Bruce Shaw has only been with AMD for about eight months, but his mission is clear: get more enterprises to opt for Opteron.
While Intel spent most of last week discussing its product roadmap to an international audience at its Developer Forum in San Francisco, Shaw was in Toronto, giving
Canadian customers an update on the competition.
As director of commercial marketing in the Austin, Tex.-based chipmaker’s computation products group, Shaw is responsible for accelerating the company’s growth in the data centre. This involves not only convincing IT managers and CIOs that AMD has the best architecture to make a transition to 64-bit computing, but also to form relationships with tier-one manufacturers. This is a territory Shaw knows well — he came to AMD last year from Dell, where was director of networking sales and marketing.
Pipeline joined Shaw for lunch following his visit with Canadian IT executives to learn more about his plans for 2005.
Pipeline: AMD certainly had a good year last year, but what drew you over from Dell?
Bruce Shaw: One of the things that attracted me to AMD was that they wanted to transform from a traditional chip maker into a more customer-centric, data-centric company. That I found very appealing, because generally, working on the other side of the fence at a hardware manufacturer, it had been my experience that the guys making the most important component were fairly unwieldy in terms of listening, doing and executing the things that customers wanted. I think AMD was really poised, with Hector Ruiz taking the CEO’s spot, to make a change.
Pipeline: How are the corporate changes translating into something you can take advantage of from a marketing perspective?
BS: The two-year anniversary of Opteron is coming up in April. It’s pretty neat to see a chip that went from non-existence in the market to a ballpark of about seven per cent of the IT market today, depending on whose count of the market you like to look at. That’s phenomenal growth for a relatively small company. The proposition was, “Come help us work and look at the customer base, look at how servers are used and applied, and tackle the marketing equation from that standpoint.”
Pipeline: And how are you doing that?
BS: Probably the biggest thing that has us having a fully-booked day is the fact that you have a plethora of systems from HP, Sun and IBM in the tier one space that’s now being based around Opteron. IBM recently announced a new blade series, and this was on the heels of some other announcements that were very interesting. HP earlier in February announced five Opteron-based products. So I think what you’re seeing from a transitional standpoint is a company that a few years ago fought its way into the enterprise in niches by high-performance computer clustering — the guys that were purely looking for a performance equation. Now they’re seeing the overall benefit of what we have from an architectural standpoint.
Pipeline: AMD has always suffered from the perception that it’s simply a maker of low-cost Intel clones. How do you market against that?
BS: If you look at the bulk of x86 servers – and certainly we want to put x86 everywhere – you have the bulk of servers today that are sold as standard IT infrastructure-type boxes. There’s file and print, the traditional areas in the enterprise that are the building blocks of most data centres. You then have mission-critical applications. You’ve got the things that, if it goes down, in traditional RISC environments, the business isn’t going to be running. Those markets, while not very large, are becoming more pervasive. So they’ve bell-curved out where the servers are. That four-piece space – the simple two-way, very fast servers were growing with compute clustering, and Opteron was a very natural value proposition there. What it enabled us to do is prove to the world that in the most computing-intensive, mission critical environments in the data centre, we were outperforming the competition at a better performance equation. So the natural thing that starts to happen after that is you start to say, “Well, if it’s good enough to run an investment house,” and you’re literally trusting it to run the business, clearly it’s industrial-strength enough to put in other parts of the enterprise where there are different requirements.
Pipeline: What kind of issues are you trying to highlight in your messaging to IT managers and CIOs?
BS: The things that are getting less acceptable in the enterprise today are power consumption, and we feel we have an advantage there, and server or asset utilization. If you walk into most data centres today and run a utilization test on a two-way server, some of the IT guys would be ashamed of the numbers that they’re giving back in terms of percentages. Given that data centres aren’t getting bigger, IT budgets aren’t growing at the rate with which they’re being asked to do more, that number has suddenly become a very important number. In 1999, the DirectConnect architecture was laid down with a clean sheet of paper, looking at how we directly connect I/O, memory, remove the front-side bottleneck, speed up the latency of the processor from all of its sub-systems. This was a winner. It wasn’t a matter of pushing more gigahertz down a single-road pipe.
Pipeline: So no more marketing based on speed?
BS: I use the analogy of the 401 highway. If the speed limit on the 401 was 200 km per hour, that doesn’t necessarily mean more traffic can go through in the morning, or any quicker. The first big semi out there that’s moving along slower is going to have a natural drag effect on that road. By directly connecting the processor to the memory to the I/O, we’re opening up a much richer infrastructure to pass that data.