This seems to track with a recent blog post from MAI regarding their approach to pre-training their first Frontier level model which seems promising but does seem to have a focus on scaling.

https://microsoft.ai/news/frontier-pre-training-at-microsoft...

Yes, but also in combination with the not-borderline-and-actually insane growth in commits and load on the platform due to AI.

this is often the argument that is trotted out, but i don't see any evidece of that.

It has been reported new user volume has nearly doubled, and commit volume 10x+ - that could certainly have an impact, but "Azure" and "scale" in my career are words that have never really gone together. Azure Cloud is the only 'hard no, will not interview' of any cloud provider that's listed on a job posting, and I've worked with just about all of them by now I think. Nothing even comes close to such a miserable experience.

The thought of migrating something at the scale of github to azure gives me the cold sweats.

I don't think that's actually the case. There were some rumors flying around about this but in their recent event they actually referred to Copilot+ PCs as generally the line targeting more casual users with support for less powerful local inference. And the new devices based on Nvidia RTX Spark (and likely the more powerful solutions from AMD, Intel, Qualcomm with large unified memory) as a "Builder" class targeting developers and heavy local inference users. They also revealed a quantized version of their coding model designed to run on these devices. So it does seem they may be somewhat working from the bottom up building smaller models or focusing on capable local inference and balancing with more powerful frontier model access. This is a lot of marketing speak but covers much of what they revealed.

https://blogs.windows.com/windowsexperience/2026/10/07/build... https://github.com/microsoft/windowsML

This is actually the case. Windows ML (https://github.com/microsoft/windowsML) is the inference framework wtih vendor agnostic support for inference on CPU, NPU, and GPU. They also announced quite a bit more including an isolation solution with MXC. There's a lot of marketing fluff in the below link but it covers the recent announcements. https://blogs.windows.com/windowsexperience/2026/10/07/build...

They did also compare to Sol and the comparisons are still favorable. However, they were most favorable comparing to Opus 5 because of the cost judging by the charts.

Newer Intel Panther Lake chips perform the same or very similarly on battery as they do when plugged in - as do Qualcomm chips. However, I don't know where they're seeing "consistently beat on performance and battery life by Intel Windows laptops". Multi-core performance can definitely beat M5 in some configurations but single core performance is still fairly far behind and battery life is comparable again depending on the exact specifications and design of the laptop. I've seen analysis showing M5 is still the perf/watt king though regardless of configurations.

Sorry, poor wording. It's the Neo that's on the bottom in many reviews, but the M5 is regularly beaten by Windows laptops. My usage of "consistently" was in the sense of regularly beaten, not in the sense of always beaten.

So…inconsistently?

So consistently beats the M5 on something, just not on everything.