Broadcom bought VMware and set every small shop on fire. Get your data out.
Broadcom bought VMware, gutted the mid-market, and the entire ecosystem paid the bill. Here is what actually happened, why it grinds, and how to get your data out.
You know what really grinds my yearn?
November 2023. Broadcom writes a check for VMware around $61 billion in cash, roughly $69 billion including assumed debt. The rest of the industry claps politely because that's what you do when big M&A happens. Analysts talk about "synergies". Journalists write the same recycled paragraphs about "commitment to customers". Executives at both companies do the customary email tour where they promise nothing will change for the people who actually run infrastructure.
Ninety days later everything changed.
If you were an ESXi shop with a perpetual license, congratulations, you were already dead, you just didn't know it yet. Free ESXi as a product? Killed. vSphere Essentials Plus for the small shops? Killed. Perpetual licensing itself? Killed as a purchasable SKU. Everything moved to subscription. Not "subscription with an escape hatch". Not "subscription that maps to what you were paying before". Full-fat "you now pay us every year forever or your hypervisor stops booting" subscription.
For the shops that survived the pricing tier reshuffle, the number went up. Not "went up a little". Not "went up because of inflation". Multiple hundred percent price hikes were widely reported. Some customers publicly reported jumps in the several-hundred to over-a-thousand percent range for feature parity with what they had the week before. And this was for existing customers. Loyal customers. Customers who had been paying VMware for a decade or two.
The response from Broadcom was the response Broadcom always gives, because this is the third time Broadcom has done this exact play. Ask CA. Ask Symantec. The playbook is: buy the incumbent, gut the mid-market and small-shop offerings, keep the top few thousand enterprise customers, extract maximum margin from them until they finally build the escape hatch themselves, then move on to the next acquisition.
The playbook has receipts
CA Technologies, 2018. Broadcom pays roughly $18.9 billion for a mainframe software company whose customers were mostly locked-in enterprise accounts running decades-old workloads they could not easily migrate. Within eighteen months, CA's mid-market and channel programs were reported as gutted. Support quality fell. Product roadmaps evaporated. The remaining customers were the ones with the least optionality, and they paid dearly for that lack of optionality. Analysts at the time described it as a value-extraction acquisition, because that's what it was.
Symantec, 2019. Broadcom pays roughly $10.7 billion for the enterprise security division. Same play. Focus on the top strategic customers, defund the rest, cut back the engineering and support layers that made the product worth buying in the first place, collect the renewal checks until customers finally build the escape hatch themselves. A lot of those Symantec customers ended up on CrowdStrike, SentinelOne, or Microsoft Defender. What none of them ended up on was "still on Symantec because Broadcom made it a good deal to stay".
VMware was play number three. The people who work in enterprise software knew the second the deal was announced. The people who work in enterprise infrastructure knew it about ninety days later when the renewal quotes started landing.
Broadcom's public post-close messaging leaned heavily on words like "innovation" and "commitment to customers", repeated across every earnings call and press channel. Anyone paying attention read those the way you read airline delay updates: technically accurate, functionally meaningless, buying time for the operational reality that was already scheduled.
Broadcom also telegraphed publicly that they would focus the post-merger business on their top strategic customers, a small fraction of VMware's total customer base. VMware had over three hundred thousand customers when the deal closed. Do the math on who wasn't in that small fraction.
The rest of us? Pound sand.
What this actually looks like on the ground
I'm a sysadmin. The shape of story I keep hearing from other sysadmins goes something like this.
A shop had been running an ESXi cluster for the better part of a decade. Perpetual licenses they bought outright years ago. Standard vSphere, vCenter, maybe some VSAN sprinkled in. They were happy. It worked. They understood it. Their staff was trained on it. Their monitoring was built around it. Their backup vendor integrated with it.
Renewal quote comes back with a number that has an extra digit on the end. Not a comma, a digit. Their support-only renewal, the one where they get security patches and phone-a-human access, is now more expensive than the entire hardware refresh they were planning for next year.
They spend the next six months on nights and weekends migrating to Proxmox. Their staff, who had never touched KVM in anger, learn it because they had to. They get it done. They save the majority of what Broadcom wanted. Their SLA is the same. Their uptime is the same. Their sleep is worse only because the migration burned everyone out, but that ends when the migration ends.
Multiply that story by thousands. Vodafone announced they were moving. AT&T made similar public moves. Even Boeing, who you'd expect to have the tolerance for high vendor lock-in costs that only a defense contractor can afford, started making public noises about exploring alternatives. When Boeing publicly says they're exploring alternatives, that isn't Boeing saying "we're exploring alternatives", that's Boeing saying "we already picked one and we're using the announcement as a negotiation lever."
Meanwhile Proxmox sits over there, quietly incrementing version numbers, giving away the software for free, charging reasonable rates for support subscriptions to anyone who wants them. Proxmox VE. XCP-ng. OpenStack for the shops with the operational chops. Nutanix if you want turnkey. OpenShift Virtualization if you're already in the Red Hat orbit. Verge.io if you want something purpose-built for the mid-market VMware refugees. There is more oxygen in the "not VMware" market right now than there has been in fifteen years.
Why this specifically grinds
There are two things about the Broadcom situation that grind my yearn more than just the price hikes.
The first is the framing. Not the pricing. The messaging around the pricing. In the months after close, Broadcom's public statements about VMware customers used the word "innovation" a lot, the phrase "committed to our customers" a lot, and produced no operational change that matched those words. Anyone who had watched the CA and Symantec playbooks knew the mismatch was on purpose. Nine months of runway to sharpen the price hike letter, delivered under the cover of feel-good vocabulary.
The second is the collateral damage in the community. VMware wasn't just a company, it was an ecosystem. Independent bloggers, homelabbers, the folks running vExpert programs on their own time, the people who built entire careers writing tutorials for VMUG meetings. VMware, before Broadcom, was actually pretty good to those people. The free ESXi tier kept the enthusiast pipeline flowing. VMUG Advantage let people spin up licensed labs at home for a couple hundred bucks a year. That whole layer of ambient goodwill got wiped in a single decision cycle. Broadcom looked at the free tier and the enthusiast programs, ran the spreadsheet, and cancelled them because nothing on those lines paid back inside Broadcom's target return window.
That's the actual crime here. Broadcom didn't just raise prices, they killed the pipeline. In three years there will be a whole generation of sysadmins who never touched ESXi at home because it was never free to touch. That's fine, they'll learn something else, but it should be named as the loss it is.
The ripple that hit everyone else in the stack
The interesting second-order effect is what happened to everyone downstream of VMware. Backup vendors like Veeam, whose product line was built on VMware integration hooks, spent 2024 accelerating their support for Proxmox and Nutanix. The customer question mix visibly flipped over a couple of quarters, from "how do I optimize my VMware backup jobs" to "how do I extract my VMs from VMware into Proxmox".
Hardware vendors got hit sideways. HPE, Dell, Lenovo, all had reference architectures and joint SKUs with VMware. Every one of those reference architectures had to be recut for Proxmox, Nutanix, and OpenShift. Sales teams at those vendors that had been trained for a decade to lead with VMware in the conversation had to be retrained, and it showed.
Independent integrators and consultancies took the biggest ambient hit. There are shops that made their entire living implementing and supporting vSphere for mid-market customers. Some saw their pipeline dry up in a quarter because their customers were either exiting VMware entirely or bunkering down and refusing to spend on anything new until they figured out their exit plan. A lot of those integrators pivoted hard into Proxmox delivery and are doing fine now. The ones that didn't pivot are gone or on their way.
Every layer of the ecosystem that surrounded VMware got broken and had to be rebuilt for a post-VMware world. None of that shows up in Broadcom's earnings call. All of it shows up in a lot of small businesses that had to reinvent themselves in a hurry.
What to actually do
If you're still on VMware and you haven't started planning your exit, you're already behind. Not "behind" in the "you should have started six months ago" sense. Behind in the "your renewal is going to be a live-fire event and you have no leverage" sense.
Start by getting a real inventory. How many hosts. How many VMs. What flavors of storage. What backup product. Which vendor integrations. Once you have that written down, you have the shape of your migration.
Then pick a target. My honest bias: Proxmox VE for the shops with staff willing to learn KVM and comfortable at a shell. It's got the biggest community, the fastest release cadence, and the pricing is genuinely reasonable. XCP-ng if you have Citrix XenServer muscle memory floating around, since it's the same hypervisor under the hood. Nutanix if you have the budget for turnkey and you want the vendor to handle everything but you want the vendor to not be Broadcom. OpenShift Virtualization if your org already runs Red Hat and you want Kubernetes and VMs in the same operational plane. Verge.io if you want a purpose-built landing zone from a smaller vendor that's actively catching migration traffic and building specifically for the refugee use case.
Whatever you pick, do a real pilot. Not a demo. Not a POC that runs for two weeks and gets dismantled. A real pilot on a real workload for at least a quarter. Run it in parallel to your existing VMware footprint. Get your team comfortable. Get your monitoring and backup working against the new platform. Find the corners you didn't expect. Fix them.
Then plan the migration itself as a real project. Budget for the pain. Budget for the staff time. Budget for the temporary double-licensing period where you're paying both sides while the move is in flight. It will still be cheaper than the second year of Broadcom's new pricing.
Watch out for the license compliance audit angle. Broadcom's playbook at CA and Symantec included aggressive audit-and-true-up letters to customers, and there is no reason to expect a softer touch here. If you're on your way out, get your license position airtight before you start moving. Talk to a licensing specialist, not just your sales rep, because the sales rep works for the vendor and the specialist works for you.
And do not, under any circumstances, sign a multi-year deal in an attempt to lock in current pricing. Broadcom's post-close SKU rework already voided any pricing assumption a customer might have made pre-close. A multi-year deal just locks you in for longer to a vendor whose incentive is to keep extracting from you as long as they can. Every year of contract you sign is a year of exit-planning you cannot do.
The bigger lesson
The deeper thing here isn't about VMware or Broadcom. It's about relying on a single closed-source vendor for a piece of your infrastructure you can't operationally live without. That is a category of risk that a lot of shops accepted for a lot of years because VMware, specifically, felt safe. It felt like the vendor would always be around, always be reasonable, always keep the ecosystem alive.
Turns out the vendor was one M&A cycle away from being a hostage-taker.
Every closed-source vendor in your stack is a Broadcom acquisition away from doing the same thing. Every one. If you own the ability to run your infrastructure on someone else's schedule instead of your own, you own that risk. If you don't, you don't.
Get your data off VMware. Yesterday. And while you're at it, make a list of the other closed-source dependencies you have, and figure out what your exit path looks like for each one, before you have to find out on a Wednesday morning that your renewal quote just doubled.
It's time to YEARN!!!!!!!!
