Published in Graphics

Nvidia-Intel licensing was really just a payment

by on16 April 2018

Settlement wrapped in a paper

Last year Nvidia stopped getting so called licensing money from Intel. After five years the financial injection was  delivered in total. This all came after the Intel and Nvidia conflict that started over chipsets, but essentially was a settlement that ended up being called licensing. It never was GPU licensing.  

The license – deal – settlement call it whatever makes you sleep better, finished over a year ago in March 2017. Part of that agreement was that Intel agreed to pay Nvidia licensing fees which in the aggregate amounted to $1.5 billion, payable in annual installments, as follows: a $300 million payment on each of January 18, 2011, January 13, 2012 and January 15, 2013 and a $200 million payment on each of January 15, 2014, 2015 and 2016.

The deal is over, and guess what, Intel didn’t announce any new license for the GPU, simply as this never was a GPU licensing deal. People close to the matter from several different independent sides have confirmed that the deal was never about  licensing. AMD’s deal with intel turned out to be a Vega GPU stitched with EMIB to a Core processor and nothing more than that.

It was simply easier for Intel and Nvidia to call it a licensing and not the payment for the damages caused by cutting Nvidia out of the chipset business. Intel had a history of bully behavior and it didn’t take competition well.

Nvidia Denver and Nforce exit

Nvidia did get the money for its loss of the chipset business and both companies called the "licensing deal" a cross patent sharing. Jensen upset Intel by hiring people to make the Denver CPU, and this was one of the things that sparked the quarrel. Denver, to our knowledge, was supposed to be a X86 solution and Intel never wanted to let Nvidia do it. Essentially Denver became an ARM solution and still lives in Nvidia automotive designs that inherited the Tegra business.

Intel is still making IGP inside its CPUs and it has more GPU market share than everyone else combined. It didn’t have to reach to AMD for GPU licensing, as it has enough patents to get other GPU players in trouble. Yes, we mean Nvidia and AMD, and this holy trinity is unlikely to ever sue each other for the patent infringements in graphics. So far that hasn't happened. Very often a company A comes with a solution and the company B comes with an answer. SLI – Crossfire, G-Sync - Freesync are some of the most obvious examples, and yet, Nvidia and AMD didn’t sue each other. The matter is especially complicated when it comes to software support, drivers, as these guys often borrow each other's ideas.  

At some points these guys were borrowing technology from each other and stepping on other patents, but since this was always a mutual thing, no one ever sued each other. The end effect was Intel paid Nvidia a settlement and stretched it over a few years. This is what that was, and it won’t happen again.

Nvidia (Jensen) tried and failed suing Samsung and Qualcomm over some GPU things and it will be incredibly hard to make a case for this in the future. Still, we tend to get surprised by some moves and let's keep our eyes open.

Last modified on 16 April 2018
Rate this item
(1 Vote)

Read more about: