Techman, wbmw, yes I agree that Sanders is an original! Now, in terms of technology, it is a question of where you differentiate between a new idea and the preceding ideas. If you want to look at truly new ideas, you have to go back to the first use of tools - wheel, ramp, etc. One can even make an argument that this was copied from nature!
Defining things in such a way to deprive AMD of any original ideas would make such a restrictive definition that Intel also gets caught up in the same net. Microprocessor? That was enabled by the transistor from AT&T. Transister? Well, isn't that just a manufacturing improvement on the vacuum tube? On it goes.
Using a more useful definition, a new idea is an idea which makes a significant difference in an industry and that idea is attributed to the first company which brings it to the public attention in the form of a product or standard.
By this definition, AMD and Intel both have a history of new ideas. Intel from the late 70s through the mid 90s, with new ideas now trailing off, and AMD starting in the mid 90s and continuing today. Let's look at the list:
First, there is HyperTransport. Generated by AMD, who pulled together the consortium and shepherded it to a standard implemented in PCs, communications equipment and consumer electronics.
Copper and SOI really are IBM's ideas, AMD has worked with IBM since early on in these developments.
AMD64 - this was an easy one for AMD because Intel dropped the batton. AMD defined the new modes, made a much more symmetrical register set, increased the registers, and created the migration path to move away from the biggest flaws of x86. IMHO, if Intel had done it instead, they would have been too mired in the past to define a chip this elegant - running in full 64-bit mode, AMD64 bears very little resemblence to the 8086 roots.
MirrorBit - very much an AMD solution to getting multiple bits per cell, and the benefits of this approach are becoming more apparent each quarter.
So, as we can see, AMD certainly is making very significant contributions to the industry through new ideas and followup implementations! The Intel folks here can show a lot of old Intel ideas (from the microprocessor through Itanium, co-developed with HP), but the list starts getting sparse after the mid 90s.
Lots of purchases of worthless communications companies, though (heh, heh). Any relationship between the lack of internal ideas and management's purchasing sprees? I think Barrett gave up on his own engineers.
AI
Market Data 
Markets 
