Earlier this week I had the distinct pleasure of joining AMD CEO Rory Read for dinner and hear about some of his big plans for AMD. Rory has already accomplished quite a bit in his first seven months as CEO including announcing AMD’s intent to acquire SeaMicro along with the restructuring of AMD’s relationship with GlobalFoundries leaving AMD free to use other fabs to manufacturer future processors.
Despite everything I knew about AMD from working with them for many years, I didn’t know quite what to expect from Rory. Sure, I knew he spent 23 years at IBM before his recent 5-year stint leading up Lenovo. I conjured up images of a hypothetical Mr. Think Pad and tried to anticipate what to expect. Three hours later I left the room totally impressed and re-invigorated about HP’s opportunities to work with AMD to develop innovative new server products to address the tectonic shifts that are facing the industry.
Unlike any mental image of a stiff, grey-suited IBM exec, Rory is instantly approachable and likeable. I quickly discovered we had both started our careers in software development and enjoyed authentic Italian food, and the evening was off to a good start. Honoring trade secrets, I won’t give away Rory’s favorite Austin Italian restaurant, but I know where I am heading next time I visit Austin. While keeping his favorite Italian restaurant a mystery, I am happy to report that at home Rory keeps an HP TouchSmart, AMD-powerred, of course.
But back to industry shifts. Mainframes to client-server. Client server to Java, the Internet, and Web services. Anyone with 20+ years experience in the IT industry has lived through several of these shifts and seen the dominant technologies and often companies of one era blindsided and left behind by the next. What shifts are underway today that Rory invisions will reshape AMD and their current and future competitors over the rest of the decade?
Rory spent a lot of time focusing on what he called the “3 C’s”, cloud, client, and consumerization and this certainly resonated with me. For most of the history of computing, the industry focus was on business computing. Think of what the letters IBM stand for. If you could build a computer that was big enough to handle the business computing (payroll, accounts receivable, etc. ) needs of the largest Fortune 100 companies, that was pretty much the pinnacle of IT. Boy, where those the easy days. Some of today’s biggest data centers, think Facebook or Google, are filled with rack after rack of computers, many with AMD processors, serving up consumer services like games and social networking. While that is good business today for AMD and HP, it is also a big challenge for the future.
Search, social networking, and web hosting alone will drive tens of billions of $ of new server demand over the coming years, as will new HPC applications as companies increasingly use digital design and simulation to speed time to market and increase product quality (think how simulated car crashes help make automobiles safer). In today’s mega data centers, it is no longer sufficient just to lash together more and more racks of low-cost servers, with more racks of network gear, and storage, all with separate power, cooling, software, and management. The next generation data center will be filled with new types of extremely energy efficient processors, meaning not ten’s of servers in a rack but thousands of simpler and lower power x86, ARM, and other processors. Now no one wants to have 1000 servers in a rack and then need to put in another 5 racks just for the network. That means server vendors will need to start to build converged infrastructure, thinking of not just the server but how to fit the server, network, and storage all in the same rack, along with high density power and cooling, all managed together versus independently as typically done today. This is exactly one of the goal’s behind HP’s Project Moonshot. It is also one of the stated reasons behind AMD’s acquisition of SeaMicro, namely for their fabric chip that helps integrate many of the networking and other functions that require discrete chips in the typical server today.
It thus comes as no surprise that AMD was one of the first to join Project Moonshot’s Pathfinder Program, and become one of HP’s partners in helping to redefine the data center of the future. Rory has a great vision for where he wants to take AMD, and it is going to be more exciting than ever working with AMD and the rest of our Project Moonshot partners in the months and years ahead.