Technology Tussle: Understanding The Intel/NV/FTC Triangle
At the time of this writing, the FTC's investigation into Intel's alleged monopolistic abuses is on hold as the government attempts to negotiate a settlement with the CPU and chipset manufacturer. If these negotiations don't result in a deal by July 22, the case returns to court, with arguments currently scheduled to begin on September 15. Intel is no stranger to these sorts of lawsuits; between AMD and the EU, the CPU giant has been battling such allegations for years.
The lawsuit between NV and Intel, however, rests on different points than the AMD/Intel allegations. Here, the battle is over whether or not Intel's already-negotiated agreements with NVIDIA give the latter permission to produce modern chipsets for the Core i3/i5/i7 processor lines. CNET has an excellent piece detailing the various technologies in question, with additional commentary by Richard Kanter (whose article on PhysX optimizations we discussed here).
To date, the Zune HD has been the highest-profile Tegra launch.
Even if the FTC held that Intel's Atom price structure was an abuse of power and validated NVIDIA's claim to possess a license to produce Core i7 chipsets, it's not clear NVIDIA would materially benefit as a result. The integration of CPU and GPU makes technological sense for everyone; the term 'chipset' is itself becoming obsolete.When NVIDIA launched the GeForce 9200-9400 series, it redefined its chipsets around the integrated graphics solutions they offered. That worked well for the Core 2 series, when OEMs were choosing between Intel graphics <i>or</i> NVIDIA graphics. In the future, OEMs will choose between Intel graphics or Intel graphics <i>with NVIDIA on top.</i> That's worked well for Atom, where ION has been able to cut clear performance distinctions, but NVIDIA may not be able to offer a low-end notebook GPU (think Core i3) that justifies the additional expense.
More Than Chipsets
If we assume that this lawsuit is part of NVIDIA's long-term strategy rather than a corporate axe-grinding, then there's something else afoot. It doesn't make sense for NVIDIA to put a major push behind a desktop/server chipset division; the industry is moving towards a level of integration that NVIDIA can't supply. Similarly, ION is a great idea, but the long-term viability of the segment is completely tied to Intel, and Santa Clara doesn't like to share. Since NVIDIA's already dealing with the fallout related to one Intel licensing issue, it's unlikely that Team Green would bet the farm on another Intel-dependent product.
NVIDIA's Tegra Design.
NVIDIA is currently pumping resources into two potentially lucrative markets. First, there's Tegra/Tegra 2, and the burgeoning handheld/UMPC market. Second, there's the realm of scientific computing, GPGPU, and the Tesla product series. NV has traded shots with Intel recently over just how much better GPU's are compared to CPUs when it comes to certain kinds of operations; the company has a keen interest in positioning GPUs as the 'go to' solution for parallel processing.
If we had to bet, we'd bet NVIDIA is interested in Intel IP that could improve one or both of these product divisions. Even if the FTC and Intel don't reach an agreement by Thursday, we wouldn't be surprised to see all three parties come to a mutually acceptable compromise before the trial actually begins.
The lawsuit between NV and Intel, however, rests on different points than the AMD/Intel allegations. Here, the battle is over whether or not Intel's already-negotiated agreements with NVIDIA give the latter permission to produce modern chipsets for the Core i3/i5/i7 processor lines. CNET has an excellent piece detailing the various technologies in question, with additional commentary by Richard Kanter (whose article on PhysX optimizations we discussed here).
To date, the Zune HD has been the highest-profile Tegra launch.
Even if the FTC held that Intel's Atom price structure was an abuse of power and validated NVIDIA's claim to possess a license to produce Core i7 chipsets, it's not clear NVIDIA would materially benefit as a result. The integration of CPU and GPU makes technological sense for everyone; the term 'chipset' is itself becoming obsolete.When NVIDIA launched the GeForce 9200-9400 series, it redefined its chipsets around the integrated graphics solutions they offered. That worked well for the Core 2 series, when OEMs were choosing between Intel graphics <i>or</i> NVIDIA graphics. In the future, OEMs will choose between Intel graphics or Intel graphics <i>with NVIDIA on top.</i> That's worked well for Atom, where ION has been able to cut clear performance distinctions, but NVIDIA may not be able to offer a low-end notebook GPU (think Core i3) that justifies the additional expense.
More Than Chipsets
If we assume that this lawsuit is part of NVIDIA's long-term strategy rather than a corporate axe-grinding, then there's something else afoot. It doesn't make sense for NVIDIA to put a major push behind a desktop/server chipset division; the industry is moving towards a level of integration that NVIDIA can't supply. Similarly, ION is a great idea, but the long-term viability of the segment is completely tied to Intel, and Santa Clara doesn't like to share. Since NVIDIA's already dealing with the fallout related to one Intel licensing issue, it's unlikely that Team Green would bet the farm on another Intel-dependent product.
NVIDIA's Tegra Design.
NVIDIA is currently pumping resources into two potentially lucrative markets. First, there's Tegra/Tegra 2, and the burgeoning handheld/UMPC market. Second, there's the realm of scientific computing, GPGPU, and the Tesla product series. NV has traded shots with Intel recently over just how much better GPU's are compared to CPUs when it comes to certain kinds of operations; the company has a keen interest in positioning GPUs as the 'go to' solution for parallel processing.
If we had to bet, we'd bet NVIDIA is interested in Intel IP that could improve one or both of these product divisions. Even if the FTC and Intel don't reach an agreement by Thursday, we wouldn't be surprised to see all three parties come to a mutually acceptable compromise before the trial actually begins.
