In the 64-bit server space, which is really what's relevant to this discussion, AMD was pretty much not part of the discussion until Dell (might have been Compaq at the time) and Sun picked them up as a supplier in the fairly late 2000s. Yes, Intel apparently played a bunch of dirty pool but that was mostly about the desktop at the time which the big suppliers didn't really care about.
But initial Opteron success was pretty much unrelated to 64-bit. As a very senior Intel exec told me at the time, Intel held back on multi-core because their key software partner was extremely nervous about being forced to support a multi-core world.
I'm well aware of Opteron's impact. In fact, the event when that info was related to me, was partly held for me to scare the hell out of Intel sales folks. But 64-bit wasn't really part of the equation. Long time ago and not really disposed to dig into timelines. But multi-core was an issue for Intel before they were forced to respond with Yamhill to AMD's 64-bit extensions to x86.
> As a very senior Intel exec told me at the time, Intel held back on multi-core because their key software partner was extremely nervous about being forced to support a multi-core world.
That's one way to explain it. Alternatively, one might say that FSB-based Netburst servers would not benefit much from multi-core because the architecture (and especially FSB) has hit its limitation. Arguably, Intel had no competitive product on the mass server market until 2006 and Core-based Xeon 5100 introduction. Only enormous market inertia has kept them afloat.
> In the 64-bit server space, which is really what's relevant to this discussion, AMD was pretty much not part of the discussion until Dell (might have been Compaq at the time) and Sun picked them up as a supplier in the fairly late 2000s.
That was one relatively small (servers number-wise) segment of the market. Introduction of Opteron servers and Windows Server 2003 64-bit has created a new segment of mass 64-bit servers which have very quickly taken over entire 32-bit (at that time) mass server market. That was the real market that Intel wanted for themselves with introduction of proprietary Itanium but failed to acquire it because of the compatibility issue. High-end mainframe-adjacent market segment indeed belonged to Itanium for many years after, but that wasn't the goal of Itanium. Intel wanted to be a monopoly on the entire PC&server market with no cross-licensing agreements but failed and had to cross-license AMD64 instead.
Opteron was a much bigger deal than you're making it sound. Market share was up to 25%.