Ok it’s my turn. Warning long post.
When talking about AMD Vs INTC as investments most of the arguments revolve around three things, production, design, and marketing.
In the production area, the once invincible INTC processing might, seems to finally be getting some competition from AMD. The combination of SOI and DSL technology starting with the “E” stepping appears to be reaping large rewards not only in greater speed, but lower heat also. So far INTC seems unable to respond with anything that has the equivalent effects. This not only applies to the P4, but also to Dotham where speed increases seem capped. Perhaps Yonah and 65nm will solve some of the problems, but for now AMD’s combination of design and process seems superior. The only AMD question remaining concerns the bin/yields of the process? From the increasing availability of 90nm parts it seems like any problems are being handled, but we’ll have to wait for the CC to find out how well AMD is meetings its’ goal of near total 90nm production by the start of H2. One other thing, there have been rumors, form more than one source, that AMD is reducing the stepping gap. That is that AMD will reduce the normal stepping lag of a year or so that INTC normally enjoys before AMD catches up. If so, given the apparent more than competitive ability of AMD’s 90nm process, this could be the thing that finally levels the process playing field.
On the design front as long as INTC remains attached to the FSB they will be limited to increasing cache with ever reduced bang for the buck. From the reviews I’ve read true/usable bandwidth also is a problem that faster memory won’t help either. Not having something like HT and an on-board memory controller also seems to be keeping INTC from keeping up with AMD. With OEMs now able to produce fan-less AMD products and the apparent large energy advantage the latest 90nm products are displaying it appears that the 90nm process will be more than able to compete with anything INTC can come up with, even at 65nm. Then there’s the matter of Yonah not having 64 bit capabilities and the Frankenstein nature of bolting 2 processors together to create Smithfield. These products hardly sound like the products INTC used to come out with. Worse, all INTC seems to be able to do is follow AMD’s lead. For the time being INTC seems more brain dead than Terri Schiavo. I don’t know, but I’m wondering just how many more design efficiencies can be incorporated into Yonah without some sort of process changes to compensate for the increased number of transistors? Anyway, the power advantages of Dotham seem ephemeral with Turion being more than a match in processing power and competitive in battery life. We should know more in the next month. Again, the combination of design and manufacturing prowess seem to be tilting the playing field AMD’s way.
As far as marketing goes, this is where the biggest chinks are showing up in the INTC armor. The regulatory events in JP and EUROland are just the expression of a greater reality. Namely, that being an INTC only shop is not the best way to run your business anymore. As usual these regulatory things trail the market place and only happen when someone starts complaining about the situation. With more demand for AMD products INTC only shops are starting to feel the heat of lost sales, or worse, lost customers. When you have a monopolist with an inferior product and better products are available, the tendency on the part of the buyers is to cheat. This has happened even with such INTC stalwarts as DELL, which has sold AMD products to demanding customers to avoid losing their business. Now take this to the screwdriver shop level where INTC compliance is less monitored and the tendency to cheat is irresistible. What’s been happening over the last couple of years is that as the mom/pops and tier2/3 manufacturers have sold more AMD stuff, less INTC stuff has been sold by the large OEMs. I know the INTC goons are going to ask for proof, and I don’t have any, but to think otherwise is to go against human nature. The result of this shift has been the increasing willingness of the big OEMs to defy INTC and take on more and more AMD projects. As far as I can tell the only things limiting this process have been the inability of AMD to guarantee sufficient product, something ameliorated by 90nm but not fixed until fab36, and the time for the snowballing of the defection process to get going. The next year should be a real trial for Otellini and the marketing dragoons.
In order to compensate for the changes resulting from the above INTC has decided that the answer is to create a more insular, vertically integrated company that can produce ever more standardized products at ever-reduced costs. Not a bad plan, but for platformization to work in the low-end of the market INTC must have the lowest costs, and in the high-end it must have the best products also. While I could see INTC having the lowest costs, though it’s getting harder to do so, I have a very hard time imagining INTC having the best products. This isn’t a hit against INTC in particular, but more a description of how mammoth companies operate. Basically any new idea is a threat to someone, and the number of someone’s threatened increases tremendously in a company with the number of levels of management INTC has. So really thought provoking/mind bending/structure changing innovation in-house is sort of frowned upon as it makes waves. A killer idea will either be ignored or maybe even get you in trouble. Not too much incentive there. At best introducing a new idea is beyond arduous and the introducer can expect all sorts of reasons to be tossed about as to why it won’t work. Look at INTC’s attempts at creating an onboard memory controller for example, it can’t be done, three teams said so.
Well if INTC isn’t going to create in-house, what about out-house. The best indication of how that might go is to look at how deworsification has worked. Need I say more?
To me it comes down to whether 100 or more mice who are having their existence threatened by an elephant can combine and defeat the beast. In that sort of environment I would expect the creative juices to be flowing non-stop, something we’re beginning to see more of in products from NVDA, SUN, ATI and a host of smaller companies now creating AMD specific products. Even 2 years ago this level of commitment/coordination was all but unthinkable, but times and pressures are forcing more and more of the mice to choose between picking between the elephants dung for morsels or partnering with AMD and getting the whole enchilada. To me that seems like a no brainer.
