The first technology wave was won by IBM and it was unbelievably powerful in the 70s and early 80s. Since then, there have been two big server revolutions: The first was from mainframes to client server configurations, and IBM was the big loser and Sun the big winner; the second was from RISC proprietary architectures to x86, and Sun was the big loser (what comes around goes around) and Intel and Microsoft were the big winners.
The next revolution is from internal services to externally supplied resources as a utility and the big winner isn’t defined yet (Google clearly thinks it will be it) and the big loser is typically the firm that was the big winner on the last wave, but neither Intel nor Microsoft want to become the next IBM or Sun.
What is interesting is the first two waves were largely owned by solutions companies; the third wave was owned by the companies that provided the operating system architecture and the lead processor. I should point out that this was one of the few places where Steve Jobs failed with the Apple servers spectacularly. But given Google really isn’t stepping up and Microsoft actually has the most powerful Web services for business at the moment (Azure/Office 365), I’m not yet convinced that we can tell who the big winner will be.
However, one of the indicators of change is the recent announcement by AMD to use its SeaMicro acquisition to create an ARM server solution. I think this move is indicative of changes to come and some rather big changes in how servers are built and used.
The last big server wave was on x86 and the end-to-end advantage that both Microsoft and Intel had in that era was unmatched. As long as PCs were king, then x86 servers made sense because the environments were similar and interoperability optimized. But we are moving away from a traditional PC world and while I could, and do, argue that smartphones and tablets are the most personal of computers, they are also very different than their x86 predecessors in terms of hardware, operating system and vendors that are successful.
The PC manufacturers have largely bounced off of smartphones and tablets to date with Apple being the exception, largely because it recreated itself as a consumer electronics company with a PC business. Samsung is now the other big company to watch and it is at its heart a consumer electronics company with a PC business.
But while we clearly have had a changing of the guard on the client side connected to the change of emphasis from PCs to tablets, we have yet to see this same change in servers and that is what I’m anticipating. Because as much as x86 clients drove synergies that made x86 servers more economical than their RISK counterparts, we are seeing the same thing with ARM.
But while most ARM vendors can’t spell “server,” AMD has been competing with Intel as the number two player in the server space for over a decade. However, catching Intel from behind in the x86 space hasn’t been a winning proposition because Intel can out resource AMD and even in the rare cases where AMD has an advantage, it tends to be over too quickly for the firm to capitalize on it. It needs to find a way around Intel into a space where Intel might struggle and ARM has become its vehicle.
Its plan is to enter the server market in 2014 with a product that is ideally tuned to address this emerging mobile device opportunity and demonstrate the same economies of scale that worked so well with x86. And also to recreate the ability for developers to more easily move between the two platforms to create compelling solutions faster with less cost. The result would be a far more compelling alternative to Intel than it can create using the x86 architecture.
However, dominance, if Intel’s success is to be recreated, would require either a merger or a tight partnership with a firm doing the client side of this solution and Qualcomm might be the obvious partner because it acquired AMD’s old mobile platform business.
Wrapping Up: Changes
I think the AMD-ARM move means that for much of this decade we will see the emergence of a new server class designed to provide services to an increasingly mobile-device-equipped world. AMD plans to be at the center of this move, but to become dominant it’ll need both sides of the solution and that means a merger or tight partnership is in its future with the most obvious candidate being Qualcomm.
We’ll see how this all plays out because Intel is pushing back extraordinarily hard with its own thin solution and has a better server position, but it is well behind on the new mobile client and would be better matched by the much more powerful, in mobile, potential partnership between AMD and Qualcomm. In any event, servers sure aren’t going to be boring going forward.