VMware to Microsoft: Cloud, On-Prem, or Hybrid

“Move everything to the cloud” was the answer for almost a decade. In episode 7 of VMware to Microsoft, Daniel Apps and I talked about why that answer stopped being good enough, and how organizations leaving VMware should decide where a workload runs.

Daniel is a Microsoft MVP based in Australia. He spent years on the VMware side before switching to Hyper-V back in 2008, so he’s argued this from both chairs. That history is exactly why his take on where workloads belong carries weight.

Four paths off VMware

Daniel sees organizations landing on one of four outcomes:

Move to cloud. Put a firm date on your last VMware renewal and start a real modernization push instead of another lift and shift.
Jump ship to Hyper-V. Re-platform onto Hyper-V or Azure Stack HCI for the cost benefit, then pick up cloud services where they make sense.
Diversify. Split workloads across platforms instead of staying fully dependent on one vendor. Some VMs, maybe some containers on something like OpenShift.
Stay on VMware. Eat the cost. Daniel was clear this is a valid choice for some organizations, not a failure to plan.

Every single one of those conversations ends up talking about hybrid, even the shops staying on VMware. They still ask about running Azure Virtual Desktop or managing part of their estate from Azure.

Cloud isn’t a destination, and repatriation is real

Daniel walked through the timeline. 2016 to 2018 was pure cloud pressure: move now or fall behind. Everybody lifted and shifted. By 2023, the industry had a real repatriation trend, organizations moving big VMs back out of the cloud to cut costs.

Daniel’s take wasn’t that cloud is bad. It’s that lift and shift never fit a lot of those workloads well. We talked through it as a lease-versus-buy problem: running the same steady-state VMs in Azure that you used to run on your own hardware is like leasing a car instead of buying one. You pay more over time for the exact same usage pattern. The premium buys elasticity and managed services. If you’re not using either, you’re paying rent on hardware you’d rather own.

What drives the decision

Daniel’s framework: it’s a dial, not a switch, and every organization sets it differently based on:

CapEx vs OpEx. He gave aged healthcare in Australia as an example of a sector that’s almost entirely CapEx-driven and resistant to a recurring monthly bill, versus organizations that want to avoid capital commitment entirely.
Cost certainty vs cost commitment. Buying hardware gets you a fixed, predictable cost. Cloud gets you no commitment, which matters when you’re not sure a workload will stick around. Daniel described a client testing a new application in Hong Kong: about $40-50k a month in cloud infrastructure for a three to six month trial, versus roughly half a million to buy the gear outright. They couldn’t commit to buying something they might scrap in three months, so the monthly cost was the right call even though it cost more.
Geography. Multi-site organizations want one template that works from a small branch office to a full data center.
Integrations. Most reputable backup, monitoring, and management tools already work with both VMware and Hyper-V, so that part of the migration is usually not the blocker people expect.
Existing Microsoft investment. If you’re already a Microsoft customer, that’s leverage. Daniel’s read: hyperscalers are hosters now, not just software vendors, and hosters compete on price when you already have the relationship.

Azure Stack HCI’s pitch

Daniel is a longtime Hyper-V advocate, and Azure Stack HCI is where he thinks that story finally works. It scales from a single node up to sixteen, so the same architecture and the same management tooling cover a small branch office and a large regional data center. You get cloud-managed infrastructure on hardware you physically control, which means a fixed cost bubble and low latency.

He also pointed out that services which used to require going to the cloud are showing up on-prem now: Azure SQL and AKS both run on Azure Stack HCI today. That’s cloud capability moving down to hardware you own, not just up to hardware someone else owns.

The nonprofit problem nobody’s talking enough about

One thing Daniel flagged as genuinely troubling: Broadcom pulled VMware’s nonprofit and education discounts. Organizations that were used to steep nonprofit pricing are now getting lumped in with commercial customers, and some are seeing costs jump 8 to 10x at renewal. For a school or nonprofit running a tight IT budget, an 8 to 10x jump at renewal can blow the whole budget.

The takeaway

Daniel’s closing advice was to treat this like any other subscription you keep reevaluating: is this still worth what it costs? There’s no single right answer for cloud, on-prem, or hybrid. Look at your CapEx/OpEx posture, how committed you can afford to be, where your sites are, and how invested you already are in Microsoft’s ecosystem. Then pick the platform that fits those answers, not the one with the better marketing deck.

This post is part of the VMware to Microsoft podcast series.

Subscribe for more of these.