Intel’s main competitor today isn’t AMD, it’s Arm and its constellation of licensees, such as Apple, Qualcomm, Samsung, MediaTek, and NVIDIA. Over the decades that followed the advent of the iPhone, Arm graduated from smartphones to PCs (eg: Snapdragon X, Apple M3), and servers (eg: Ampere Altra and NVIDIA Grace). Intel seems to have had the sudden realization that Intel Products should open up in the same way as Intel Foundry Services, and that just as IFS in Pat Gelsinger’s words aspires to be the “TSMC of America,” x86 should aspire to be the “Arm of America.” The only way to go about doing this is to adopt an IP licensing model similar to that of Arm, where third parties are licensed the x86 machine architecture, and should pay Intel royalties based on their chip volumes, and other factors such as CPU core counts. Much like Arm, Intel could set up separate licensing models for SoC designers who want Intel’s various IA cores as IP blocks, or just the x86 license to design their own cores, like AMD does. Since Intel is a founding member of the UCIe alliance, it could even offer up Compute tiles as chiplets.
Intel Realizes the Only Way to Save x86 is to Democratize it, Reopens x86 IP Licensing
Back in the glory days of x86 in the 1990s, you could buy an IBM PC-compatible x86 processor from not just Intel and AMD, but also the likes of Cyrix, IDT, Transmeta, and NEC. Competition among the various x86 licensees propelled innovation through the first half of the 32-bit era, but toward the end of the century, and with the Y2K frenzy looming, the PC processor market saw a slew of consolidations, such as Cyrix and IDT (later Centaur) being acquired by VIA Technology. As of 2000, only two companies made high performance x86 processors, and processors for servers—Intel and AMD, with VIA Technology limiting itself to the entry-level PC market. Then came along Arm Cortex 32-bit, graduating from the embedded computing market to client computing, driven by smartphones.