Disclaimer: This article describes an ongoing customer evaluation. All figures are simplified and based on customer-specific information and indicative assumptions. Future growth and potential savings outside the described edge environment are hypothetical and have not been included in the TCO calculation. Actual pricing, migration effort, hardware compatibility and transformation costs depend on the individual environment, commercial terms, support levels and final technical design.
I believe we all agree, that infrastructure decisions are rarely made on licensing costs alone. Architecture, operational experience, existing investments and migration effort usually matter just as much. But there are situations where a change in licensing model has such a significant financial impact that it becomes difficult not to evaluate alternatives.
I am currently involved in such a discussion with a customer looking at the future of its distributed edge infrastructure.
The customer has not yet decided to move away from VMware vSphere. Nutanix Cloud Infrastructure with AHV is currently being evaluated as an alternative, and there are still technical and operational questions to answer before a final decision can be made.
The financial side of the evaluation, however, already provides a strong argument for taking the alternative seriously. Based on the current environment and the assumptions we are using, the difference could amount to almost $1M over five years.
In the context of a large enterprise IT environment, $1M spread over five years may not immediately sound transformational. But this calculation covers only a relatively small and clearly defined part of the customer’s infrastructure.
The edge environment is expected to grow. Additional locations are likely to be added over time, together with more clusters and potentially dozens of additional virtual machines. If that happens, the financial difference between the two licensing models could become considerably larger.
The customer’s overall virtualization environment is significantly larger than the edge footprint discussed here. I would not simply extrapolate the edge economics across the entire data center estate, because workload density, architecture and licensing characteristics are different. But if the same evaluation identifies additional areas where NCI and AHV can improve the economics, the potential value goes well beyond the initial edge use case.
A relatively focused edge discussion could suddenly become the door-opener for a much broader infrastructure conversation.
What makes the initial case compelling is that the potential savings do not depend on a major hardware refresh or a fundamental redesign of the edge architecture. Let’s take a closer look.
The Situation – A Small Workload on a Resilient Infrastructure
The customer operates distributed edge environments with four physical hosts at each location. The clusters provide the resilience required for applications running locally, but they are relatively small from a workload perspective. A typical cluster runs no more than around ten virtual machines.
During one of our discussions, the customer mentioned that the VMware licensing for these environments currently costs approximately $5,000 per physical host per year.
For a four-node cluster, that translates into roughly $20,000 per year.
Yes, on its own, this is not necessarily a number that would trigger an infrastructure transformation, but the economics change when the same architecture is replicated across many locations.
The environment we are discussing could eventually consist of approximately 60 physical nodes, or 15 four-node clusters. At that scale, the VMware licensing would amount to approximately $300,000 per year.
Assuming those costs remain at a comparable level, the five-year licensing expenditure would be around $1.5 million.
And this was exactly the point where the discussion became less about comparing two virtualization platforms and more about whether the licensing model still fits the actual workload.
Looking at the Workload Instead of the Physical Capacity
There is nothing wrong with the four-node architecture. The physical infrastructure is there to provide availability and resilience at remote locations. Reducing the number of servers simply to lower software costs would solve the wrong problem. The more relevant question is how that infrastructure should be licensed when only a maximum of ten virtual machines are running on it.
This is where NCI Edge changes the economics.
NCI Edge is licensed based on the number of virtual machines instead of the physical CPU capacity underneath them. For a highly consolidated data center environment, this distinction might not necessarily create the same economic advantage. For a distributed edge environment with relatively few workloads per cluster, it can make a substantial difference.
For the current evaluation, we are working with an indicative “NCI Edge Pro” cost of approximately $500 per VM per year.
Ten virtual machines would therefore result in annual licensing costs of approximately $5,000 per cluster, compared with roughly $20,000 under the customer’s current model.
Across 15 clusters, the difference becomes significant. The estimated annual software cost falls from approximately $300,000 to $75,000.
Over five years, the comparison is $1.5 million versus $375,000.
That represents a potential gross licensing difference of $1.125 million. So far so good, right?
Preserving the Existing Hardware Changes the Equation
Software savings alone do not make an infrastructure transformation worthwhile. A lower licensing bill quickly becomes less attractive if achieving it requires replacing servers that still have several useful years ahead of them. This is another important part of the current evaluation.
Nutanix is not limited to proprietary appliances. NCI and AHV can run on a range of supported standard x86 server platforms. Provided the customer’s existing servers pass the required compatibility checks, there is therefore a realistic path to continuing to use the hardware already deployed.
If the business case required 60 new servers before the first dollar could be saved on software, the economics would look very different. In this scenario, the objective is to change the infrastructure software while preserving as much of the existing investment as possible. The four-node architecture does not need to disappear, the applications can remain local, and the required resilience can remain in place.
The change happens primarily at the software and operational layer.
A Platform Change Still Has a Cost
It would nevertheless be misleading to describe the $1.125 million licensing difference as the actual saving.
Moving from VMware vSphere (ESX + vCenter) to AHV requires work. Existing brownfield environments need to be migrated. The virtual machines have to be moved, tested and validated before the old environment can be retired. Existing automation needs to be reviewed and adapted to deploy and manage the Nutanix environment.
There is also an organizational investment. Administrators need to become familiar with Prism and AHV (and AOS). Operational documentation and runbooks have to be updated, and existing backup, monitoring, security and support processes need to be validated against the new platform.
Not all of these costs increase linearly with the number of clusters. Training is largely an upfront investment. The same applies to documentation and much of the work required to adapt deployment automation. Once the processes have been established, they can be reused for subsequent locations.
This becomes particularly relevant if the edge footprint grows as expected. The initial transformation creates capabilities that can be reused for the next cluster and the one after that. The marginal effort of introducing another Nutanix location should therefore be considerably lower than establishing the platform for the first time.
Greenfield deployments make the equation even simpler. A new location does not have an existing ESXi environment that needs to be migrated first. NCI with AHV can become the target platform from the beginning.
To avoid presenting an unrealistically optimistic business case, I included an estimated $150,000 transformation budget in the calculation. This is intended to account for migration, automation, testing, training, documentation and internal project effort.
That brings the estimated five-year cost of the Nutanix scenario from $375,000 to $525,000.
Compared with $1.5 million for continuing the current licensing model, the potential net difference is still $975,000 over five years.
Image disclaimer: The image was created with the assistance of ChatGPT.
The Eight-Month Payback
The payback period is perhaps the most useful way to put the transformation cost into perspective.
The difference between the two licensing scenarios is approximately $225,000 per year, or $18,750 per month.
At that rate, a $150,000 transformation investment would theoretically be recovered after approximately 8 months.
The transformation therefore does not need several years before the economics begin to work. Based on the assumptions in this model, the investment would be recovered within the first year. The remaining period contributes directly to the overall TCO advantage.
Over five years, the model suggests a reduction of approximately 65% compared with continuing the current licensing approach.
Of course, this remains a model. Pricing can change, migration projects can become more complicated than expected, and technical requirements can influence the final architecture. That is precisely why the customer is evaluating the platform rather than making a decision based on a spreadsheet.
The Decision Has Not Been Made
This is not a migration success story. At least not yet. 🙂
The customer is still evaluating NCI with AHV and has not made a final platform decision. Technical compatibility, operational requirements, automation and the migration approach still need to be assessed in sufficient detail.
But the financial part of the discussion has already become considerably clearer.
The potential savings are large enough to justify the cost and effort of evaluating a different platform. Even after allocating a substantial budget for the transformation, the five-year economics remain attractive. And because the business case does not depend on purchasing an entirely new hardware platform, there is a realistic opportunity to preserve existing infrastructure investments.
There is also a strategic dimension to the project that goes beyond these 15 edge clusters.
If NCI with AHV proves itself technically and operationally in this environment, the customer will have gained practical experience with the platform. The operations team will have been trained, automation will have been developed, processes will have been documented and the migration path from ESX to AHV will have been tested in production.
At that point, evaluating Nutanix for other parts of the infrastructure becomes a very different conversation. The edge project would no longer be an isolated cost optimization exercise and it could establish the operational foundation and confidence required to look at a much larger virtualization estate.
Whether the economics will be equally attractive elsewhere remains to be analyzed. A large data center environment with higher VM densities requires a different TCO calculation than a small edge cluster. But even a substantially smaller percentage improvement applied to a much larger infrastructure could translate into significant absolute savings.
That is why I see this edge evaluation as potentially more important than the initial $975,000 business case suggests. The immediate opportunity is to reduce the cost of a clearly defined infrastructure environment by almost $1M over five years.
Edge is not the only use case where this licensing approach can significantly change the economics. VDI can show a similar pattern. Instead of licensing the physical cores required to provide resilience and capacity, NCI-VDI is licensed per concurrent user. This can make the TCO very attractive in larger VDI environments, especially when additional capacity or a disaster recovery site is required. It is another example of why the right licensing model can matter just as much as the underlying technology.
Coming back to our edge use case, the longer-term opportunity is to prove whether the same platform can create value elsewhere and the customer now has to determine whether the technical and operational findings support the business case.
So far, the evaluation is moving in the right direction, yes.
And sometimes, a relatively small edge (or VDI) project is exactly where a much larger infrastructure transformation begins. 🙂
