It wasn’t long ago that the Nvidia name was liquid lightening. The company’s main competitor, ATI, began to lag Nvidia’s performance significantly starting with the GeForce 6 released in 2004. As the generations moved on, the gap between Nvidia and ATI seemed to only become wider, a trend that climaxed with the Geforce GT 8800. Released in 2007, the 8800 was all things to all men. It was extremely fast, but also very affordable.
This, however, was Nvidia’s peak – everything since then has been downhill. ATI, in a surprise move, delivered a blow to Nvidia with the ATI Radeon HD 4000 series. Released in June of 2008, the HD 4000 series managed to match or exceed the performance of Nvidia’s products. But it was worse than that. The GPU used by ATI was far smaller and more efficient than those being used by Nvidia. This meant that Nvidia was using more raw materials to build their chips. Knowing this, ATI instigated a price war.
Now the results from that war are becoming clear. Nvidia shipped less discrete graphics products than ATI in Q2 2010, the first time Nvidia has been bested in four years. While some might want to brush this away as a fluke, I think this trend will only accelerate. Consumers in this market seem to response slowly to changes in the performance crown. Nvidia was the default choice of gamers for four years – to this day, there are laymen gamers who believe Nvidia cards are far-and-away faster than similarly priced ATI products.
Nvidia’s challenges in the discrete graphics market are only part of the company’s woes, however. In fact, they may well be the least of Nvidia’s worries. The larger thorn in the company’s side is Intel and, to a lesser extent, AMD”s processor business. The problem is this – the integrated and low-performance discrete graphics segments will be gone in two years. They’ll have almost completely vanished. Why? Because both AMD and Intel are incorporating fairly powerful graphics technology into their processors. Anandtech’s recent preview of Sandy Bridge, the processor architecture Intel will be releasing next year, tested the performance of Sandy Bridge’s new integrated graphics component. It proved capable of matching the ATI Radeon HD 5450 in a series of gaming tests including Batman: Arkham Asylum and Dragon Age Origins. And this, keep in mind, was a pre-production sample.
AMD’s upcoming Bulldozer architecture, while it has not yet been tested in any way, is also focusing heavily on graphics performance. Considering AMD’s experience, and what we know about the architecture’s design so far, it seems likely that Bulldozer’s graphics performance will exceed that of Intel’s Sandy Bridge.
So, Nvidia is struggled to keep up in the discrete graphics market and the low end of the market will soon be taken over by processors integrated graphics. Not a pretty picture, is it? And it gets worse.
Nvidia’s CEO, Jen-Hsun Haung, is no idiot. He is a driven business leader with a good sense of where the market has headed. As a result, Jen-Hsun has undertaken numerous projects aimed at broadening Nvidia’s business. These include Tegra, a line of system-on-a-chip solutions meant for cell phones and other low-power devices, and Tesla, a brand of GPUs aimed at supercomputing applications. Both of these new brands are good ideas, but the company so far has had little success. Tegra’s biggest design wins were the Microsoft Zune HD, Kin One and Kin Two. The Kins were complete failures. The Zune HD, while a good product, has never sold in large volumes and probably never will. Tesla has had even less success.
This doesn’t mean that Jen-Hsun won’t pull a miracle out of his bag of tricks, but the clock is ticking. I will be keeping a close watch on Nvidia’s future developments, and I’ll report them in future deathwatch articles.
The deathwatch idea is blatantly stole from TTAC, one of my favorite blogs. Check them out!