Source: The New York Times
Andy Grove, the renowned chief executive of Intel, who died in March, coined a phrase beloved in Silicon Valley: “Only the paranoid survive.”
That sounds cool, if you like your capitalism fierce. That idea, however, turns out to have some significant downsides.
Intel is the world’s biggest semiconductor company because when Mr. Grove was in charge, it dominated the personal computer industry and was an important player in the associated business in computer servers.
Today, the PC market is shrinking, hurting Intel’s profits. The server-chip industry is still strong, thanks to the rise of cloud computing at companies like Facebook and Google. Cloud companies engineer server chips in ways that make very powerful and flexible systems used by millions of people.
But Intel missed joining a number of other markets that did not look like the PC business, particularly smartphones. It is scrambling for a place in sensors (or what is called the internet of things), wireless networking, autonomous vehicles and other hot areas, as computing spreads from traditional computers to nearly every machine.
“The mistake was, when you grow with great leadership and great success, your paranoia is focused on the one thing that is delivering that,” Brian Krzanich, Intel’s current chief executive, said in an interview.
From PC-dominated computing, he said, we are now “where computing is constantly shifting.” Mr. Krzanich says he is still a strong believer in paranoia, but he must turn that vigilant anxiety in a lot of directions at once. Cloud computing may be Intel’s core, but Mr. Krzanich has to feed it from chips in many markets.
The effect of those changes will be seen this week as Intel holds its Developer Forum in San Francisco. The conference is where the hard-core geeks of computing gather to hear about what’s coming in the realms of processing power, memory and other technology that enables the modern world.
Giant tech trade shows like the International CES, held in Las Vegas every January, are focused on hot products. Developer events may be nerd-fests, but they say more about the technology world, and where a company hopes to be in two or three years.
At the Developer Forum, Mr. Krzanich will deliver a keynote speech, in which he is expected to dwell on topics like virtual reality, sensors and artificial intelligence, which was in the news last week when Intel bought an A.I. start-up, Nervana Systems, that has technology that will be used in Intel’s data center chips.
Shifting gears for the giant chip company is no easy task. “Intel has four main business lines — PCs, mobile, internet of things and servers — and it’s hitting on one of them,” said Patrick Gelsinger, a longtime Intel executive who is now chief executive of the software company VMware. The factories where Intel builds chips “now cost $10 billion, at least,” each to set up, he said, “and the competition is catching up.”
In April, Intel announced it was cutting its work force by 12,000 people, about 11 percent of workers. About a quarter of the executives who report to Mr. Krzanich come from other companies, a big deal at a place where leaders rise through the ranks over decades.
In his speech, Mr. Krzanich may stay away from talking about wearable devices. Last month, Intel withdrew a health-monitoring watch that had some nice capabilities but sometimes burned its wearers’ skin. Even so, Mr. Krzanich said, the heart-monitoring capabilities of that watch will find their way into a device Intel plans to make with the athletics brand New Balance.
The part he is looking forward to, Mr. Krzanich said, is a one-hour question-and-answer session with 200 elite developers. “It’s a much broader group” than even a couple of years ago, he said. “People writing for machine learning and A.I., networking chips, how virtual reality feeds into PCs.” Even in decline, PC chips still make up half of Intel’s revenue.
If these things catch on, Intel will sell more chips to those businesses and create new demand for cloud computing, because most devices are now connected to the cloud.
Even so, another pillar of the Intel mythology, the unchanging nature of Moore’s Law, is getting revised. Named for the Intel co-founder Gordon Moore, it states that a chip’s power doubles every 18 to 24 months without a rise in costs. Mr. Grove’s paranoia championed and fed on Moore’s Law, as Intel strove to achieve profit from that computing cadence.
Silicon Valley treats Moore’s Law as if it is immutable, and with even more reverence than it does paranoia. But it was not a scientific law; it was always an observation about the behavior of a market for computers and software, which paid off at a rate to justify increasing investment in making chips.
It is changing, Mr. Krzanich said, because phones, sensors and cloud systems develop at different rates. “It’s lengthened to 24 to 36 months,” he said. “The performance of the ecosystem is much more than Moore’s Law.”
That is why Intel is in the wireless and networking fields, and is working on a new kind of three-dimensional memory chip, which Mr. Krzanich said would be out at the end of this year, that can speed performance of big-data-type calculations sevenfold.
There would be a historical precedent to that shift. Years ago, Intel’s main business was in memory chips, a market that collapsed three decades ago, leading the company to move toward making number-crunching processors for PCs. That shift, which arguably saved the company, was the lesson that led to Mr. Grove’s paranoia.