Microsoft 'Hybrid Intelligence' isn't intelligent
Disclosure: Virtified did not attend Microsoft Ignite 2026 in person. This analysis of Microsoft’s current positioning is based on watching the November 17 keynote, which tells the marketing story rather than the engineering reality.
Virtified Summary
Bottom line: you cannot have hybrid intelligence without hybrid infrastructure.
Microsoft's Ignite 2026 keynote paid lip service to enterprise AI needs beyond the cloud and the desktop
Microsoft’s ‘hybrid intelligence’ covers the PC and the cloud, but not the data centres and distributed sites where enterprises will increasingly run AI. At the same time, consumption-based pricing is prompting Microsoft customers to reconsider their on-premises infrastructure. That expands the opportunity for hybrid infrastructure providers such as Red Hat and Nutanix.
Details
The keynote is further evidence of a strategic blind spot for Microsoft. Whilst cloud was a catalyst for rapid AI adoption, the AI journey is just beginning.

AI will expand beyond cloud to data centres, manufacturing plants, clinical healthcare, mining sites, remote offices, branch offices and so many other distributed locations.
The Microsoft blindspot
Microsoft is a cloud success story. Its pivot from Office and Windows to Azure was the foundation for much of its growth over the last decade. But the obsession with cloud was a strength that is becoming a weakness; assuming it can succeed in AI risks it falling behind in cloud and forgetting on-premises infrastructure.
Whilst pivoting to AI is a strategic necessity for Microsoft, the infrastructure software needs of enterprises are now two steps removed from Microsoft's focus.
A key example is Azure Local. Despite more than a decade promoting Azure Stack in various patterns, Microsoft has not bridged the gap between on-premises and public cloud infrastructure. Microsoft’s ‘hybrid intelligence’ shows little connection to Azure Local. For example, Microsoft's announcement does not mention Azure Local, Windows Server or on-premises data centres (refer to 'Building Windows for hybrid intelligence'). Perhaps these are covered elsewhere, which would hopefully be the case although it would introduce another gap.
Virtified’s own assessments highlight additional challenges for Microsoft. In the data centre use case, at this time, the Virtified Loop score for Windows Server (7) is higher than Azure Local (6). On this measure, and with a confusing array of deployment options, Azure Local remains a step backwards from Windows Server. Calling it an intelligent hybrid infrastructure strategy would be a stretch.
Microsoft has a heritage in on-premises infrastructure but is becoming less suited to the era of physical AI.
The Microsoft blindspot becomes a gaping hole if non-cloud inference expands. Virtified asserts that many physical AI use cases cannot tolerate any unnecessary network latency or unreliability; others will mandate local inference for privacy and other forms of confidentiality. Compliance will amplify the challenge and show up in everything from self-driving vehicles to occupational health and safety.
But it's not all bad news, is it?
The good
Microsoft's October 7 announcement of Microsoft Execution Containers (MXC) is a necessary capability for isolating agents in both cloud and non-cloud infrastructure. MXC is generally available on Windows 11, and Microsoft says it supports macOS and Linux through platform-specific backends.
That is the extent of it, and MXC was announced before Ignite. Unless the Ignite breakout sessions go beyond high-level messaging, our position stands. Given ongoing discussions with Microsoft over many years, Virtified has little confidence that anything will change.
The bad
Microsoft’s examples of token costs were tone-deaf. Multiple speakers made claims along the lines of '...and this only cost $X' where X varied and did not seem to be estimable in advance.
Budgeting and licensing for Microsoft has always been a challenge and it is becoming harder as Microsoft has moved from predictable costs to a 'death by a thousand cuts' wherein small incremental unit costs balloon out of control. Virtified has discussed these challenges with hundreds of Microsoft customers since the Azure era began. The shift to consumption-based pricing has upended IT budgeting and prompted many Microsoft enterprise customers to reconsider their on-premises infrastructure choices.
Implications
In Virtified’s view, Microsoft is under-serving AI inference needs beyond the cloud and the desktop. In the process, Microsoft is opening the door for other providers—with Red Hat being the most notable beneficiary. Nutanix will increasingly benefit too, as it expands into containers and AI and extends its storage hardware options. Broadcom will also remain relevant for many conservative enterprises.
Other observations: productivity dominated the keynote
The Microsoft Ignite 2026 keynote focused heavily on workplace productivity. As it has for decades, Microsoft is acting as a fast follower. Like other AI vendors, it is trying to catch up to Anthropic in agents, skills, connectors and overall experience. Agents are making the client device and productivity suite more vendor-agnostic, which would compound the erosion of Microsoft's relevance for on-premises enterprise needs.
A recent interaction highlights the productivity blindspot: a senior Microsoft executive claimed that Microsoft was the only 'full stack' solution for the needs of small and medium businesses. Virtified retorted that the claim was incorrect for two reasons. It arrogantly ignored both Google and the need to supplement Microsoft for basic business applications (examples: payroll, book-keeping).
If Microsoft fixates on catching up with Anthropic, and continues to neglect on-premises enterprise infrastructure, it will do so at its own peril.
Keynote replay
As at the time of publishing this Virtified Insight, the keynote replay was available at https://ignite.microsoft.com/en-US/home
What Microsoft said
The net content on hybrid intelligence that was sent to Virtified by Microsoft (1) was:
Windows Hybrid Intelligence
GitHub Copilot Fusion can now use local models such as MAI Code 1.1 Flash on NVIDIA and AMD-powered PCs in preview.
Adobe, Autodesk, Dell, HP, Insight, NVIDIA and PwC are Hybrid Intelligence pilot partners, with builders creating local AI experiences on Windows.
Copilot+ PCs will soon get local context, actions and models, plus read-only local PC diagnostics.
Notes:
Virtified does not rely on material sent by any vendor—and particularly not the polished 'spindoctoring' of Microsoft. We shared this portion to highlight how little top-level new capabilities are considered 'hybrid intelligence'



