Nvidia has recently announced that they will be licensing their GPU technology to other chip makers. This meant mainly for mobile chip makers, who can use their GPU instead of Imagination or Vivante’s GPU, coupled with ARM CPU cores, but other companies such as Intel or AMD could theoretically license Nvidia’s GPU tech in the PC space, too.
Considering the Tegra line of chips hasn’t been doing so well lately, I think this is exactly the thing Nvidia needs to do right now. The mobile market is vital for their long term survival, considering both Intel and AMD can make their own CPU’s and GPU’s now, and Nvidia can only make GPU’s in the PC space, which means they are very vulnerable in the future. Intel will keep pushing its own GPU’s more and more into laptops, and Nvidia will be relegated to an increasingly smaller market at the high-end.
The problem is Tegra hasn’t been as successful as they hoped it would be, with the exception of winning the Google Nexus 7 contract, and not because for lack of marketing or hype, but because Nvidia has consistently failed to deliver on what they were promising, or if they did deliver, it was at least half a year later, when the competition has already surpassed whatever they promised with the new chip.
Another reason why licensing Kepler, Maxwell and future GPU architectures to other chip makers is because Nvidia has been working on getting exclusive games for its GPU’s. But that’s very hard to do when they don’t have a lot of market share, because then developers aren’t as interested in making games for their chips and using their specific GPU features. So by licensing the GPU for others to use it, too, they can expand their market for their GPU, and thus make developers a lot more interested in making games optimized for Nvidia GPU’s.
If Nvidia can get a large customer like Huawei, LG or Samsung, that would huge for them. At least on paper, their Kepler GPU, looks to be the most advanced GPU that will be released next year. I could see why Samsung would want to use that for their Exynos chips, especially now after going back and forth between Mali and Imagination GPU’s, which means their choice of GPU is not set in stone yet.
If you’re wondering how can Nvidia still differentiate in the future if they do this, I say do not worry. They plan to release their own ARMv8-based CPU core called “Denver” in 2 years, so then they will reverse things a bit. Their GPU will be commoditized, but the CPU will not be. As far as strategies go, this one is pretty brilliant, but as always, it will depend on how well Nvidia can execute on it.