Can They Do No Wrong?

I just got back from a whirlwind trip from Boston to New York... There seems to be a lot of questions flying around about ATi and NVIDIA’s competitive landscape. I have received many emails asking how things are looking from within Voodoo so I thought I would write a quick non-fiction, generalist standpoint so you can do your own analysis based on my experience.

First of all, what people need to understand about the GPU business is historically a major slaughtering of the competition only occurs when the other falls flat on their face. So if the green team falls flat, the red team skates by, and visa versa.

As an example, let’s look back a few years ago when Nvidia lost two product cycles. Back then Nvidia failed simply because they fell flat on their face with the much delayed much disappointing launch of the GeForce FX 5800 - while ATi skated by with their outstanding Radeon 9700. Prior to the beating, Nvidia was focusing many of their important engineering resources on the original XBOX and perhaps it was because of this that they missed out the win on the high end space. In this business if you lose on the high end you will ultimately lose on the mainstream.

If you asked me at that time roughly what our internal ratios between ATi and Nvidia were on the desktop I would have told you 100% ATi and 0% Nvidia. Nvidia knew they were losing the high end, and they were determined to change it.

The legend states that the CEO of Nvidia ultimately admitted defeat before his entire team and vowed to never take a back seat to anyone again. They joked publicly about their failed cooler unit on the GeForce FX 5800, and it seemed as if Nvidia had been humbled by their competition. Nvidia was a changed company (a serious heavyweight in the making); they started focusing on platform designs (SLI/Nforce4), better GPU technology (6800+), and of course the coolest notebook concept in years (MXM). Nvidia was determined to, at the very least, catch up in performance while schooling their competition on platform designs.

If you asked me two years ago roughly what our percentages were between ATi and Nvidia I would have told you they leaned towards Nvidia – eventually leading towards 100% Nvidia – this prior to ATi launching their x800 series. That’s a pretty insane change – not based on the other’s success, rather the competition’s failures.

Essentially ATi had fallen flat on their face with x800 and their initial Crossfire launch probably for the very same “Xbox Curse” that Nvidia happened to hit a couple of years earlier. The bottom line is ATi failed to recognize the potential NVIDIA’s focus on platforms such as SLI, and they focused huge resources on XBOX 360, therefore Nvidia conquered on multiple fronts.

We never really took ATi seriously at that point and the mainstream had no idea what was going on with the high end. The bottom line was ATi was not producing cards properly, they were mismanaging some relationships, and they misfired on Crossfire a few too many times. This led to the point where Voodoo was selling 100% Nvidia on the desktop side up until late 2005 to early 2006. Voodoo is on the bleeding edge - and what happens within our four walls typically expounds to the enthusiast market which ultimately leads to mainstream.

Thanks to Nvidia SLI they were able to sell two GPUs for every high end machine for a very long time, which meant we sold nearly twice as many boards as we ever had. That again confirmed their performance lead was dramatic. They gained more market share in the chipset market than ever before, even to the point where it seems like VIA and SIS all but gave up competing. Nvidia even bought ULI recently (a company that makes fantastic south bridge chipsets) for a song, basically eliminating the potential of any company competing with Nforce4. ATi tried but could never make a chipset worthy of a true enthusiasts PC. It seemed as if Nvidia could do no wrong. ...and really, they have done nothing too “wrong” since. They continue to be managed under their centralized approach, while ATi manages with their de-centralized approach. Both have advantages and disadvantages, which I may get into in a future article.

It was in late 2005 that we realized ATi was onto something with their X1900 series. ATi was about to release a card that would take them out of the stone age and back in competition with Nvidia. They may not be way ahead of the game, but they are way ahead of where they were and they have certainly caught up with their competition. The only drawback was that ATi launched Crossfire with the RD480 chipset, which was merely considered “okay” for us.

Meanwhile it seems as if ATi had locked a bunch of computer gamers in a room, one in particular who is very well known for his overclocking skills, and they were secretly designing the Xpress 3200 chipset. This is one bad-ass chip, it’s everything that an enthusiast chipset should be – and it’s amazingly overclockable. It’s probably among best motherboard chipset that I have seen to date, and that’s saying a lot – because I absolutely love Nforce4.

Let me be clear, ATi has never really made a killer motherboard chipset until now; and as of March 1st if you asked us our ratios I would tell you that ATi has done very well within Voodoo just based on the fact that they made a great package. We are about 50/50 at the moment in the extreme high end, based on all the positive press that we have been receiving, and that ratio seems to be sticking. Xpress 3200 with Crossfire technology combined with the Radeon X1900XTX has proven to be ATi’s white knight.

So again, that’s how it works in this space – in order to have a definitive slaughter over your competitor - they must absolutely fail miserably. If no one fails it's likely that we will be looking at a very even, very competitive horse race. At this point neither ATi nor Nvidia has failed and as far as many of us are concerned they are walking on nearly even territory with both shoelaces tied tightly.

With the success of the XBOX 360 (it could be MUCH more successful, I will be writing about that soon) and the upcoming Nintendo Revolution ATi will probably have a much better year than they’ve had in years. Of course there is the much delayed wearingly anticipated Sony PS3 for Nvidia – doubtful to launch in volume this year, but then again that’s typical with any console launch and it’s probably out of Nvidia’s hands. However, one cannot take away from Nvidia's dominance in the chipset market - including server.

There is the challenge that ATi faces of breaking into existing platforms. Nvidia practically owns the enthusiast chipset market on current platforms. While many people don’t mind going out and dropping a G-Note on a pair of video cards, once they have to spend another $100 or so on a motherboard the mind barrier takes over. Typically people believe once you need to replace the motherboard you might as well buy a new PC – that brings me to my next point.

ATi, still a bit bruised, may overcome this once Intel Conroe and AMD AM2 are launched. If ATi executes their chipsets on a timely manner, and they perform as well or better than the RD580 then it is most likely fair game. If ATi extends their chipsets into the server market then we're talking about an entirely different landscape.

ATi also has the distinct advantage of being able to work in Crossfire mode on an INTEL chipset – something I haven’t seen from Nvidia yet. Could that mean serious implications, or does it mean that Nvidia just has to flick a switch to make it work? Perhaps Nvidia prefers to eat Intel’s chipset market share with NFORCE 4 instead…. Perhaps both ATi and Nvidia will take chipset market from Intel as well. Time will tell.

This is going to be a great year for graphics; we finally have competition coming from both sides: Bring it on boys and girls, let innovation prevail!