Intel Will Pay Nvidia $1.5 Billion to "Maintain Patent Peace"
Intel has agreed to pay Nvidia $1.5 billion to settle their long-simmering legal dispute that had been set to go before a Delaware Chancery Court in December.
Intel will pay Nvidia in five annual installments beginning Jan. 18, and in return will receive full access to Nvidia’s full range of patents, which had been part of the dispute. Nvidia will retain use of certain Intel patents that had also been in dispute.
“This agreement ends the legal dispute between the companies, preserves patent peace and provides protections that allow for continued freedom in product design,” said Doug Melamed, Intel senior vice president and general counsel, in a statement.
The fight had been over the terms of a 2004 agreement under which Intel granted Nvidia access to some of Intel’s technology for use in its chipsets, the chips that sit between the microprocessor and the graphics chip like connecting tissue. The cross-licensing agreement allowed Nvidia to make chipsets that were compatible with Intel microprocessors.
The trouble began in 2008, when Intel released its Nehalem generation of PC chips. The two companies disagreed over whether the 2004 agreement allowed Nvidia to make chipsets that would work with Nehalem chips and generations of chips that would follow. They filed dueling lawsuits in the Delaware Court of Chancery in early 2009. Intel asked a judge to rule that the agreement didn’t cover Nehalem and future generations of chips, while Nvidia sued for breach of contract, and sought to terminate Intel’s right to use some Nvidia patents that had been part of the agreement.
The larger backdrop here is the growing threat Nvidia’s chips, known as graphics processing units (or GPUs), pose to Intel’s chips in servers and supercomputers. Engineers often refer to this as the CPU-GPU debate, where Intel’s chips are referred to as CPUs.
GPUs are common in most PCs, and usually handle the processing required to make games look good and run smoothly, working in concert with the CPU.
Since GPU chips do certain kind of math known as a floating point operation a lot faster than a CPU, they’re increasingly being used in systems that Intel has traditionally considered its primary domain: Heavy-duty financial modeling (oil and gas exploration is a good example). They’re also making a huge splash in the rarefied world of supercomputing: Nvidia GPU chips are being used in three of the top five systems on the elite Top 500 list of the world’s most powerful supercomputers. And as we all saw at CES last week, they’re starting to show up in tablet and other PC-like devices running Windows with the full support of Microsoft.
The dispute between them, which effectively put Nvidia out of the business of making chipsets that were compatible with Intel chips, certainly hurt. Though for Intel’s part, losing the Nvidia patents in question could have conceivably hurt its new Sandy Bridge chips, which combine a GPU and a CPU into one single component. Intel formally launched Sandy Bridge at CES last week.
And as recently as last week, sources familiar with the matter were saying that a new trial date was scheduled for February. Nvidia CEO Jen-Hsun Huang was careful not to directly answer a question about that from Mobilized’s Ina Fried in an interview at our D@CES event last week:
Intel and Nvidia had mysteriously withdrawn the case from the court’s calendar days before opening arguments were set to get underway on Dec. 6. Bloomberg News then reported that settlement talks were underway, though by mid-December there were signs that those talks had stalled, and sources said that a new trial date had been agreed to. That was until today, when sources at both companies started to drop hints that news was imminent.