Re: AMD is losing money hand over fist and there's no end in sight except in the dreams of the Dans and NiceGuys of this world.
You've been posting that since I started reading these boards in 1999, but AMD's market share and financial position, relative to Intel, keeps improving.
What do you think of Intel's sudden realization that they do need SOI after all? (We've heard, we've heard, Intel's SOI is unique in all the universe) The bottom line, though, is that Intel has been following a non-SOI low-K strategy for .13 and below, and they've suddenly decided they need a high-K SOI process, but that they won't have the details of production worked out until .09 or .065.
Meanwhile IBM, AMD, Motorola, the Taiwan Foundries - pretty much everyone except monkeys in the zoo and Intel :-) - have been working on SOI volume production techniques for several years, and have started introducing it and gaining production experience at the .13 and even some at .18 nodes.
With what certainly appears to be a glut of foundry capacity for the .13 and .09 generations, AMD is minimizing capital outlays and leveraging its production and design capabilities with partnerships. AMD continues to recognize as CAPEX costs about as much as it actually spends each quarter
Intel is continuing to spend more on CAPEX each quarter than it recognizes as costs each quarter, so even as they present Pro Forma earnings, they have been experiencing negative cash flow.
Their poor yields hold back what should be a significant cost advantage.
Intel is capacity constrained after having spent $15 Billion in the last two years, and has about 18 FABs with about 1/3 of those targeting CPUs. AMD has spent about 1/5 that amount on its FABs during the period, has 1 1/2 FABs targeting CPUs, and is having less trouble meeting demand than Intel. With 4 time the FAB space, and 5 time the FAB expense, Intel is having trouble producing 3 times the number of chips AMD produces.
Were those poor yields at Intel or AMD that you were referring to?
And the chips AMD is shipping are higher performing chips than those that Intel is shipping. Even with all of Intel's efforts at controlling benchmarks and compilers, AMD single processor and multiprocessor, workstation and server chips, are so much faster that they outperform Intel's offerings most of the time even on Intel compiler code running Intel benchmarks. (We've heard, we've heard, "next quarter Intel performance will bury AMD"). But didn't you say the same thing last quarter? And the quarter before? And the quarter before? |
