In a world where ARM (ARMH) is now a viable alternative instruction set architecture in the broad computing market (tablet, notebook, desktop) and, indeed, some could say the dominant architecture, why doesn't Intel (INTC 1.77%) put an end to AMD by simply buying it?

The pros
What would Intel get out of such a deal? Right off the bat, a number of things come to mind:

  • Hundreds to thousands of graphics patents that could potentially be enough to keep Intel from needing to license NVIDIA's (NVDA 3.71%) patent portfolio
  • Solid graphics IP that could become even more potent if given Intel's R&D and manufacturing leadership
  • 100% market share in the x86 PC market, and the last dregs of the x86 server market
  • A discrete GPU business that could, once ported to Intel's factories, drive some decent revenue growth
  • Intel would immediately inherit all of the semi-custom designs that AMD has in the pipeline

The cons
How about some of the cons? 

  • Intel is likely on the verge of internally developing GPU IP that is more competitive with AMD's, although it's not yet clear how the next-generation Broadwell and Cherry Trail parts from Intel performs in graphics
  • Intel would inherit a modest net debt load, as well as a bunch of PC/server CPU inventory that it would need to continue selling 
  • There still may be antitrust issues, despite the growing prevalence of ARM-based products in the compute continuum
  • AMD is not a meaningful competitor in HPC, and buying AMD's FirePro lineup would still not position it very well against NVIDIA's Tesla
  • AMD would not likely be accretive to free cash flow and, since there is so much overlap, the company would be hit with restructuring charges as it lays off employees
Buying NVIDIA Would Probably Make More Sense
While also unlikely, it really does seem that, other than the shut-AMD-down-in-PCs angle, Intel could get everything that it needs vis-à-vis graphics from NVIDIA, and would even be able to completely dominate HPC with NVIDIA hardware/software. If NVIDIA, under Intel's wing, were to build its GPUs on Intel's in-house process, then Intel/NVIDIA could potentially increase its GPU market share to such a degree that it could potentially put an end to AMD/ATI, as AMD would not be able to compete on performance or power consumption.

It also helps that NVIDIA's balance sheet is clean as a whistle, with billions in net-cash. Further, ex-Tegra (which is losing money), NVIDIA's cash flow (which is already robust) would look even better under Intel's wing. The main thing blocking such a deal, in this Fool's view, is that NVIDIA CEO Jen-Hsun Huang seems dead set to leading NVIDIA to glory on its own (which, if successful, could generate more value for shareholders than an Intel buyout would).

Foolish bottom line
The real reason that Intel is unlikely to buy AMD is simple -- there are better buys out there, including Intel/AMD competitor NVIDIA. Further, with Intel continuing to push AMD out of the x86 business with its own products, it seems that the idea of buying AMD is only compelling from a graphics perspective, or perhaps to keep AMD's x86 business out of the hands of a competitor (although the cross-licensing deal between AMD/Intel would make this difficult for a competitor, anyway).

AMD doesn't really seem to be a compelling take-out target in a world where ARM is so pervasive and where everybody is investing heavily in graphics. There's still probably a good amount of value in AMD's GPU IP, but the value will diminish as others ramp investments and AMD cuts R&D. In short, AMD needs to build value organically to drive a meaningful share price increase.