The NY Times thinks the FTC is setting up NVIDIA for x86 products
Subject: Processors | December 20, 2009 - 02:00 AM | Ryan Shrout
A very interesting blog post over at the New York Times is speculating that the newly filed lawsuit by the FTC against Intel could have some interesting implications for NVIDIA in the technology market. The Times seems to think that the FTC is going to purposely open up the market in such a way that will enable NVIDIA to make the oft-rumored x86 processor that could make it an equal of AMD and Intel.
There are couple of specific passages that Ashlee Vance seems to point directly to this possibility:
necessary to interoperate with Intel’s CPUs or chipsets) to others, via
licensing or other means, upon such terms and conditions as the
Commission may order, including but not limited to extensions of terms
of current licenses.
Vance seems to think that this measure is "opening the door for Nvidia to obtain an x86 license from Intel." I don't quite follow that logic in any manner. The measure above clearly states that the FTC wants to see Intel make available technology to "interoperate" with Intel's processors; obviously this is meant to address the complaints NVIDIA has had about Intel locking them out the chipset market.
NVIDIA CEO Jen-Hsun Huang would love nothing more than to see Intel get smacked by the FTC
With the recent move to integrate the various functionality of traditional chipsets into the processor itself, this move will not immediately create a force for change. Unless the FTC plans on dictating that AMD and Intel are not allowed to do things like integrate PCI Express and graphics onto a single chip (which seems incredibly counter to the concept of improving technology), NVIDIA will still need to create a compelling option that has features that will differentiate it from Intel's on-board solutions. This is essentially what NVIDIA is doing with the upcoming ION 2 chipset and what NVIDIA would probably like to do in conjunction with the Intel Clarkdale/Arrandale launch next month. But, because of the legal dispute between the two companies, NVIDIA has been unable to produce a chipset to work with the DMI-enabled Lynnfield/Clarkdale processors.
NVIDIA current ION platform with the Intel Atom CPU
I think the goal of Measure #17 is likely to create a better atmosphere for customers by creating more options in the chipset market, but I don't think it will have any affect on the x86 processor market.
There is another interesting point that Vance points to though:
licensing agreements that restrict the ability of licensees to change
ownership, to obtain investments or financing, to outsource production
of x86 microprocessors, or to otherwise partner with third parties to
This is where the NY Times blogs things the FTC is really helping NVIDIA enter the x86 market. If you read it closely though, it is saying that Intel should not be able to dictate "change of ownership" or "outsourcing" of x86 licenses. It is definitely worded in a vague enough manner to cause concern for Intel, but it would seem to address the debate that raged between Intel and the separate AMD/GlobalFoundries partnership. That was settled in the $1.25B settlement between the two companies already.
What this could create is the ability for NVIDIA to purchase VIA and legally keep its existing x86 license - something we have theorized and debated at PC Perspective for some time.
The next question: does NVIDIA have the ability or desire to make an x86 part? Vance says it "takes about four years and close to $1 billion" to make an initial x86 processor but this seems like a very optimistic and obtuse estimate. Could you make an x86 processor in that time frame and budget? Sure. Would it be worth a damn? Uhh, questionable to say the least.
Vance also points to the many Transmeta employees currently working at NVIDIA that would seem to have no other purpose than to make and/or test x86 parts. That is interesting evidence to see say the least - and the fact that NVIDIA's R&D budget has increased about 2x in two years.
What would NVIDIA gain by making an x86 processor though? Ask VIA how much good it has done in the market to have a (very competent actually) x86 processor: the company continues to struggle to stay alive and compete with the various Intel offerings like the Atom and CULV lines. But then again, maybe the new rules and regulations being presented by the FTC would give these kinds of oppositions a boost.
In truth, NVIDIA and Intel (and AMD) know that even though the CPU is still a vitally important part of a modern PC, the GPU is becoming more and more relevant in terms of what it can do and how valuable it is to overall processing. Just look at the world at HD video and even Adobe Flash to see those trends at work. And even though the project was a failure initially, the fact that Intel has been spending so much money and time on Larrabee proves that computing is going in a highly parallel direction. Since we know that GPU architectures have been moving in the direction of programmable design each subsequent generation, the question is whether or not the standard x86 CPU will remain as dominant and important as it is today long enough for NVIDIA to even WANT to have an x86 offering in its portfolio. Projects like NVIDIA's Tegra (that utilize licensed ARM CPU cores rather than x86 cores) might be the true direction to the company's survival.
NVIDIA needs to tread carefully in the next few years and hope that it can fend off the battle with AMD's GPU team long enough to even consider expanding into x86 territory. And whether or not the FTC is really interested in creating that kind of environment, in my mind, is still up for debate. Thanks goes to Ashlee Vance for bringing up an interesting debate once again, if nothing else.
Intel and Advanced Micro Devices have set out on a path to combine
computing and graphics functions on a single piece of silicon, instead
of selling two different chips to handle these functions. That’s nice
for Intel and A.M.D. because they both have main (computing) chip and
graphics chip expertise.
Nvidia has been looking like the odd man out in this scenario, since
it only makes graphics chips. As it stands today, Nvidia does not have
a license to make the x86 computing chips it would need to compete
against Intel and A.M.D. in this new era of hybrid products.
Parse through the legalese in relief measure No. 17 and you find the
F.T.C. opening the door for Nvidia to obtain an x86 license from Intel.
That’s big news, since Intel had very little motivation in the past to
grant Nvidia such a license. With an x86 license in hand, Nvidia could
go the hybrid route as well, and the world would end up with not two
but three chip powerhouses going at it for mainstream computing