Logo   | Search
About
Project Feasibility Study
Hardware Design &
Fabrication
Program Development & Debugging
Product Characterization & Correlation


 
   
 
 
Enter
 
 
Intel's architect of ever smaller things
(Financial Times - Aug 6, 2007)

Paul Otellini's challenge is to make sure the world's biggest chipmaker gets smaller. Smaller circuitry, smaller power demands, smaller chips for smaller devices - the process is never ending.

The pace was set more than 40 years ago by Gordon Moore, one of Mr Otellini's predecessors as chief executive of Intel. Moore's Law, as it became known (from an article he wrote in 1965), states that the number of transistors on a chip should double every 18 months to two years.

In trying to maintain this pace in the current century, Intel lost its way. It failed to update the design of microprocessors and focused on speed. The crisis came to a head in 2006, one year into Mr Otellini's tenure as the fifth chief executive since 1968, the year the company was founded. "I prefer not to dwell on it, because we're over it, but we'd let our micro architecture get too old," he says.

In previous generations Intel would update its designs over two or three years, but the Netburst architecture "was six years old when we finally replaced it". The architecture also generated too much heat and hogged power.

Intel had decided that the clock speed of its processors should be the standard for performance improvements. This allowed its smaller rival, Advanced Micro Devices, to gain market share by developing a more advanced architecture and features.

By the time Intel finally introduced its Core micro architecture a year ago, earnings were suffering. Three years of double-digit growth in revenues and profits were followed by a 9 per cent decline in sales and a 53 per cent fall inprofits in 2006.

Mr Otellini carried out the biggest review of the company in more than 20 years. He sold off non-core businesses, eliminated more than 1,000 management posts and cut a global workforce of more than 102,500 by 12 per cent.

It was perhaps the worst year for Intel since 1974, when Mr Otellini, now 56, joined the company. "I think Intel had its first significant lay-offs in July of 1974, which was the month I started," he says.

That year the US was in recession. "I liked the fact that during a tough time for the industry they respected their need to hire from college, which has really set a pattern to this date."

Compared with his gruff and outgoing predecessor Craig Barrett, Mr Otellini seems shy, sensitive and thoughtful. He is slightly nervous when speaking in public and reveals few personal details. Alifelong San Franciscan, he is a graduate of the city's university, a member of the board of its symphony orchestra and a keen sailor.

He is the first Intel chief executive not to be trained as anengineer, but has always been intrigued by technology. "It was also nice that the technologycompanies that were on campus were within 50 miles of San Francisco, so I gravitated to those."

He had a choice between jobs at Intel, Fairchild Semiconductor and AMD, which was to emerge as Intel's big rival. He chose Intel because he liked the people. "Everyone I met was just bright and energetic and I liked the prospects for the company."

Intel grew from 3,000 people when he joined to 100,000-plus last year. Its share of the global market for the microprocessors that power computers is 80 per cent.

Mr Otellini is speaking at Intel's Santa Clara headquarters in Silicon Valley. As in all its offices, a cubicle culture prevails (see below). Viewed from above, the corridors and partitions resemble the circuit paths etched on its chips, data routes that, by the end of this year, will have been narrowed to widths of just 45 nano-metres, or billionths of a metre.

"If we do the math, almost 90 per cent of the revenue that we get in December of any year is from products that we weren't shipping in January," Mr Otellini says.

Intel has restored its metronome-like pace of development, the "tick-tock model", he calls it, "the analogy being a clock or pendulum".This year Intel is swinging to 45-nanometre chips. In 2008 it will introduce a new microprocessor architecture, followed in 2009 by the next level of miniaturisation - 32 nanometres - and another new design in 2010.

He first set the company back on this path when, as head of the microprocessor business, he turned to Intel's Israeli outpost for the development of what became the Centrino platform for notebooks.

"There is some element of success breeding complacency," he says of the Netburst era. "The need to change, and the creativity, usually comes from outside."

Centrino balanced power-saving against performance and led to the development of an architecture based on "parallelism" - two or more cores, or individual processing units, working together on the same chip. More tasks could be carried out at lower clock speeds and temperatures.

Chip architecture is again mirrored in the company itself. Parallel teams work on each new generation of products. These leapfrog one another, as one team creates one generation, hands over to the next and then picks up the development cycle again. "I'll be the first one to admit that there is some internal competition there, but that's very healthy," he says.

The chief executive has also got rid of the Intel tradition of "two in a box" - two executives helping each other to manage a department. "In some cases, we had two in a box managing one or two people, which is just ludicrous."

Taking out layers of management has reduced costs, which Mr Otellini sees as essential in a world in which smaller chips on smaller devices means smaller margins. Chips for consumer applications are sold at lower cost, particularly in emerging markets.

He sees growth in a new class of devices - ultra-mobile PCs, mobile internet devices and consumer electronics gadgets that will usea future chip codenamed Silver-thorne. It offers ultra-low power from an ultra-small chip but is still internet-enabled.

"We're not abandoning computing. The personal computer is going to be around forever, in my view. But I think that these new classes of devices are going to be additive," he says.

"These kinds of chips can address the needs for ultra-low-cost computing in emerging markets, new devices for living rooms. So this expands the potential market for Intel dramatically."

Life outside the box: squaring up to the cubicle culture

Paul Otellini insists that plans were under way to modernise Intel's office space well before the company's cubicle culture was satirised on US television.

Conan O'Brien filmed a visit to Intel this year for his late-night talk show. He mockingly praised Intel for creating a working environment where people could feel there was no individuality, no hope and no sense that life had possibilities.

He admired the imaginative colour scheme of grey, grey and greyish blue and likened signs on columns such as B9 and H10, defining sections of a maze of partitions, to a carpark.

All Intel employees, even the chief executive, work in cubicles. Mr Otellini has a corner cubicle next to the windows. He does not like the fact that engineers two cubicles apart send an e-mail rather than get up and talk.

"The whole nature of sitting down and hashing out ideas and collaborating is a bit stymied by the construct of the cubicles," he says. He is promising more common spaces and more colour, though not like that in the 1970s.

"[The cubicles] were ugly orange. I remember, when they first put them in, they looked cool. But by the time you got to 1985, they looked pretty bad. And at the time, grey was très cool. I mean, how neutral can you get?"

Now, he says, it's time to move on. "Modern office furniture is much more configurable, so you can do things that meet an individual department's need."



 
Copyright © 2012 ChipTest, All Rights Reserved | Disclaimer
Designed & Developed by Cherry
Home