team, these grads were just youngsters, barely older than some of their own kids. But they brought a new dynamic and energy to the team that couldn’t be created any other way. They were proud, outspoken, egotistical, very confident in their skills, and convinced that they had something unique to contribute. As lean as we were, several of these wonder kids quickly moved into critical roles, taking on far more responsibility and workload than some of their more senior colleagues. These youngsters may have held some totally unrealistic expectations of their careers—like believing they would move from junior engineer to CEO in a mere handful of years—but they remained resilient, willing to learn, and highly competitive.
In addition to college graduates, another very attractive talent pool was the significant number of experienced engineers, like me, who left IBM at the height of the technology boom of 1998–2000 to find fame and fortune in the startups. Kahle and I conspired to hunt down some key former colleagues and make offers they couldn’t refuse. It was something of an uphill battle, because even if we could entice one of our buddies to return to the ship, it was always possible that the IBM executives would reject the request for authorization to hire. Many true-blue executives flatly disagreed with the practice of rehiring former employees, believing them to have been disloyal in their previous abandonment of the company. Fortunately, Akrout was not so adamantly opposed. He couldn’t afford to be.
If we were to have any hope of filling out a team in time to meet Sony’s goal for a product launch in 2005, we were going to have to do whatever it took to bring in the talent. I leaned back in my chair and propped one sandaled foot on my desk, while I casually phoned a former IBMer, a guy I worked with several years before. We chatted for a while, bemoaning the demise of the startups, remembering the excitement and the dream we shared of the pot of gold just waiting for us to claim it. The potential certainly existed, and a few people we knew landed in startups that went public or were purchased by some bigger corporations.
My friend sighed. “Some people got lucky, but most of us just got worn out. It was hard work, long hours, and, in the end, all you wanted was your paycheck, which was two months late.” He was more than happy to come in for an interview.
The startups were drying up, innovative work was suddenly scarce in the industry, and the economy in general had tanked. Everyone was hungry for something exciting—and stable—to work on, and they recognized the game processor as a once-in-a-career opportunity. We had no trouble lining up eager engineers to interview. The promise of a new high-performance microprocessor design team allowed IBM to cherry-pick top talent from some of their competitors. Given the right incentives, good engineers were willing to go wherever the best work existed. Company loyalties didn’t have a chance against the spectrum of design work we offered.
Each new addition to the STI team, whether an internally transferred loyal IBMer or a prodigal son/daughter returning to the fold, brought with them priceless contacts with other engineers who might be interested in hearing from IBM. It was always easier to convince potential candidates to join a team of old friends.
Sony and Toshiba did not have as much luck filling out their part of the STI team and, like IBM, they were hampered by the staffing demands of other high-priority projects within their companies. Ongoing work on PlayStation 2 derivatives consumed a significant number of Sony’s workforce, while Toshiba was busy with their own products. Out of necessity, Akrout gave the order to continue bringing in IBMers to achieve the critical mass necessary for the STI team to achieve their goals.
This led to an unequal split in the team, with IBM having approximately three-fourths of the resources. Akrout predicted the occurrence of this situation, but kept a low profile in front of his Japanese peers. Since each company contributed one-third of the funding, this resulted in a lucrative deal for IBM (three-fourths of the resources, but only one-third of the salary bill), and a good means for Akrout to continue to hold onto a larger pool of talent for any future IBM needs.
While Sony and Toshiba did not contribute as many engineers as IBM, the ones they did bring into the Design Center were top notch. The problem, as might be expected with this international mix of players, was communication. My first close encounter with the Japanese engineers came at a lunchtime gathering that Tony Truong organized. Tony rounded up eight of us, and we headed off to his favorite Vietnamese noodle shop in North Austin. The place was a hole in the wall, but the food was excellent. I always deferred to Tony to order for me, and that day he chose a large bowl of rice noodles with spicy meatballs for me. I was never sure how to eat this tasty mixture of chicken broth, rice noodles, and meat. Should I use a spoon, a fork, or chopsticks? Or should I just pick up the bowl and slurp? I mostly chose the latter.
As we ate, I tried to make casual conversation with one of my Japanese lunch mates, Takeshi Yamazaki, who Tony introduced as one of Sony’s chief engineers on the project. What Tony had not told me was that Yamazaki spoke very little English. I muddled through several attempts at conversation, but Yamazaki returned only blank stares. He said nothing.
At first, I thought he was one rude fellow, but then I finally figured out the problem. I thought to myself, “This is going to be an interesting project if none of my partners can speak English!” I certainly couldn’t speak Japanese.
At just about the same time, IBM halted all external hiring, proving true the predictions of that wise human resources representative. Corporate management informed us there was a surplus of design engineers in the company who we must re-deploy before considering any more external applications. I can be as true-Blue as the next IBMer, but I knew right away that this was not going to be any easier for us than hiring off the street. In fact, it would probably be harder. As in any large family, our remote cousins were not enthused about letting us take charge of the family fortune.
First, we looked to Rochester, Minnesota, the hotbed of high-performance microprocessor design in the mid-1990s. Their AS/400 minicomputers had been popular business and accounting computers, but those machines relied on proprietary operating systems and software. They eventually lost market share to the more popular open source Unix computer systems. With this erosion in the customer base, IBM corporate offices pulled a sizeable chunk of funding from Rochester, forcing a downsizing. Despite the reductions, some top engineering talent remained in Rochester.
My first task was to figure out how to integrate a large group of engineers into my Austin-based design team and keep them motivated and happy while working remotely from Rochester. After several brainstorming sessions, I decided we needed to partition the microprocessor core in such a way that we could give the Rochester team a self-contained, “meaty” portion of the design, sort of a mini-project. This would serve two purposes: first, it would minimize the interaction required with the rest of the core team in Austin and, second, it would build pride of ownership at the Rochester site. The alternative was to give them responsibility for bits and pieces of the design work scattered throughout the core, a choice that was far more likely to complicate communications and make the Rochester team feel like hired guns.
I deployed a similar technique with a team from the Research Center in Yorktown, New York. The Yorktown folks were some of the best engineers in the company, who had made major contributions to the Power4 microprocessor design. I was counting on a similar performance from them for this new design. I gave them a self-contained piece of the design so that the Yorktown contingent could work on their own mini-project.
Since external hiring for IBM was no longer an option, we worked out a one-time good deal with Toshiba where they agreed to hire a hotshot ex-IBMer, Jack Bell, who we desperately needed for our verification team. It sounded like a great idea to everyone, and we all got what we wanted. The reality was very messy, though. Bell was looked upon by IBMers as “one of us,” so he was privy to lots of secrets that we didn’t necessarily want to share with our Japanese partners. We conveniently forgot he was in the room when we discussed IBM-only topics. Thankfully, he didn’t report everything he heard to Toshiba, because in truth, at the beginning at least, his loyalty was still to IBM. Meanwhile, Toshiba engineers
