First hybrid cars, now hybrid chips?
Grose started by explaining how manufacturing fits into the company's new three-pronged philosophy. On two of those three points, it sounded like a stretch to me. "Energy-efficient processing" is tenet No. 1, and fabrication technology can certainly help there. AMD is aiming for "optimal performance per watt, per square mm" of silicon wafer by working on "transistor-level" improvements.
I don't mind being told more about AMD's asset-light tactics, cross-company research partnerships, and focus on highly integrated products, but how these things fit into headings like "ultimate visual experience" and "affordable Internet access" -- the other two philosophy pillars -- is beyond me.
Enough about the presentation structure. The meat of Grose's message began when he launched into the nitty-gritty lowdown on the "unique hybrid manufacturing model" of the "new AMD."
In short, it's a combination of AMD's own manufacturing facilities, foundry partners available on demand, outsourced assembly and testing operations, and those research partnerships I mentioned above.
The chipper Sunnyvale chipper has pumped billions into building and maintaining those fabs, and of course, it only makes sense to squeeze every ounce of potential capacity out of them. The foundry relationships got a serious boost when AMD and ATI became an item, because that's the only manufacturing technology the Canadian graphics specialist had ever known. Now, AMD wants Chartered Semiconductor
Specialists like Amkor
Ramp me up, Scotty!
Grose was happy to report that ramp-up times have improved with every new generation of process technologies. The 90nm and 65nm lines at Fab 36 started off at what AMD calls mature yield levels, whereas it took months for the 90nm and 130nm processes to reach that output goal in Fab 30. Hopes are high for the 45nm process to get off to a good start, too -- but not until the middle of 2008.
The details of how this will be achieved might not interest everyone here, so I suggest a different source to fill your ultra-low-K-dielectrics, strained-silicon, and immersion-lithography hunger.
Pilot lines for 45nm chips are already up and running in the Dresden plant, and the equipment needed to start converting real production lines to this process is on order for delivery early next year. AMD's cash crunch forced a delay of that conversion, even with new credit facilities in place.
So what do I make of this nuts-and-bolts presentation? Well, it's not too far from what Texas Instruments
But it's still the same basic philosophy, one that has improved TI's margins considerably while reducing its exposure to market changes. Anecdotally, I'm told that the heavy outsourcing has also led to longer order fulfillment lead times for TI, and that some customers would have preferred the old ways, even at higher product costs.
For now, AMD's main issue is to get its Barcelona chip out the door on time and in volume. Then there's the 45nm conversion, and the output drop that will come when production lines are getting new equipment installed and tested. Focusing on order times will have to wait a bit. Finally, assuming that AMD makes it through this difficult time, the next couple of process changes appear to be on track.
That wraps up our report from this presentation with AMD, but stay on the lookout for more "Fool on the Street" reports, bringing you juicy information that only the analysts have heard.
- Fool on Call: AMD Might Make It After All
- Texas Instruments Hangs Loose
- So You Think You Can Dance, Intel?
Intel is a Motley Fool Inside Value recommendation.
Fool contributor Anders Bylund is a hopeful AMD shareholder who holds no other position in any of the companies discussed here. You can check out Anders' holdings if you like. Foolish disclosure hears all and sees all.