Visit Your Local PBS Station PBS Home PBS Home Programs A-Z TV Schedules Watch Video Donate Shop PBS Search PBS
I, Cringely - The Survival of the Nerdiest with Robert X. Cringely
Search I,Cringely:

The Pulpit
The Pulpit

<< [ It Beats Flipping Burgers ]   |  Alienation of Inflection  |   [ Unprotected Hex ] >>

Weekly Column

Alienation of Inflection: How Intel is Reinventing Itself the Wrong Way

Status: [CLOSED]
By Robert X. Cringely

This week, PBS held its annual meeting just up the rail line from me in San Francisco. It was my chance not only to remember that I pretend to work in television, but to work for the first time on-stage with Andy Grove of Intel. For those who don't recognize the name, Andy is the chairman of Intel and its employee number one, a point made very clear to me years ago by Intel founder Gordon Moore. "Bob Noyce and I are the founders of Intel," said Gordon. "Andy is employee number one."

Well, Andy is still very much on the job and shared with the PBS audience not only word of a major Intel investment in enhanced digital TV (particularly PBS enhanced digital TV), but he also shared his big view of the digital future for both Intel and PBS.

Alas, I think Andy got the Intel part wrong.

He did a very good job of setting up for the audience the idea of product cycles, and how to tell when it is time to abandon the old and embrace the new. Intel has done this twice. As you may remember, Intel was originally started to build memory chips, something it pretty much abandoned to the Japanese more than a decade ago. In that instance, the Japanese just plain cleaned Intel's clock, applying new technologies and new ways of doing business so furiously that the memory market was lost almost before Intel knew it was being contested. So Intel jumped to its second business, microprocessors.

In an instance where the external threat may not be so obvious and ferocious as the Japanese DRAM attack of the 1980s, companies typically reach what Grove called an "inflection point." This is a moment where sales and profits have been rising and one might guess they'll rise some more, but they won't. At the inflection point, sales of a particular product or service are at their peak. The managers of the business can either decide at that moment to expand in a new direction, generating not only new sales but new types of sales, or they can let the existing business decline and die, often taking the company along.

Certainly, there is a trick to knowing exactly when the inflection point is reached. Jumping to the new technology too early is risky because the new market may be slow in maturing, or maybe the new technology isn't really the "next big thing" at all. This was certainly the case in the late 1970s with bubble memory, a nonvolatile storage medium that got a lot of press and almost no sales. Had Intel bet the ranch on bubble memory, there might be no Intel today.

There is a temptation, too, to delay the jump even when it is quite obvious what's the next big thing, because jumps cost money and even declining markets can be very profitable if you have the combination of economies of scale and no need for further capital investment. An executive can look really good by milking old technology for all it is worth. That's exactly what John Sculley did at Apple in the late 1980s. Sculley was acclaimed as a genius at the time, then fired for cause when his unwillingness to really embrace change sent Apple into the toilet.

That's a technical business term — "into the toilet."

Andy Grove led Intel's jump from memory chips to microprocessors around 1986. This involved literally abandoning memory altogether and working both to dominate the microprocessor business and to turn it into a brand through, among many thing, the "Intel Inside" campaign. It is easy to forget that until the Pentium came along, Intel licensed other manufacturers to make official clones of all its processors. This was certainly true for several chipmakers up through the 80386, and IBM had the right to build i486s, too. But once Intel committed to processors as its core business, it became all business, with the result that the company underwent double-digit growth and eventually got in anti-trust trouble with the government.

Then last year, according to Andy, Intel hit what he calculated to be another inflection point. By 1998, there was much less certainty that the world would stay PC-centric, and the source of this uncertainty was the Internet.

So Intel reinvented itself again, according to Andy. They certainly didn't abandon the microprocessor business, but this time added growth in a new direction — Internet commerce. Somebody ran the numbers and found that 42 percent of Intel's business was in some way related to electronic commerce, yet the company was making little direct investment in that area. No more. Intel is now pumping billions into new Internet ventures, almost none of them having to do with selling more microprocessors.

So far this is all very good. I, a guy who calls technology trends really well but can't seem to make any money at it, agree completely with Andy the billionaire. Then he started talking about "bit factories," and I had to disagree.

In his speech to PBS, Andy Grove identified building bit factories as Intel's core strategy for the time past the Internet inflection point. A bit factory, according to Intel, is a room or a building connected to the Internet through very fast redundant Internet connections. This room is filled with thousands of servers and those servers host Web sites, databases, support FTP software downloads — whatever the virtual Internet companies of the future will require. Intel likes bit factories because, as a builder of microprocessors, motherboards, and entire PCs, they can build bit factories cheaper than anyone. Bit factories use lots of Intel components. Bit factories encourage businesses of all types to get on the Internet. And to Intel, it looks like a great business. They can sell products and services over the wire, eliminate dependency on middlemen, and get closer to the customer's wallet.

So what's wrong with bit factories? They go against the single major force behind Intel's success — the broad distribution of microprocessors. Intel shouldn't be building bit factories. It should be building bit cottages, thousands of them.

Intel's success came at the expense of mainframe computers, yet a bit factory is just a virtual mainframe. Internet users are widely distributed, yet the concentration of server capacity in a few places not only doesn't match Internet demographics, it hurts Internet performance. Just take a look at's recent trouble carrying a Victoria's Secret fashion show to prove that.

In the 1970s oil crisis, the U.S. government began funding research in alternative, renewable energy sources and the U.S. electric power industry responded by proposing a variety of programs for massive, centralized solar and wind power facilities. Saying that they'd gather solar power in a few hundred acres of desert then ship it hundreds of miles to my house was ludicrous, but it maintained the political and economic power structure of big business. It would have made far more sense to put a solar panel and a windmill on every roof with lower resultant capital costs and transmission losses. But since putting a panel and a windmill on every roof would have also pretty much eliminated the need for a power company, the research logically went in a stupid direction.

Now we have Intel, which has made billions putting computers on 200 million desks, suggesting that the way it will make money in future is by sucking power back from those desks and into bit factories. This is a bad idea. It would make much more sense to leverage the power of the microprocessor by building a network of servers that's as close to the end-users as possible. It's harder to do, sure, but that just means Intel sells more processors to itself.

The bit factory idea might work for awhile, but the inflection point is looming, Andy. Don't forget, you heard it here first.

Comments from the Tribe

Status: [CLOSED] read all comments (0)