Microsoft 為推動 Hybrid Intelligence 改善 Windows 11

After years of neglect, Windows users are experiencing a renaissance of attention and commitment from Microsoft. And while Windows enthusiasts are right to be cynical and wonder when this magical moment will end, I have good news: That won't happen anytime soon.

The reason I know this is that I've now confirmed with two sources the real reason that Microsoft is working so hard to improve Windows 11 this year.

Back in June, I wrote that AI--or, more specifically, local AI, agents, and hybrid AI--was almost certainly behind all the Windows 11 improvements we've seen in 2026. My theory--which is more of an educated guess kind of thing, given the evidence--was that the enshittification of Windows 11 had left this OS platform in tough shape technically and reputationally, and that was proving to be a serious issue because Microsoft strategically needs hybrid AI, what it calls Hybrid Intelligence, to work, and work well.

If you think about Microsoft in terms of eras, we have moved past the cloud era and into the AI era. The cloud era was hugely successful for Microsoft, of course, but this was also the trigger for Windows 11 enshittification: Because Windows 11 couldn't contribute much to Microsoft's cloud ambitions, Microsoft neglected Windows 11 during this time. This was the final straw for the top-level talent at Microsoft that had always pursued the next big thing, meaning the products and services with the most career advancement potential. And that left Windows 11 in the hands of B-teamers at best, with no leverage among the senior leadership. It was almost an embarrassment because it's been a consistent revenue maker and Microsoft needs Wall Street to focus on its new growth engines.

The AI era is different and not just for the obvious reasons around capital expenditures, the datacenter build-outs the public doesn't want, and the financial shenanigans Microsoft, like others, uses to hide all the spending. Unlike with cloud computing, Windows can play an important role in the AI era because cloud AI workloads are punishingly expensive. The only way that Microsoft can lower its costs and make sense of AI financially is to bring as much AI as possible to customers' local devices in the form of local AI models, local AI agents, and hybrid AI orchestrators that use policy-based rules to select local AI when possible and cloud AI only when needed.

And that's quite a thing. Based on my weeks of experiencing trying to get local AI working effectively on the new breed of Nvidia RTX Spark computers has taught me a few things, and not just about entirely new levels of frustration I never knew existed. Local AI isn't ready yet, not for mainstream users, and there's a lot of work to be done before this will make sense as an integrated, normal feature of Windows 11. This would never work with Windows 11 in the state it was in a year or so ago. And not just because the platform had been ignored and even actively undermined for so...