2GuysTek · Comment Ledger v3 · complete pull · lpHsSaL-NJw

Where did everyone actually land?

713 comments, hand-coded by declared destination, organization scale, platform sentiment, and recurring argument. 189,306 views · 4,550 likes.

Sample and method 713 of 841 comments YouTube reports = 84.8%, up from 696 on the previous pull. The missing 15% are deleted, held for review, or from terminated accounts. 468 top-level and 245 replies; 51 creator replies excluded from all counts. A commenter is counted once per platform they name, so hybrid answers count twice and totals exceed the number of people. Only 211 of 653 coded statements reveal organization size, so the scale splits below are drawn from that subset, not the whole. Timestamps from yt-dlp are rounded to 27 distinct days and are unusable for time-series.
713
comments coded
84.8% of all reported
82,100
VMs and hosts named
across 27 organizations
400%
median increase cited
n=12, range 15 to 5,400%
53
comments pointing at
the 2027 contract cliff
15
who say they are
staying on VMware

01 — Where people landed

Counts anyone running a platform now or actively executing a migration to it. Colour shows organization scale where the commenter revealed it; the dark segment is unstated. Proxmox leads on mentions but Hyper-V is closer than the first read suggested, and VMware's 24 is not all inertia: 15 of those are people explicitly choosing to stay.

Enterprise Education SMB / mid-market MSP / consultant Homelab / individual Unstated

02 — The segmentation, using only commenters who revealed their scale

Share of each platform's landings that came from enterprise or education, among the 211 commenters who said. OpenShift and Nutanix are enterprise-only. XCP-ng is the opposite. Proxmox is not the SMB-only tool the first sample suggested: a third of its identified landings are enterprise, including 4,000-VM and 3,500-VM estates.

03 — Weighted by workloads, the ranking inverts

27 commenters named a VM or host count, totalling 82,100 workloads. Counting machines instead of people, Nutanix takes a third of everything moving on the strength of two very large estates, and Hyper-V collapses from 31 mentions to under 5% of workloads. One person equals one vote; one person can also equal 20,000 VMs. Both charts are true and they disagree.

04 — How people talk about each platform

Explicit praise minus explicit criticism, per commenter, independent of what they run. HPE Morpheus is the surprise: net negative, and the criticism is about HPE as a vendor rather than the product.

05 — What the comments argue about

Recurring arguments across all 653 coded statements.

06 — The price shock, in their own numbers

IncreaseCommenterDetailLanded
5,400%@Dfk429S9fo3$600 for three years of ESXi on 3 VMs became a $33,000 renewal quote. Attributed by another commenter to the 96-core minimum.Unstated
~900%@Potzii85MSP already paying licensing in the millions; quote would have been 9x. Moving every customer.HPE Morpheus
740%@mikeyfoofooOne month to decide despite a support deal to 2027. Paid, then still got a cease and desist.Unstated
567%@jonathanbuzzard1376University central IT, 75k to 500k per year. Migration hardware paid for itself in a year.Proxmox
479%@markdownsouth1500$19,000 SnS became roughly $110,000 per year. Kept perpetual sockets, ignored Broadcom.KVM on Rocky
400%@WhyMustThisBeSoLong$10k to over $50k. Already had Windows Datacenter licenses, so went bare metal Hyper-V.Hyper-V
400%@noelreynolds2356A school. Called it a non-starter and did not renew.Undecided
317%@oltepCustomer since 2006.Unstated
300%@MarkKnightsNegotiated down, locked three years to buy planning time.VMware, for now
300%@TrackyMattTriple cost to renew next month, on a three year term. Same licensing, no more product.Deciding now
200%@JD-Dallas65Refused to pay double.Hyper-V
15%@TheCynysterMindThe outlier that matters. Too small to be worth squeezing, so they stayed. Prepared to go 100% Microsoft if that changes.VMware

07 — Fleet sizes named

WorkloadsCommenterDestination
20,000@mouazaljamil2276Nutanix. Single largest move stated in the entire thread.
10,000@captchacaptchaNutanix for prod, Proxmox for non-prod. One-year mandate. Reusing old servers because new ones are unaffordable.
10,000 → 4,000@jamespfearn4326GCP Compute Engine plus cloud-native rewrites. Estate shrank 60% in the move.
10,000@rijojohn85OpenStack, across many clients, by a dedicated migration team, in one year.
4,000@prabhakarpatil8085Proxmox, in flight.
4,000 vCPU@RoyOlsenProxmox, as contracts end.
3,500@stefanh7948Proxmox across 14 clusters. Complete.
3,500@ericedeStill evaluating. Demoed Morpheus, liked it, cost killed it. Has Hyper-V from acquisitions and won't consider it.
3,000@theLaminator88Oracle Linux Virtualization Manager, and not happy about it.
2,000@kataklysm812OpenShift.
1,500@FranckEhretStaying, reluctantly. SAP and Cisco VOIP dependencies; legacy SAN rules out Nutanix.
1,500@DS-pk4ehHyper-V.
1,500@tech-daddyOracle VM.
1,400@athrvanaik5036Bank: Nutanix + OpenShift Virtualization + KVM.
1,000+@AlistairMaxwell77Hyper-V 2025 clusters, state government.
1,000+@kamlshunHyper-V, manufacturing, deep in Azure and M365.
1,000@VikingDeveloperAzure Local, cloud-first strategy, now second-guessing it over sovereignty.
800+@thomasfuchsberger3912Proxmox. A commercial cloud provider, not an SMB.
600-700@syllysayanami5032Nutanix AHV with Prism Central. Nine clusters collapsed into two clusters of five nodes.
400@triradioProxmox. Shut down another VMware cluster the day they commented.

08 — Platforms they asked you to cover

SUSE Harvester / Rancher6 Apache CloudStack3 oVirt3 Sangfor3 VirtualBox3 Raw KVM / QEMU2 Oracle OLVM2 Incus2 Virtuozzo2 H3C1 Kubermatic1 StorMagic1 Nutanix CE1 IBM LinuxONE1 TrueNAS Scale1 Citrix / VDI market1 Public cloud as a landing1

09 — Signals worth building on

Buried lede

HPE is giving away hardware

One commenter reports HPE approached them offering free hardware to move off VMware onto Morpheus. An MSP is moving every customer to Morpheus. But Morpheus is the only platform in the set with net negative sentiment: unimpressed after evaluation, monthly cold calls from multiple reps, GreenLake support complaints, and "HPE is the next lock-in."

Free hardware plus bad word of mouth is a story. Nobody has covered it.

Counter-narrative

The people staying have reasons

Fifteen commenters say they are staying, and almost none say it is because they are trapped. Full VCF stack with vSAN, NSX and Automation has no single-vendor equivalent. SAP and Cisco VOIP certifications. Legacy SAN that rules out Nutanix. A VDI shop on Omnissa. One org whose increase was only 15% because they were too small to bother squeezing.

Your video's framing assumes leaving is the rational default. Two dozen comments disagree with specifics, and one flatly disputes your Nutanix pricing claim with their own numbers.

Strongest thread, 53 comments

Everyone is waiting for 2027

The single most repeated argument, by a wide margin. Three-year ELAs signed just before the acquisition expire next year. Several add a second trigger: the hardware refresh cycle. One commenter puts full exit at five years and names 2031.

Underweighted

Digital sovereignty is now a buying criterion

Thirteen comments, and it is no longer abstract. Five separately point out Proxmox is Austrian and therefore European. One org went to Huawei FusionCompute in South Africa. A consultant reports Sangfor, Huawei DCS and H3C driving bank migrations across Asia, LatAm and Africa. One commenter moved to Azure Local under a cloud-first strategy and is now being told to abandon US cloud.

Untold, 12 comments

The individuals nobody serves

Workstation Pro refugees running Cisco Modeling Labs, Windows 2000 for a customer, retro OSes, air-gapped labs. One moved to six physical PCs and a hardware KVM because nothing else does 3D passthrough. One asks directly: "What's in it for us individual users? We're stuck with Broadcom?"

Proxmox's real ceiling

Support, not capability

Six criticisms and they cluster. The SLA and support model disqualifies it at companies with contractual SLOs. Multipath iSCSI needs config-file surgery. Snapshot behaviour on shared storage. Doubts about maintenance burden at scale. Notably, a Proxmox user rebuts each one in the replies, including that LVM-on-iSCSI snapshots landed in version 9.

And in this pull, the argument resolved itself. The person who said snapshots were gone came back, acknowledged version 9 fixed it, and said that having sorted their backups out it's now "more of a goodie than a necessity." Two operators worked it out in your replies without you.

Reality check

Nutanix cuts both ways

One org consolidated eight or nine VMware clusters into two Nutanix clusters running 600 to 700 VMs, exactly the pitch Nutanix makes. But three separate operators, unprompted, describe the same failure on the other side: mismatched end-of-life dates meant they could not realize ROI on their hardware, one concluding "AHV only, never HCI again." Paired with a hardware compatibility list strict enough to disqualify the older servers everyone is stuck extending, that is the closest thing in this dataset to a documented pattern rather than a one-off complaint.

Second-order

The cost crisis stacked

VMware's increase collided with RAM and storage prices up 5x and cloud spend doubling. One org's AWS bill went £300k to £1.8m; another €35k to €80k per month. Multiple commenters say they cannot buy new servers at all and are refurbishing old ones, which is itself driving platform choice: Nutanix's strict HCL loses to Proxmox on second-hand hardware.

New in this pull

The certification glut

A commenter running 3,500 VMs raised something nobody else did: the market is about to be saturated with VMware certified engineers, because everyone got certified when it was the only platform and now there's nowhere to use it.

Same comment, better detail: one of their global offices emailed asking for support and discovered their licenses had expired the previous year. Nobody had budgeted for it. That is a failure mode nobody is talking about, and at that org's scale it cannot be unique.

New in this pull

OpenShift's clean record just broke

Until this pull OpenShift had nine pieces of praise and one criticism, all of it from Red Hat shops saying it was hard but worth it. Now there's an operator saying large VMs with terabyte virtual disks on Pure Storage killed productivity: any latency causes Windows guests to blue screen or orphan inodes. Their verdict was that it isn't built for enterprises.

Every alternative in this dataset now has at least one detailed failure report from somebody running it in production. That is arguably the real finding.

Skills

The retooling problem nobody budgeted

Several comments make the same argument: the blocker is not the product, it is that Windows and VMware admins do not know KVM, and one org's SME advising against migration was the person whose skills were locked to VMware. Meanwhile the VMware certification job market is collapsing and VMUG lab licenses now sit behind an annual VCP-VCF exam.