an assignment with IBM’s research group, the beginning of a career-long tie to the folks in Yorktown, New York. His first assignment took him into design tools—money-saving, software-based programs that improved the quality of a design by automatically checking for errors, cross talk in the wires, and other design points that exceeded acceptable limits. He integrated those cutting-edge tools into the design process for the chip that eventually put IBM in the lead in high-performance Unix servers.
Kahle spent nearly twenty years of his life at IBM, serving as a key player on several highly successful development efforts including the breakthrough server chip known as the Power4 Gigaprocessor. He was very smart and very aggressive, big reasons why he was one of the most influential engineers at IBM.
I recalled the first time he invited me to play soccer. I scrounged around in my closet, found my old cleats, and brushed off the dust from a few years of neglect. Half an hour on the field and I was stiff, sore, and exhausted, but not giving up. I joined Kahle on the front line, and we tag-teamed the ball toward the goal. When we failed to score, Kahle ran across the field and screamed in my face for not hustling enough. He prided himself for being the fastest, most aggressive player on the field. He made up for any lack of ball-control skills by out-hustling his opponents. He was a bully, but I admired his competitiveness.
Whether playing soccer or designing a new chip, losing was not an option. Kahle was quick to reject those who didn’t agree with him, and more than one IBMer’s career suffered for crossing him. The Sony chip lay on Kahle’s career path; it was his next stepping stone. Nothing would keep him from success. I was more than a little flattered that he saw me as an asset in that venture, but I remained on the fence about joining up. I enjoyed the impact I could have in a small company. Going back to a big company like IBM wasn’t exactly appealing.
I drained my mug and then motioned to our waitress for another round.
Kahle described his new design center, which fell under the auspices of IBM’s Microelectronics Division. “I spent the first few months of the STI partnership serving as the director in charge of both business and technical issues, even helping to write the contract. Boring stuff. As you can imagine, I was dying to get involved in chip design. The technical complexity of the chip soon consumed my interest, my energy, and my time, leaving little for the business side of things. Paperwork fell into complete disarray, I had no time for staffing, and the executives started to worry.”
“I doubt anyone could handle both of those challenging jobs,” I sympathized. As I well knew, Kahle was a technical wizard. He would perform much better in an engineering position than he would in a management or business role.
“Maybe not, but I quickly found out that I didn’t want to handle both jobs. I took the title of chief engineer, focusing on what I do best, chip architecture. I let others handle the contracts, financial controls, staffing, and other adminis-trivia. Finally, I was in the right place.”
I had no doubt he’d made the right decision. He wielded a strong influence over any IBM team, and they generally accepted his technical command as absolute rule.
Kahle finally zeroed in on the point of our meeting. One of the first jobs he needed to fill was that of the chief architect for the PowerPC microprocessor core, the “brains” of the chip. “That’s where I need you, Shippy. You’ve always delivered high-quality innovative solutions for me.”
I liked the sound of that. As the chief architect, I would take on the coveted role of the technical lead, defining the fundamental blueprint for the PowerPC microprocessor core. It was a once in a lifetime opportunity, the kind of job every engineer desires. It would allow me to exercise all of my creative and innovative engineering muscles and have a major influence on the design. I leaned back in the booth, ignoring the creaking of the old leather seat, and looked Kahle in the eye. Cutting-edge technology, a chance to be in the thick of things, a key leadership position, ground-floor opportunities—who wouldn’t be interested?
Kahle sensed a win and immediately moved to close the deal. “This project is extremely important to the IBM Corporation, and folks like you, hired into key positions, will be heavily rewarded.”
I smiled at that. He definitely knew the way to my heart. I said, “You just caught yourself a fish.”
We spent another hour or so reminiscing about the good times we’d enjoyed together in the mid-1990s at the Austin-based Somerset Design Center, a joint venture between IBM, Motorola, and Apple. Somerset challenged Intel for the PC market with a sleeker, cleaner, PC version of IBM’s standard Power server chip, called PowerPC. The Somerset office lay outside the main IBM campus, purposefully free of most of the bureaucracy found in industrial giants, thus creating the impression of a small, privately-run company. It was IBM’s bid to give its entrepreneurs the freedom to create and to invent. Kahle’s team delivered the first Somerset microprocessor, the PowerPC 603, and then Kahle brought me on board to architect the follow-on microprocessor, the PowerPC G3. Exciting times! The engineers on my team actually believed they could beat Intel, and I guess I did, too. We worked long and hard on our chip designs, and we frequently ended the day on a sand volleyball court at the design center or at the nearby Arboretum drinking beer together. We were a tight group.
Kahle threw big parties for the design team at his house, nestled on the cliffs overlooking Lake Austin. Always a savvy investor, he built his house back in the 1980s before the rest of the world discovered Austin. The folks at his parties were rowdy and consumed lots of beer and margaritas. Kahle cooked his famous “Kahle burgers,” delicious, if technically inelegant—thick, hearty, and smothered in cheese. Classic rock tunes from the 1960s and ’70s played in the background. There were generally two camps of folks at the parties: The first camp wanted to talk nonstop about work (shoptalk). They could pontificate ad nauseam about the details of their work. The second camp, which I belonged to, wanted to talk about anything but work. Both camps liked to drink a lot.
We were having so much fun, we didn’t even realize we’d awakened a sleeping giant. It took several years, but ultimately Intel crushed Somerset just as they did all of their competitors. Kahle fondly referred to those days as “taking on the dark side.” There was a negative perception of Intel in the engineering world due to the technology giant’s monopoly of the PC microprocessor design space. On the software side, there was a similar perception of Microsoft, which dominated the software used on PCs. Microsoft and Intel, nicknamed the “Wintel” pair for their combination of the Windows operating system and the Intel microprocessor, presented a formidable foe. Even facing that serious challenge, Somerset was still a fun and intellectually stimulating environment that Kahle and I both badly wanted to re-create in the STI Design Center—but this time, we meant to win.
The landscape of high-performance microprocessor design had changed significantly since those days in Somerset. Digital Equipment Corporation (DEC) and their once formidable Alpha microprocessors had disappeared. Sun Computer had lost a lot of their market share with their SPARC architecture. Supercomputer maker Cray Research had also gone by the wayside. Other minicomputer and mainframe shops like Data General, Amdahl, and Hitachi had mostly disappeared. There were really only two horses left in the high-performance microprocessor race: the X86 architecture produced by Intel and AMD, and the PowerPC architecture produced by IBM and Motorola. These microprocessors accounted for the majority of the PC and high-performance workstation business.
The only other exciting microprocessor technology was the nimble ARM Architecture—Advanced Reduced Instruction Set Computer (RISC) Machine, which was pervasive in the emerging mobile and handheld space. Most of the rest of the electronics companies adopted a System on a Chip (SOC) philosophy, where they embedded X86, PowerPC, or ARM microprocessor cores into their own custom chips. They differentiated their products by adding their own “secret sauce” intellectual property (IP) to the chip. For example, a disk drive chip might include an embedded microprocessor core alongside a patented
