Sadly things all changed due to an ill-conceived buyout of ATi which led to AMD dropping behind in R&D as well as actual manufacturing of CPUs. This buyout combined with an extended legal battle with Intel on multiple fronts nearly spelled the end for AMD and their newly acquired GPU/chipset company. We saw a number of products from both sides of the house that just were not able to really compete properly. All of them seemed to be one generation behind in terms of performance. AMD also went through multiple CEOs and saw a number of long-time employees leave the fold for good.
Now, in 2014, we have to wonder (again) just exactly where is AMD going? They have tried to get back into the game with Intel by re-launching their FX brand and even tried to push a massively overclocked CPU out to the public. None of these bought them back much market share although they have been slowly building momentum with their APUs. On the GPU side, AMD seems to be trying to pin down the sweet spot in terms of performance and price. Their GPUs have been getting steadily better while their gaming API Mantle is gaining more traction in the development community (even without a large budget to push it).
AMD also bought themselves a complete HPC (High-Performance Computing) company and brokered a deal to manufacture ARM based products. They are now one of the more diverse companies when it comes to what they are actually capable of producing. The do have to rely on the likes of TSMC or Global Foundries to get things from the design stage to the real world, but they have considerably more ammunition to work with than ever before.
This puts AMD into an interesting position, if (and this is a big if) they can execute on their plans. We expect to seem AMD dive into a few new markets in the next few years including high density server. They have an advantage over Intel here in that they can provide multiple product bases from a single company including ARM, APU and traditional x86 CPU bases products. AMD also can (and is likely to) push back into the professional graphics market. This is an area they have allowed to drop off while catering to the enthusiast and gamer crowd. It is another unfortunate side effect of having a limited R&D and manufacturing budget. AMD simply cannot do everything they want to.
This also means we are unlikely to see a CPU that competes with Intel or to keep the AMD enthusiast happy. We might see something in this vertical in 2016-2017 time frame, but even then it is not likely to be much of a step up until AMD gets their next wave of professional/enterprise products out to market. Until that time AMD is likely to stay on the 28nm process for their CPUs and APUs, putting them very far behind Intel and others that are already exploring 14nm. That having been said, if AMD can work out the kinks on their HBM (High-Bandwidth Memory) and die stacking we could see some significant improvements in performance in whatever product comes out. AMD might also simply leave their CPUs at 28nm until 2016 so they can push APUs and ARM based products into smaller processes for greater power/performance ratios. We have heard they are moving to FinFET instead of sticking with their Silicone on Insulator style they have traditionally used. This shift could slow down their plans to move their CPU/APU product line to smaller processes in the near term.
Over the next 3-4 years AMD will be making a short term shift to the pro side of the world in terms of CPUs and to a lesser extent GPUs. We will still seem them push out GPUs for the gamer and also improve on their APU offerings, but the main focus is likely to be on enterprise and professional products. While many AMD fans will be sad to hear this, it is a needed move for now so that AMD can eventually move some of the new technology down into their enthusiast and consumer products once they are profitable. 2015 is likely to be the year that AMD tries to make all of the partnerships and purchases payout… let’s hope they can do it.
Tell us what you think in our Forum