Failure Is an Option

Meanwhile, George Taylor and his engineering team completed the first "tapeout" of their microprocessor. (Tapeout is a critical milestone in the creation of a chip.) But there were sticking points. Hitachi refused to begin fabricating the Exponential chip without approval from IBM, which said it needed approval from Motorola -- and Motorola was developing a chip that Exponential's chip would be competing against. "We kept thinking, it shouldn't be this hard,'' Dorris recalls. Eventually Motorola gave its consent, and Exponential's blueprint was sent to Hitachi in January 1996.

Then the meteor shower began.

Amelio announced that Apple would lose $740 million for the second quarter of 1996. The company had been hemorrhaging, but this loss was staggering. By this time, Shriner's team in San Jose had grown to nearly 60 people, and Exponential's burn rate was approaching $1 million per month. Shriner and Dorris had to secure another round of funding, and they hoped that the venture-capital community would not look too unkindly at Exponential's alliance with Apple. Valdes, hoping to secure more orders, intensified her talks with Power Computing and Umax. "It was becoming clear to us that we would need an escape door from Apple,'' she says.

It was also becoming clear that Exponential would miss the target date for the rollout of its first chips: Hitachi was late in delivering parts. On July 17, Taylor and his engineers loaded the X704 microprocessors into a Macintosh machine for the first time. The processors ran at 410 MHz -- which was 40 MHz below what Exponential had promised Apple for its first milestone, but double the speed of Intel's fastest chip. The results were by no means perfect, but they were remarkable. "It was 4:45 in the afternoon when the system booted up," recalls Dorris. "Everyone was pretty excited. Champagne corks were flying.''

In October, Exponential unveiled its microprocessor at a bash that it threw for Apple and for the press at the Fairmont Hotel in San Jose. And in November, at Comdex, the company unveiled the processor before a more general audience, earning rave reviews by using the device to run amazingly realistic 3-D animation.

"I remember the kickoff party," recalls Isabel Gonzalez, 44, Shriner's administrative assistant at Apple, who rejoined him at Exponential. Her husband, Victor, also joined the company, as a systems administrator. "You could see the excitement on everyone's face. On that day, we introduced our product to the world.''

The world liked it. Despite Apple's troubles, Exponential raised more than $12 million in additional venture money. That was the good news. The bad news: The money was already spoken for. Shriner had not been blind to the problems at Apple. So he had recruited another team of engineers, and charged them with developing superfast chips for the Intel-compatible market. It was a smart move -- but it carried financial and organizational costs: The two teams of engineers were spending piles of money. And the Mac-compatible team was doing its best to run the Intel-compatible team out of San Jose. "It got real ugly,'' Shriner recalls. "It came down to a battle for limited resources.''

Hoping to avoid a civil war, Shriner talked with Paul Nixon, a manager and microprocessor engineer with whom he had worked at Apple, about moving the Intel team to Austin, Texas. Nixon liked the idea and began recruiting engineers in Austin. Within two months, he had 20 employees -- most of them engineers -- on his unit's payroll. Now there were, in effect, two Exponentials.

Then came a pair of back-to-back blows that would ultimately seal Exponential's fate. On December 20, Apple agreed to buy Next Computer Inc. for $430 million. Along with the company came Steve Jobs -- Next's CEO, Apple's former CEO, and now a key adviser to Gil Amelio. Among Jobs's first pieces of advice: Drop the OS licenses, ditch the clones, reverse course.

A little more than two weeks later, Amelio announced that Apple had lost $120 million during the previous quarter and that layoffs would follow, as would other severe cost-cutting measures -- including a reduction in R&D spending. "It was clear then that our future would be with the clones,'' says Shriner. As if to reinforce the point, Power Computing stole the show at that year's Macworld convention, held in San Francisco -- and Power Computing executives were proselytizing Exponential's chips. Macworld magazine, in the issue coinciding with the convention, named Exponential's chip "the most promising technology'' for the year to come.

"Rick was the belle of the ball,'' Valdes recalls. "I'm telling you, our phone didn't stop ringing.''

Meanwhile, Hitachi had produced an inventory of more than 7,000 410-MHz chips, and Exponential began to stockpile them. But there were no buyers. Apple balked, saying that the chips weren't up to standard. Furthermore, John Rubenstein, Apple's newly installed vice president of engineering and formerly Jobs's lieutenant at Next, announced that Apple had become intrigued by a promising PowerPC chip that IBM was developing.