Clearwire, a WiMax network provider.

Earlier, Intel made a series of acquisitions designed to catapult it into the communications market. It spent billions without generating much of a return, and scaling back its interest in the telecommunications industry with the added insult of hundreds of millions in impairment charges.

Barrett's greatest legacy, however, may eventually have little to do with technology.

He has spent his three years as Intel's chairman circling the globe evangelizing education and investment in scientific research. He has especially urged the U.S. government to improve educational standards in this country, worrying that a lack of interest in science and technology among young students could lead to a decline in U.S. competitiveness in the years to come.

Barrett's huge Montana ranch will probably be a fine place to retire and reflect on his 35-year career at one of the world's most important companies. Intel may be in a bust cycle in the famously boom-and-bust semiconductor industry, but a decade ago he correctly identified where Intel needed to go: now all the company has to do is make it happen.

[ferretboy88]

If Intel is in Bad shape then we are all doomed. How could they not be doing well.

[lmasanti]

Why is that so many companies appears to be a "one shot companies," in the sense that they "do hit the spot [almost only] once?"

On the other hand, Apple has done "several really big success": Apple II, Mac, iPod, iPhone...

And speaking of "persons," the one behind those aforementioned Apple's success also made NeXT and Pixar!

[ilya_cilk]

Multicore is here ? dual cores have been out for a couple years now, and quad-cores are fairly common. In fact, it?s hard to find a single-core desktop machine on any vendor site. This scenario is playing out in a similar manner in the data center, where increasingly ?fatter? nodes are featuring more and more cores. Indeed, just a couple months ago, Intel announced a 6-core Xeon processor.

Yet the vast majority of applications have not been multicore-enabled. Why? Well, first of all, parallel programming is hard ? in fact viewed by many as an unsolved problem. And second, here?s the rough calculus done today by software development teams: On a dual-core system (the typical configuration today), one can typically expect a two-fold improvement at best. This performance improvement is just not compelling enough to make the (potentially considerable) development investment. Once the 8- and 16-core systems become mainstream, however, a single-threaded application may be leaving an order of magnitude in performance and throughput on the table. That?s when multicore enablement moves from ?nice to have? to a ?must have.?

For more about the multicore challenge, check out this e-Book: http://www.cilk.com/multicore-e-book/