It has been nearly two years since Broadcom closed its acquisition of VMware in November 2023 and set in motion the largest enterprise virtualization re-evaluation in over a decade. When we wrote about the deal closing and then about the move to subscription-only licensing and the migration options it created, much of the analysis was necessarily speculative — organizations were assessing alternatives, but few had completed a migration. Two years on, the picture is clearer. We can now say something more concrete about where the workloads actually went, and the answer is more nuanced than either the alarmist or the dismissive predictions suggested.

Recapping the Trigger

The catalyst was licensing. Following the acquisition, Broadcom eliminated VMware’s perpetual licenses in favor of subscription-only agreements, consolidated the product range into a smaller number of bundles, and restructured the partner program. For many organizations, particularly smaller ones and those with modest deployments, the effect was a substantial increase in the cost of continuing to run VMware — in some reported cases, a multiple of what they had been paying. This converted a stable, taken-for-granted part of the infrastructure into an urgent line item demanding a decision.

The decision space had four broad options: absorb the new cost and stay; migrate to an open-source alternative; migrate to a commercial alternative; or move the workloads to public cloud. Two years of decisions have sorted organizations across all four, and the distribution is instructive.

Where the Workloads Went

A large share stayed — at least for now. The most under-appreciated outcome is inertia. Virtualization is deep infrastructure; migrating it is disruptive, risky, and labor-intensive. Many organizations, especially large enterprises with extensive VMware estates and deep operational dependence, concluded that the cost increase, however unwelcome, was less painful in the near term than a large-scale migration. They renewed under the new terms while beginning to plan longer-horizon migrations or hedges. The lesson is that switching costs in core infrastructure are enormous, and a price increase that would trigger immediate flight in a commodity market produces grumbling and gradual planning in a sticky one. The exodus was real but slower than the loudest early predictions implied.

Proxmox captured a great deal of the small-to-midmarket migration. Among organizations that did move, Proxmox VE emerged as a leading destination, particularly for smaller deployments and for organizations comfortable with open-source infrastructure. Built on KVM and LXC with an integrated management interface, Proxmox offered a recognizable operational model and a price point that contrasted sharply with the new VMware economics. Its momentum over these two years has been one of the clearer market signals — it absorbed a meaningful portion of the organizations for whom VMware’s repricing was most acute.

OpenNebula and OpenStack drew the platform-independence seekers. Organizations whose motivation was not merely cost but a strategic decision to escape single-vendor dependence gravitated toward the open cloud-management platforms. OpenNebula’s relative operational simplicity made it attractive for organizations wanting cloud-style management without the full operational weight of OpenStack, while OpenStack remained the choice for larger and more demanding private-cloud deployments, particularly at telecommunications providers and research and government infrastructure operators. For this site, with its roots in the distributed-systems research lineage that produced OpenNebula, the renewed enterprise attention to open cloud management has been a notable development — concepts that lived for years primarily in research and specialist deployments found a broader enterprise audience because of a licensing shock.

Nutanix was the principal commercial alternative. Among organizations that wanted to leave VMware but preferred a commercial, supported, turnkey hyperconverged platform rather than assembling and operating an open stack, Nutanix was the most prominent destination. It offered the operational familiarity and vendor support that risk-averse enterprises wanted, while providing an exit from the specific VMware relationship. The trade-off — exchanging one commercial dependence for another rather than achieving genuine independence — was acceptable to organizations whose goal was escaping Broadcom’s terms rather than escaping vendor lock-in as a principle.

Public cloud absorbed some, but less than expected. The early speculation included a scenario in which the VMware repricing would push large volumes of workloads off-premises entirely into the public hyperscalers. This happened to a degree, particularly for workloads that were already cloud-suitable and for organizations already pursuing a cloud-first strategy that the VMware situation simply accelerated. But the migration of a VMware estate to public cloud is itself a major undertaking, and for many workloads the economics of running them continuously in public cloud were no more attractive than the new VMware pricing. Public cloud captured the workloads it was always going to capture eventually; the VMware shock advanced the timeline more than it changed the destination.

What Two Years Taught the Market

Several lessons stand out from the migration experience.

Migration is harder than vendors of alternatives suggest and easier than incumbents claim. Organizations that migrated generally found the process demanding but achievable — the tooling for converting VMware virtual machines to other platforms matured rapidly as demand materialized, and the alternatives proved production-capable. The narrative that leaving VMware was practically impossible did not survive contact with the organizations that did it. Equally, the narrative that migration was trivial did not survive contact with the operational reality of moving production virtualization.

Operational capability is the real constraint. The organizations that migrated most successfully to open platforms were those that either had, or were willing to build, the operational competence to run them. The cost saving of an open platform is partly offset by the operational ownership it requires. Organizations that migrated to open infrastructure without investing in the corresponding skills sometimes found the total cost less favorable than the sticker comparison suggested.

The licensing shock accelerated a structural shift already underway. The renewed interest in open private-cloud infrastructure was not created from nothing by Broadcom; it accelerated a pre-existing trend toward platform independence and open infrastructure. The VMware situation gave that trend a concrete, urgent trigger, converting a slow philosophical drift into immediate procurement decisions. The forced re-evaluation made organizations confront infrastructure dependencies they had not examined in years.

Single-vendor dependence is now a recognized risk category. Perhaps the most durable effect is attitudinal. The VMware experience served as a vivid, expensive demonstration of what deep dependence on a single commercial platform can cost when the vendor’s commercial strategy changes. Infrastructure teams that previously treated their virtualization layer as a stable given now treat vendor concentration as a risk to be actively managed — a mindset shift that outlasts the specific VMware episode and informs decisions across the rest of the stack.

The Outlook

Two years in, the VMware migration story is neither the catastrophe nor the non-event that competing early narratives predicted. A significant minority of organizations migrated, distributed across Proxmox, the open cloud platforms, Nutanix, and public cloud according to their size, motivation, and operational appetite. A larger share stayed for now while planning hedges and longer-horizon moves. The migration wave is not finished — many of the organizations that renewed under the new terms did so as a stopgap, and their migrations will unfold over coming renewal cycles rather than all at once.

The deeper significance is the normalization of infrastructure-vendor risk as a managed concern. For infrastructure planners, the practical takeaway from two years of VMware migrations is to treat platform independence as a property worth maintaining deliberately — through portable abstractions, documented exit paths, and the operational capability to exercise them — rather than discovering its absence the next time a vendor’s commercial strategy shifts.

Further Reading

Back to Blog