> By March 1989, however, Hugh Barnes—now Compaq’s vice president of engineering—started to notice that Intel’s best chip people were being reassigned to other teams. Some quiet investigation revealed the cause. Sun Microsystems, one of Intel’s rivals, had announced chips based on a new design approach called reduced-instruction-set computing (RISC). For Intel, this presented a threat to its higher-end, large computer and mainframe markets. It was now shifting to focus on that threat instead.

> At a hotel room in Silicon Valley, in April 1989, Canion and Gates met with Andy Grove and Intel chair Gordon Moore to try to persuade them to stick with 486 development. After considerable back and forth, Intel reversed course. The new chip launched in late 1989.

Could this be the moment that forever saddled us (and Intel) with the cumbersome legacy of x86? It seems like a great cultural win for PCs in the moment, but in hindsight this decision almost feels backwards somehow.

I'd say no. We would have had the Itanium drama 10 years earlier. The i860 had some flashy features in it (integrated graphics, SIMD) but also the same Achilles' heel as Itanium of compiler-directed pipeline ordering. That made it much slower than other processors and any analyst of the time would have recommended staying on x86 for the performance and not having to change all your software. So AMD and Cyrix sell more chips and Intel has to backpedal to avoid completely losing the PC market.

It sounds like they are talking about the i860 which was launched and failed in the market. While it is possible that it might have had more success if Intel chose to abandon the x86 line with its launch, I think it is more likely that it would have accelerated AMDs move from a second-source x86 manufacturer to designing its own x86 chips.

The lesson of the article is that backwards compatibility is everything. By sticking with the 486, Intel succeeded fantastically. They even eventually succumbed to building AMD64 compatible chips.