Photo by Towfiqu barbhuiya on Pexels
For years, the virtualization market operated under a quiet assumption: once a large enterprise built its data center on VMware, it would stay there indefinitely. The platform’s dominance — powering the majority of virtualized workloads in telecom, finance, and government — gave VMware a kind of gravity that made migration seem not just expensive but strategically unwise. Then Broadcom acquired the company in November 2023, and the gravity shifted. Now T-Mobile, one of the largest VMware customers in the telecommunications sector, is doing what few thought possible: pulling tens of thousands of virtual machines off the platform in what amounts to the most dramatic defection since the acquisition closed.
The announcement, made public via a court filing in the company’s ongoing lawsuit against Broadcom, reveals that T-Mobile has already begun a large-scale migration away from VMware’s hypervisor and management stack. The carrier wants Broadcom to continue supporting the perpetual licenses it purchased pre-acquisition — a request Broadcom has apparently resisted as part of its aggressive push to convert all customers to subscription-based pricing. The situation is more than a contract dispute; it is a real-time stress test of whether any single vendor can dominate enterprise infrastructure while also alienating its largest users.
Why T-Mobile Is Dismantling Its VMware Stack
The surface-level issue is licensing. T-Mobile holds a significant number of VMware perpetual licenses — the traditional model where you buy the software once and pay annual support fees. After Broadcom took over, the new management not only discontinued perpetual sales but also began pressuring existing perpetual customers to switch to subscriptions, often with steep price increases. T-Mobile’s lawsuit, filed in 2025, argues that Broadcom is refusing to honor the support terms of those perpetual agreements, effectively forcing the carrier into a more expensive model without a contractual basis.
But the deeper story is about trust and architectural sovereignty. Telecommunication networks are not like a typical corporate IT environment; they require near-100% uptime, low-latency data paths, and rapid scalability for events like Super Bowl traffic or natural-disaster surges. T-Mobile cannot afford to be held hostage by a vendor that changes the rules mid-contract. By moving tens of thousands of VMs — a process that involves re-provisioning workloads, re-architecting network functions, and re-training staff — T-Mobile is signaling that the cost of staying on VMware under Broadcom’s terms is higher than the cost of leaving.
The Technical Lift: Migrating Tens of Thousands of Virtual Machines
To understand the scale, consider that a single large telecom data center can host anywhere from 500 to 5,000 virtual machines, depending on the function (core network, billing, customer management, etc.). T-Mobile operates multiple data centers across the U.S. Moving tens of thousands of VMs means touching every layer of the stack: compute, storage, networking, orchestration, and monitoring. In plain language, it is like transplanting every organ in a body while keeping the body awake and working.
T-Mobile is likely moving to a combination of open-source KVM (Kernel-based Virtual Machine) and alternative hypervisors such as Nutanix AHV or Microsoft Hyper-V. KVM, already widely used in cloud providers like AWS and Google Cloud, offers a cheaper but less integrated alternative. The carrier will also need to replace VMware’s management tools — vCenter, NSX for networking, vSAN for storage — with equivalents from either the open-source ecosystem (OpenStack, Kubernetes) or a rival vendor. The migration is not just a technology swap; it is a reorganization of the engineering team’s daily workflow and incident-response procedures.
The timeline matters. T-Mobile has not disclosed a completion date, but migrations of this size typically take 12 to 24 months even with dedicated teams. During that period, the carrier must maintain dual environments — the old VMware estate still running some workloads while the new platform comes online — adding complexity and increasing the risk of configuration drift or security gaps. That T-Mobile is willing to accept that risk indicates just how untenable the Broadcom relationship became.
A Familiar Story: When Vendor Consolidation Sparks Exodus
This is not the first time a dominant infrastructure vendor has triggered a mass migration by changing its licensing model. In 2009, Oracle acquired Sun Microsystems and promptly raised support prices for Sun’s hardware and Solaris operating system, prompting enterprises to accelerate moves to Linux and x86 commodity servers. More recently, Cisco’s decision to bundle licensing and raise prices on its networking gear after acquiring companies like Meraki led some large customers to explore white-box switching and open networking.
The VMware-Broadcom case follows the same pattern: an acquisition creates a new entity with a strong incentive to extract more revenue from the installed base. But VMware’s position in the virtualization market is even more entrenched than Sun’s or Cisco’s ever was. According to industry estimates from 2024, VMware still ran on over 70% of virtualized servers in large enterprises. A report from The Register noted that while many organizations had started evaluating alternatives, few had actually begun full-scale migration. T-Mobile’s move turns that evaluation into action, potentially creating a tipping point for rest of the telecom sector.
Who Loses, Who Wins: The Ripple Effects on Telecom and Enterprise IT
The immediate loser is Broadcom. Losing a customer of T-Mobile’s size means forgoing not only the subscription revenue but also the reference value — the ability to tell other telecoms that VMware is still the standard. If T-Mobile successfully runs its network on an alternative for a year without major incidents, the decision becomes easier for every other carrier. AT&T and Verizon, which also run large VMware estates, will be watching closely. Both have been quiet publicly, but internal discussions about migration are likely already underway.
The winners include Nutanix, Microsoft, and the open-source community. Nutanix, which offers an integrated hypervisor and storage solution, is perhaps best positioned because it already targets large enterprises with a single-vendor experience similar to VMware’s. Microsoft’s Azure Stack HCI and Hyper-V provide a natural path for organizations already using Microsoft’s management tools. The open-source ecosystem — KVM, OpenStack, Kubernetes — benefits from more investment and talent attention, but remains a fragmented option that requires significant in-house expertise.
For the broader enterprise IT industry, T-Mobile’s migration is a case study in the cost of vendor lock-in. The legal battle, though specific to licensing terms, has exposed a structural vulnerability: when a vendor is acquired, the new owner has no obligation to preserve the old relationship. Companies that once saw VMware as a safe, long-term bet now face a choice — pay more or rebuild. The financial math will differ for every organization, but the emotional math has changed. Trust in VMware as a stable platform partner has been broken, and that trust will not be restored by a discount on a subscription plan.
The Broader Lesson: Cloud Infrastructure Independence
Beyond the telecom sector, this episode reinforces a lesson that the cloud era has taught in fits and starts: infrastructure independence is a strategic asset, not just a cost center. The hyperscalers (AWS, Azure, Google Cloud) have long argued that their abstractions free customers from worrying about physical hardware and hypervisors. But T-Mobile’s situation shows that abstraction does not eliminate dependency — it merely shifts it from one vendor to another. The carrier is now paying the price of having ceded too much control to a single virtualization layer.
The practical implication for IT leaders is straightforward: every organization should have a credible plan to run its most critical workloads on at least two different hypervisors or cloud platforms. That does not mean running everything in active-active mode, which is prohibitively expensive. It means maintaining the skills, tools, and architectural patterns needed to migrate if the primary vendor becomes untenable. For many, that will involve investing in Kubernetes as a common orchestration layer that can run on top of KVM, Hyper-V, or even bare metal.
Looking ahead, T-Mobile’s migration will likely be studied in business schools alongside other inflection points in vendor lock-in. If successful, it will embolden other large enterprises to accelerate their own diversification plans. If it falters — due to cost overruns, performance degradation, or operational complexity — it will give other companies a reason to pay Broadcom’s higher prices rather than risk the upheaval. Either way, the assumption that virtualization infrastructure is too hard to change has been shattered. The question now is not whether large migrations will happen, but when the next one will be announced.
Editorial Note: This article was produced with AI assistance and reviewed by the Celloraa editorial team for accuracy and clarity. It is intended for informational purposes only. Read our Editorial Policy.
Leave a Reply