Something New is Coming from Windows, But … What? ⭐

Last week, Microsoft hosted an event to give us a look at what's new from Windows, Surface, and Nvidia. It meandered drunkenly from topic to topic for 90 minutes and didn't feel in any way cohesive. And it's taken me three viewings to fully understand what it is that Microsoft was trying to communicate.
What lies beneath
To be fair, this isn't my fault. To fully understand this event, called Something new is coming from Windows, you have to be up-to-date on a wide range of topics, not all of which are interesting to the same people.

There's Windows, of course, which just underwent a year of Microsoft fixing pain points and reversing many functional regressions in Windows 11 after five years of ignoring them.

There's AI broadly, which today happens mostly in the cloud and is thus expensive.

There's local AI, which is always on the cusp of becoming important but never seems to get there.

There's the evolution of local AI-capable PCs from AI PCs to Copilot+ PCs and now to RTX Spark computers.

There's hybrid AI, some mix of local and cloud AI that requires intelligent orchestration, powerful local AI, and some policy-based understanding of when it is OK to use paid cloud AI.

There's Surface, an unprofitable PC business that Microsoft keeps going for some reason, but now reduced to two basic model lines, Laptop and Pro.

There's the broader PC ecosystem, which follows Microsoft down every rabbit hole it can come up with, more often than not running into a dead-end and having to recalibrate.

There's Copilot, Microsoft's unloved AI assistant, and how it may or may not make sense in a multi-model hybrid AI world.

There's Microsoft AI, which seeks to displace Microsoft's reliance on OpenAI and Anthropic frontier cloud models and is racing to build out families of both cloud and local AI models.

There are AI agents which, like local AI, are always on the cusp of becoming important but never seem to get there despite being positioned as inevitable.

And there are the various programmatic and interactive ways in which developers can access AI capabilities of all kinds, using a growing range of local, cloud, and hybrid models and a growing range of hardware platforms and PC form factors. (Which, yes, can easily be broken down into many sub-topics.)

Beyond these high-level topics, understanding last week's Microsoft event also requires some perspective of the past, in this case what Microsoft announced over time about each topic, whether at major conferences like Ignite and Build, smaller events like this one, or via more traditional press releases and blog posts. And it helps to understand what the rest of the industry is doing across these topics, whether it's Big Tech companies like Amazon, Apple, and Google, Big AI companies like Anthropic and OpenAI, Little Tech and Little AI companies of all kinds, or anything and everything else in-between.

I don't know about you, but this all feels daunting to me sometimes...