In a significant move, NVIDIA has announced that it will soon begin licensing its GPU and visual computing IP to third-parties. What this means, is, companies who wish to implement NVIDIA’s technologies into their products will be free to do so, without having to get the product from NVIDIA. For example, an SoC maker might want a different CPU than what’s available in Tegra; they could build their own SoC and implement NVIDIA’s IP in the way it best sees fit.
Technologies that power this could soon power other companies’ products
This is an interesting way for NVIDIA to get its technologies into more products and more markets than it would be able to accomplish on its own. By licensing out its IP, NVIDIA might not have any clue what companies are planning to do with it… it’d be out of the company’s hands, and its wallet would be the beneficiary. It’s an interesting concept, and one that works out well for the likes of ARM. It really does make sense that NVIDIA would make this move.
Will this result in a killing-off of NVIDIA’s current product-lines, like GeForce and Tesla? Not at all. It’s highly unlikely that any company would license NVIDIA’s IP to build a verbatim graphics card; Instead, NVIDIA will continue doing its innovating, releasing brand-new parts, and once the current IP formula is complete, will license it out to other companies. Our desktop GPUs are safe. Where we’re really likely to see NVIDIA’s IP used most is in mobile, although the possibilities are huge.
Currently, NVIDIA is offering its Kepler architecture IP, but it’s noted that future architectures will be licensable as well. This whole move seemed worrisome to me at first, but after more thought, I actually think it’s a great one by NVIDIA, and it could result in some very interesting products on the market.