With “eye candy” – snazzy media processing – being the most important aspect of personal computing these days, Intel can ill afford to be at war with graphics chip maker Nvidia. Not while AMD is fielding decent CPU chips and arguably better GPU chips, and then merging them into its Fusion products – a hybridization tack that Intel is itself making with its just-announced “Sandy Bridge” family of Core processors.
And so, even as Nvidia said last week that it is jumping into the CPU racket and will field its own ARM-derived processors for servers and PCs in 2013, the lawyers at Intel and Nvidia were working feverishly away on a cross-licensing deal.
Under the agreement announced today, Intel is shelling out $1.5bn to have continuing access to all of Nvidia’s patents, and Nvidia retains the right to use certain of Intel’s patents. The cross-licensing deal has a six-year term and gives Nvidia access to unspecified Intel technologies and excludes the patents relating to CPUs, flash memory, and certain chipsets.
Nvidia had an existing six-year licensing agreement with Intel for certain technology that was set to run out on March 31, 2011, which is why Nvidia was eager to seal a deal with the chip maker. Nvidia needs access to Intel chip and system technology if it is to develop GPUs that are compatible with Intel’s CPUs and chipsets and get them to market in a timely fashion.
As part of the agreement, Intel and Nvidia are settling all outstanding lawsuits.
The way the bookkeeping works, less than $100m of the dough coming from Intel will be booked as settlement of legal disputes, with the remaining bag of cash coming to Nvidia in six equal installments. The graphics chip maker says the deal will add about $233m in operating income and about 29 cents per share of net income to its bottom line. So, if you bought Nvidia shares last week on the ARM news, you just got a bonus.
This agreement signals a new era for Nvidia,” Jen-Hsun Huang, the company’s president and CEO, said in a statement announcing the settlement of hostilities between Nvidia and Intel. “Our cross license with Intel reflects the substantial value of our visual and parallel computing technologies. It also underscores the importance of our inventions to the future of personal computing, as well as the expanding markets for mobile and cloud computing.”
The settlement agreement between Intel and Nvidia comes four months after the US Federal Trade Commission settled an antitrust lawsuit that it had brought against Intel in December 2009 for a laundry list of anticompetitive practices, including trying to lock AMD out of the CPU market and trying to smother upstart GPU maker Nvidia just as these adjunct processors were starting to displace workloads that might have otherwise run on CPUs.
The FTC said last August that its consent decree with Intel curbed the “disturbing behavior” of Intel, and included measures that prevent Intel from paying PC and server makers to not use non-Intel chips and compels Intel to maintain a PCI-Express bus on its chips for the next six years and make that specification open to third parties. The FTC consent decree spans CPU, GPU, and chipset technology.
The cross-licensing agreement follows a similar deal that Nvidia inked in December 2004 with Sony for the graphics used in the PlayStation 3 game console. In a conference call explaining the deal, Huang said that this deal had generated $500m in licensing and royalty revenues.
The Intel-Nvidia agreement does not cover any of the patents and intellectual property covering Intel’s chipsets and processors, and this is so by design. “We have said for a long time that we have no intention of building chipsets for Intel processors,” Huang said. “We have no intention of building x86 processors.”
But, Nvidia has every intention of building GPU co-processors that hook into Intel and AMD processors, since that will remain a very large portion of its revenues for the foreseeable future.
The best possible thing that could happen to AMD would have been for Intel and the FTC to not have settled last summer, and for Intel and Nvidia to not have come to terms on licensing, with all parties then remaining embroiled in legal battles. Now, Intel will come at AMD’s Fusion APU hybrids with its own Sandy Bridge Core chips and then tag team with Nvidia for those customers who want or need more powerful discrete graphics controllers or GPU coprocessors to do number-crunching work. ®
This article originally appeared in The Register.