Despite the figures put forward by ARM, it's still hard to see the big picture.
We've been promised the "ARM revolution" in the PC world for many years now, and the aging x86/x64 architectures have more than held their own. In fact, it's quite simple: every ARM attempt to revolutionize the PC world has ended in failure. Every attempt? No, not quite, because the latest offensive launched by ARM with the support of Qualcomm and Microsoft seems to be bearing more fruit. This offensive dates back to May 2024, when Microsoft presented a brand-new PC family called Copilot+ PC. At the time, there was no question of doing away with x86/x64 processors, but Microsoft was clearly highlighting the new Qualcomm chips and the power of their NPU(neural processing unit) for artificial intelligence in general, and for Copilot+ in particular.
At the time of this launch, ARM computers could certainly count on a version of Windows 11 specifically designed for their specific architecture... but that was about it. For many applications that didn't exist natively on ARM architectures, an x64 emulation layer was required, with a certain loss of performance. Logically, ARM, Qualcomm and Microsoft pulled out all the stops to get most of the major publishers to port their applications to Windows 11 for ARM, so that they could be supported natively and no longer via an emulation layer. These efforts seem to have paid off, with ARM now announcing that over 100 major applications are available natively on Windows 11 for ARM, enabling users to spend " 90% of their time on native applications ". In fact, the emulation problem is almost a thing of the past.
" Almost ", because we shouldn't take everything ARM tells us at face value, as they are both judge and jury in this matter. First of all, the figure of 90% is communicated by ARM, but it is difficult to verify. Secondly, we mustn't forget that spending 90% of our time on one application doesn't take into account the number of applications we use, nor their importance and resource requirements. For example, you could imagine that the remaining 10% of your time is spent on dozens of applications on which you spend very little time. Or you could imagine that the applications in question are of crucial importance to a particular activity. Finally, it should not be forgotten that ARM does not really detail the applications concerned, contenting itself with a few examples. It is therefore possible that native ARM applications are not the most demanding and that the remaining 10% of time is spent on particularly poorly optimized applications.
Last but not least, ARM refrains from detailing the user profiles concerned, or even the number of such users. We're deliberately stretching the point, but the fact that 10 "average peasants" are happy with the applications natively available today doesn't mean that we can draw any conclusions about the uses of many millions of users around the world. In fact, ARM's progress is real, and Windows for ARM is now a tangible reality. ARM is on the right track, but it's probably too early to claim victory.
