True about the laptop and Apple markets, however AMD just signed up the biggest customer of all, Dell. They already have IBM and HP as customers. In the segment of the market where the margins are very high, AMD is making major inroads.
Intel spent billions on the Itanium and ended up having to issue processors based on AMD's x64 instruction set (the Yamhill processor) to keep up.
This reminds me of when IBM was king, and Compaq shipped the first 386-based PCs. Intel may have the market now, but that doesn't mean so for the future. Oftentimes decisions made for the long term a few years back come back to bite you in the future. This is esp. true for Intel. While Conroe and the new chips may be excellent chips, they may be a stopgap that shows less innovation in future designs.
Intel's Israel division basically handed them the Centrino, Core, Core Duo, and Core 2 Duo chips because they utilized sound engineering practice and built on proven technology, while the rest of the company pushed NetBurst, which was not so sound
.
What I believe that TW is saying is that Intel's misstep with Netburst, which lasted approx. 7-8 years, and their dalliance with Itanium, which has lasted much longer, may have shifted valuable engineering resources away from much more practical long-term projects.
AMD has come in with a long-term plan for x86-64, and now has the backing of the major hardware and software vendors, including Microsoft, Red Hat/IBM (since IBM Global Services is providing a large chunk of their enterprise support), Novell, EMC (and VMWare), HP, Oracle, and many of the major Open Source operating systems. They also did not lock up their interconnect technology (HyperTransport) in licenses and costs.
Intel, on the other hand, has waffled incredibly on this front, esp. with Itanium, NetBurst, and the x64 extensions. CIO-level people are beginning to see this, and it is damaging.
It's going to really hurt HP first with the Itanium decisions. The current PA-RISC to Itanium transition involves a very complex migration to Itanium, as binaries from PA-RISC don't run very well on Itanium. This means that you have to re-qualify the software you utilize on Itanium, and possibly purchase new licenses for Itanium, which cost a lot of money. By the time you factor in what you pay for performance, those Opteron boxes seem a lot more attractive just on price alone. When you also realize you can utilize the same staff and tools to maintain hardware across the enterprise for your large-scale applications (SAP, Peoplesoft, Exchange, Oracle) as you do for your middle-tier and departmental applications, you also see the power of what AMD has brought to the market.
The big companies that buy truckloads of this stuff see this, and also have a much higher profit margin on what they buy. I think Dell actually loses more money than they say on their consumer PCs due to the fact that they can charge much higher margins for their business lines to make up for it. The overall cost of equipment is not just in the equipment, but how much power and manpower it takes to provide a certain amount of computing power to get the job done for the customer. AMD's solution provides a very large amount of power at a very low cost per watt for the large-scale applications, and scales out to a very large scale once reserved for non-x86 chips. They also have backward compatibility which has been tested back to DOS 2.11 with the AMD64 chips, and will support unmodified versions of Windows on their chips. You can run what you already have on their chips without the cost of upgrading the applications as well.
Intel doesn't have a long-term plan for that scenario. People like tw and I, who work within those parameters, understand what is brought to the table by both parties. If a proposal from a vendor comes across my desk requiring a very large amount of hardware to be purchased, vs. a solution which is more sane in the hardware, software, and infrastructure requirements, the latter will always win.