Six hundred stores moved and nobody left the office
Gary Sliver, director of platform engineering at Sheetz, has watched more than 600 convenience stores change hypervisor without dispatching a single technician. StorMagic announced the project on 15 July 2026, dateline Bristol, England, and the numbers it published are unusually specific for a vendor release: 830-plus stores in scope, 600-plus already migrated, roughly 200 stores per month, four months from start to finish.
The detail that matters is the one that sounds like a footnote. Every store ran the migration on the Dell R440 and R450 server pair it already had. No hardware was replaced. No site was visited. The work was done remotely, store by store, at a rate that would embarrass most European branch rollouts.
Sliver's own words set the tone: "StorMagic worked to understand our needs and requirements, to include a 24/7/365 operating environment, and ultimately delivered a product and plan that is allowing us to migrate hundreds of locations from VMware quickly and with minimal downtime, without requiring hardware replacements or having to send technicians to every site. This project is proof that large-scale edge migrations don't have to be disruptive, drawn-out or scary affairs."
What the release actually says
The document is a migration story, not a grievance. StorMagic said Sheetz is replacing VMware across its retail estate with StorMagic SvHCI, gave the store counts and the pace, and framed the benefit in generic terms: reducing cost, complexity and operational disruption. Scott Mann, StorMagic's SVP of global sales, added the line the company clearly wanted quoted: "Sheetz is demonstrating that enterprise-scale VMware migrations can be completed rapidly, remotely and without costly hardware refreshes."
There is no price in it. No contract value, no savings figure, no comparison of what Sheetz paid before and pays now. The terms were not disclosed and neither was the saving. The only cost language beyond that generic phrase is StorMagic boilerplate about rising hardware prices, which is a statement about servers, not about software licences.
The motive nobody sourced
Broadcom is never named in the release, and the trade press named it in nearly every headline. Neither licensing nor subscriptions nor a price increase appears in the document at all, yet the coverage universally reads the project as a Broadcom-licensing exodus, complete with verbs like quitting and phrases like VMware's grip. The most careful independent analysis we read concedes the point openly: the Broadcom motive is the author's own inference, and no Sheetz source states that reason.
Two other figures deserve the same scepticism. The 11,000 virtual machines that circulates through the coverage is not in the primary; it is arithmetic somebody performed on the store count. And the 500 percent price increase that appears nearby refers to a generic unnamed VMware customer, not to Sheetz. Attach it to Sheetz and you have invented a fact.
This is not pedantry, it is procurement hygiene. An owner who cites Sheetz in a renewal negotiation as proof that Broadcom pricing drives customers out is citing a motive that the source document does not contain, and the counterparty across the table can read the release too.
The incumbent was already inside
StorMagic did not win Sheetz from VMware; it was already underneath VMware. SvSAN, StorMagic's storage product, had been the storage layer beneath the hypervisor across hundreds of Sheetz stores before any of this started. Sheetz was an existing customer. The headline that says StorMagic snared Sheetz from VMware's grip describes a competitive win that did not happen.
What happened instead is more useful to an owner. An adjacent vendor already inside the stack expanded upward into the layer above it, and it did so with the enormous advantages that position confers: it knew the estate, it knew the hardware, it had a working relationship with the platform team, and it had already been trusted with the data. That is our reading, not the companies' claim.
Know this before you convene a competitive evaluation, because it substantially decides the outcome. The vendor best positioned to replace your hypervisor is the one already sitting in your stack, and a bake-off run without acknowledging that is a bake-off with a predetermined winner and a lot of wasted supplier time.
What the reused servers prove about exit cost
The Dell servers are shouting the real lesson: the exit cost of a platform is dominated by the hardware refresh, not the licence. Sheetz moved 600 stores in about three months precisely because it never touched a server, never scheduled a truck roll, never staged equipment and never coordinated a technician's calendar against a store's opening hours. Strip those out and a migration that owners describe as a multi-year programme becomes a quarter.
Owners benchmark their escape against licence spend, conclude the numbers are impossible and renew. That is the trap, and it is arithmetic done on the wrong line item. Before you accept the next renewal, find out what your escape actually costs, measured in hardware you would have to buy and sites you would have to visit, not in licence line items you already know by heart.
The honest scope limit: Sheetz is a US convenience-store chain and this is a US retail estate. It does not bind or describe a European operator, and nothing in it transfers automatically to a retail or logistics estate of several hundred sites in the EU or the UK. Its value to a European owner is as a measured precedent for what edge migration costs and how fast it can run, and that read-across is ours, not the companies'.
Read next: Nobody Forbade It. The Filter Refused Anyway. | Two AI Giants Just Became Your Integrator



