What Has Happened To AMD?

There is an excellent 2-part series on AMD on Ars Technica, that provides a succinct review of AMD’s history, and a concise review of the company’s strengths and weaknesses. I, however, want to take a different viewpoint on AMD’s future. It isn’t that I disagree entirely with some of the conclusions in the article. It is just that I think I have a different perspective on why AMD is in more trouble than they realize, and that they may be heading the wrong way.

From Darling To The Dump

Remember when AMD was going to eat Intel’s lunch, back in the late 1990’s and even into the early 2000’s? It sure looked good for them, at least from the outside. As the Ars piece points out, though, internally AMD had (and still has) problems. Lots and lots of problems, set in motion by the reckless spending and unpredictable management style of its founder. On paper, when you looked at PC’s in the late 1990’s, it sure seemed that AMD was going to be a true competitor to Intel. Anecdotally, it seemed as if 1 out of every 3 new desktop and notebook PC models offered AMD exclusively, or as an alternative to Intel. While choosing AMD in their early days felt like a risk, buying a PC in 1998 with AMD instead of Intel didn’t require a huge leap of faith.

But things clearly went south, around about the time they bought ATI. Now, AMD CPUs as an option seem few and far between. Sure, they’re still out there, but there doesn’t seem to be the vocal and vociferous fanbase trumpeting AMD. Their processors are a step behind Intel. They don’t have a real marketing presence. The company has, effectively, become invisible to people. It’s hard to make any headway when your engineers aren’t doing a very good job, your product really isn’t competitive, and you’ve disappeared from the landscape.

But What’s Ahead?

AMD has some problems, because their mainstream markets are either heavily entrenched with Intel (servers) or are seeing a modest amount of contraction (notebooks and desktops). AMD has effectively zero presence in the mobile space, and seems have less wherewithal than Intel at being able to gain even a small toehold. AMD is betting on high-density microservers, but I’m not convinced this is as big a market as AMD believes it to be. With better underlying “big” processors, coupled with the continued rise of virtual environments from the likes of VMWare, having dedicated small servers for most companies just doesn’t make sense to me. Even if the “software fabric” gets sorted out, there are more and better options for resource utilization with virtual servers, and creating a new server instance is a trivial event, compared to having to include hardware setup as part of the picture.

And I have zero confidence that AMD can figure out the “software fabric” problem of trying to create scalable and manageable environments over these microservers. It isn’t that there isn’t a solution. It’s just that smarter and more skilled people have been working on this sort of thing for decades. Some headway has been made, and there are frameworks out there to help build software to take advantage of clusters of discrete servers. But there is still a long way to go to get a general purpose “distributed operating system” to make a bunch of servers look like a single unit to any piece of software, without requiring it be modified to work differently.

In the mean time, cloud computing is taking over a lot of what companies want and need for their server-side computing needs. These cloud services are run on virtual machines on larger hardware. It’s easier to manage, easier to scale and easier to secure than trying to sell customers discrete little servers. There certainly can be a role for microservers (I have some future projects that could certainly benefit from them), but I’m just not convinced it’s a large market.

A Reinvention Of the Company

When you look at AMD’s problems, it comes down to two things: engineering and execution. On the execution front, AMD still seems to be stumbling and bumbling around without direction or purpose (other than “sell some stuff”). Their ATI division still seems to be moving forward, and being selected as the CPU and GPU for the Sony Playstation 4, as well as being rumoured to supply the parts for the next XBox and Wii, helps ATI and AMD tremendously. Being at the core of the console gaming market, a space that appears to be able to grow substantially still, is a big deal.

But execution is only part of the equation. Knowing that AMD faces big-time competition for the server market, as well as needing to fight harder for a shrinking PC market, AMD needs to beef up its engineering. There was a brief time where it was leading Intel. AMD processors were generally faster, and were certainly cheaper, than what Intel offered. AMD’s 64-bit approach has become the standard. But the AMD engineering team has lost their way, and the result is product that always seems to be about a year behind Intel. This is not a sustainable situation. Whether it’s the skill of the engineering teams themselves, or whether it’s how they are managed, AMD really needs to sort this out, and fast.

The challenge, though, is that it seems that what AMD needs is a serious, big-time shakeup in the culture. Apple’s resurgence since 1997 was due, in part, to a revival of the “old Apple” culture, something that got diluted, if not lost entirely at times, under the auspices of CEO’s from Scully to Amelio. IBM’s rebound from near-death in 1993 was due, in part, to a culture shakeup when Gerstner took over. HP’s current problems stem, in part, from the company departing from some of its own heritage and culture.

Create a Vision and A Plan

Part of the cultural rebuild requires that the company first have a vision (which outlines their purpose in life), and then immediately have a plan to make the vision happen. A vision without a plan is meaningless drivel, a soundbite without substance. A plan without a vision is navigating in the dark, because you don’t really know the destination. You need both, together, coupled with solid execution, to make it all work. AMD seems to have neither at the moment.

It will be a scary time for investors, because you can’t undertake this kind of large-scale renovation without first creating a bit of a mess. It will take courage and patience to see the process through, and it will not happen overnight. Realistically, the risk is minimal, because AMD is on a course to irrelevance anyways. While profitability returned for a little while, they are back to their money-losing ways, and the slope of the line isn’t encouraging. They continue to lose marketshare. At this point, heroic measures are needed to truly turn the place around, and to make it relevant again.

And there is one possible direction that needs to be put on the table: get out of general purpose x86 CPUs, and focus on other components, such as being the dominant provider for game consoles and possibly home theatre devices, as well as the GPU business overall. It’s radical, it needs some thought, but it shouldn’t be discounted out of hand. No one would have expected IBM to stop selling PC’s, and that move has proved to be the right thing for IBM. There is no rule that says you have to keep building or selling a product just because “we’ve always done it that way”. There may also be opportunities in the solid-state storage space, since there is still a tremendous amount of growth to come in that area. AMD may also find it useful to explore an ARM licence and find other ways to enter the mobile space. Sure, they aren’t one of the early entrants, but being first (or even in the first wave) doesn’t give you an automatic right to success. I’m sure that companies like Google/Motorola, HTC and Nokia wouldn’t mind having another alternative for ARM processors, particularly from someone who isn’t also making their own mobile devices (like Samsung). Further, the potential explosion for wearable technology could open up space for parts manufacturers. These are certainly radical ideas, and the odds of success in some cases are probably slim. But that doesn’t mean that they are wrong.

Could AMD be great again? There’s no reason why it couldn’t. But it will take a massive teardown and rebuild of the company, a new attitude toward engineering and execution, and a fundamental rethink about what the company does and how it does it. Until AMD is ready to take that step, it will likely continue to stumble along as it has been for years.