CNET también está disponible en español.

Ir a español

Don't show this again

The secret history of the sub-$1,000 computer

In the face of relentless competition from Intel, processor upstart Cyrix dressed up the pig and created the sub-$1,000 PC market in the process.

Once upon a time there were no iPods, iPhones, Xboxes, Blackberrys, or Tivos. Really, I'm not kidding. There were PCs, though. And they were really expensive. But we didn't have anything else to spend our money on, so that was OK. We paid $2,000 for our PCs and liked it.

Back in those days, there were three microprocessor companies--Intel, AMD, and a little Texas (it's an oxymoron, I know) company named Cyrix. If you don't recognize the name, that's because Intel had such a lock on PC makers back then that Cyrix's processors were sold primarily through the third-party reseller channel.

It's a popular misconception that Cyrix "cloned" Intel's processors. Cyrix's processors were actually all original designs. In fact, Cyrix's manufacturing partners--initially Texas Instruments, later IBM and ST Microelectronics--licensed Cyrix's designs for their own branded processors.

Cyrix had early success with math coprocessors and 486 chips, but when it came to the Pentium generation, things got a little tougher, to say the least. Still, for a brief time, the company's 6x86 (aka M1) was the highest-performance x86 processor on the planet.

Then everything fell apart.

Like a bad sequel, the 6x86MX (aka M2 or MII)--the 6x86 processor's successor--failed to live up to expectations. And Intel's design, manufacturing, marketing, and sales machine were relentless. By the end of 1996, revenues were down, losses were piling up, and the future looked bleak for Cyrix.

So, we (yes, I was there) did what all desperate marketers do. We dressed up the pig. We took an older processor architecture and positioned it for a new market--sub-$1,000 PCs. In those days, PCs sold for an average price of around $2,000. You could get low-cost machines, but they were performance- and feature-crippled and not particularly useful.

In February of 1997, we launched the MediaGX processor along with Compaq's $999 Presario--the world's first sub-$1K multimedia PC, complete with all the bells and whistles. We simultaneously positioned the M2 for $1,000-$1,500 PCs and we were off to the races.

What we did is called market segmentation. Intel followed with the Celeron brand, as did AMD, and lo and behold, PC prices plummeted.

The back-story:
Cyrix co-founder Tom Brightman alone had the vision for a $500 PC when everyone else was going in the opposite direction. In October of 1995, a story entitled "Cyrix faces skeptics: Technology plan for a $500 personal computer encounters technical doubts" appeared in the San Jose Mercury News.

I was sitting right next to Tom, in the press room at the Microprocessor Forum in San Jose, when Dean Takahashi interviewed him for that story. I admit to thinking Tom was nuts. But when we later ran into trouble with the M2 design, and I began thinking about PC price elasticity, a big light bulb went on. Sometimes I'm a little slow, but I get there eventually.

Brightman had a skunk works operation in Longmont, Colo. His team, under Forrest Norrod, developed the entire MediaGX platform, including the software and a demo PC that formed the basis for Compaq's $999 Presario. Sure, the MediaGX was slow and the M2 was a disaster, but Brightman's vision and the Longmont team's execution gave us some really nice clothes to dress the pig in.

If it were up to Wintel, you'd still be paying $2,000 and up for a PC. But according to Gartner, today's average selling price for all PCs--desktops and notebooks--is around $850. And you can get a Dell Inspiron for $349.

The point is this: competition doesn't just give buyers options. The cause and effect may not always be obvious, but competition also creates new price points and spurs everyone--including market leaders who would rather sit on their fat profit margins--to innovate and charge less for it.

Moreover, competition doesn't just happen. It takes vision, strategy, and execution by extraordinary individuals and teams. Sometimes a little desperation and creativity doesn't hurt, either.

The ending:
National Semiconductor acquired Cyrix in November of 1997 for around $550 million. About two years later Mark Bluhm--the M1's architect and project leader--founded Navarro Networks and took 15 of Cyrix's best engineers with him. Cisco later acquired Navarro for $85 million.

Meanwhile, National launched the Geode family of integrated system-on-a-chip processors based on the MediaGX architecture. In 1999, National sold what little remained of Cyrix's processor group--sans the Geode product line--to Via Technologies for $167 million. National later sold Geode, including the Longmont operation, to AMD.

It was a sad and drawn-out ending for a great company whose impact is still felt today. Still, it was fun while it lasted.