It’s been just over a year since Intel and AMD reached their landmark settlement, and two years since the company entered into a joint agreement with ATIC on the fab. Those were huge movements involving huge sums of money and they take a long time to be digested. AMD never stopped doing what it does, and neither has its graphics group, the former ATI, but with these two monumental changes, the management of AMD has been able to focus better, and get their house in order. You can see the results of that new focus and vitality in some of the re-organizations, new hires, and change in marketing.
Here’s what I see on the horizon for AMD.
Fusion. Probably the biggest thing to impact the industry and certainly AMD is the absolute marriage of ATI and AMD technology in one chip. No one has anything like Fusion, nor will they for several years, if ever. But it’s not a slam dunk, just yet. There are still fabrication issues, balancing of x86 CPU performance vs. wattage, getting the OEMs and ODMs to think outside the box. A lot of questions still exist about the benchmark performance of the heterogeneous processor AMD calls an APU. Based on discussions with AMD, the graphics performance should be quite good. However, the first units to the market, the Ontario and Zacate, built in 45nm at TSMC’s fab will be average to good we think. The real performer will be Llano built in 32nm at GlobalFoundries. With Fusion processors AMD has the potential to claw away at Intel’s dominate market share. They also have the potential to eat away at Nvidia’s low end GPU market. But to do one they have to do the other – and therein lies the gottcha. If AMD can’t convince OEMs that a Fusion is better than an Intel and Nvidia combo then all the company will accomplish is to replace their IGP sales. Neither Intel or Nvidia seem concerned, and talk on street is AMD will not gain much market share. This is a waiting game – we have to wait for the HPUs to show up and see what kind of traction they’ll get. I’m cautiously optimistic.
Digging holes – Bulldozer and Bobcat. AMD has been preparing two new processors, a big one for servers (but not workstations), and a little one for tablets and netbooks. AMD was criticized for not having a processor for the netbook segment in 2009 when that segment was busy stealing market share away from notebooks. This year tablets will steal market share from netbooks and notebooks and now AMD shows up with Bobcat for those markets looking for the sweet spot, the company’s new theme. In the server space, the company is going to offer 16 core and 8 core versions and try to segment the server market into two categories instead of the four that now exists. This is another wait and see situation for AMD. The strategy sounds good – more cores for less money, but the traditional server OEMs will make the buying decision of sellable benchmarks so AMD is going to have prove it.
Sweet Radeon. ATI got its act together in 2008 and brought forth its sweet-spot strategy and new scale up/scale down architecture. Renamed but carrying on with that strategy AMD has just introduced its latest part the Cayman series. Like Nvidia, AMD has a full line of new versions coming out. They will go toe-to-toe with Nvidia. First results haven’t been that impressive, and there seems to be some supply issues. Final benchmark results with driver tweaks probably won’t show up till CES, but we don’t think the Radeon 6850 and 6970 are going to be exceptional and it will take a dual GPU like the HD 5970 to win the gamer’s attention.
FirePro step sister. Like Nvidia AMD takes their consumer GPU and applies it to three markets: Consumer, GPU-compute, and professional. The professional brand is FirePro and due to limited resources AMD hasn’t been able to come close to matching the emphasis and support Nvidia puts into the professional space, and their market share results show it. Although the company has gained a little ground, it’s not likely they will capture even 10% of the market in 2012. The actual share percentage isn’t as critical as the ramp – if AMD doesn’t have a market share growth ramp (CAGR) of at least 10% by this time next year it’s going to hard to understand why the company doesn’t spin off the group.
FireStream hobbled. AMD’s GPU-compute product line suffers the same resource issues as the workstation boards. And whereas Nvidia has made Tesla a clear server part, AMD doesn’t even include it in their server product list. AMD has opted to offer the product relying on open source tools from Khronos known as OpenCL. And whereas an open system should win out in the long run, OpenCL has a ways to go. FireStream is a place holder for companies that don’t want to be locked into a proprietary system, although Nvidia supports OpenCL just as well as AMD does, but they don’t make a lot of noise about it.
Open everything. AMD has carried the open systems theme to other aspects of the industry for stereovision, physics, and ray tracing.
Super Catalyst. ATI had been criticized in the past about their drivers, and that has cost them dearly in market share in the professional space. It’s taken a couple of years, but those dark days are behind them now. In addition to having stable drivers, there are OC tools, and new built in processor features like stability programs for handheld video.
AMD was in pretty bad shape a few years ago and has made a remarkable and impressive comeback. Things don’t change overnight and the company still has work to, and it’s doing it. The stock market isn’t showing much enthusiasm for the company. Since September AMD share price rose about 20% while Nvidia’s rose (from a historic low) 30% in a market that rose about 15%. The investor community seems to be taking a wait and see attitude with a few of them betting on the company and buying in now at what they hope is a discounted price.