From 16 to 72 to 96 Cores: Inside the Licensing Maze VMware Shops Are Trying to Navigate

For years, VMware licensing existed in that weird corner of enterprise IT where nobody felt great about it, but most folks at least knew the rules. You could size a cluster, run the numbers, make a case for your budget, and walk away without feeling like you'd accidentally stepped on a financial landmine. But the moment Broadcom took the wheel, that sense of predictability evaporated. The confusion didn't hit all at once. It started as a ripple — a stray conversation between admins comparing notes, a reseller hinting that new minimums were "coming soon," a support engineer who wasn't allowed to say more but clearly wanted to. And then the stories multiplied. One engineer running a modest 24-core setup said he'd been told he needed to buy licensing for 72 cores anyway. Another claimed Broadcom's own docs still said the minimum was only 16 cores per CPU socket. Someone else said they'd heard 96 was the floor for version 8 stacks. The tone wasn't panic so much as fatigue: seasoned infrastructure folks trying to understand rules that seemed to shift depending on who they talked to. And underneath it all was something familiar — the sense that smaller customers were being pushed into a corner where they didn't really get to decide anything anymore. It wasn't just about licensing. It was about whether they were still welcome. ## The 16-Core Rule That Somehow Doesn't Answer Anything If you open Broadcom's official documentation today, the rule looks almost friendly: **16 cores per physical CPU is the licensing minimum.** Doesn't matter if the CPU has 6 cores or 64 — you cover at least 16, and you move on. That sounds simple. Clean. Predictable. Except that's not the number VMware customers are talking about. Admins in the trenches say the **minimum is effectively 72 cores per purchase**, regardless of how small your hardware footprint is. If your servers only have 24 cores each, too bad — you're paying for 72. And if you ask why the 16-core-per-socket rule exists at all when the bundle minimum overrides it, the answer usually sounds like: "Well… it depends." Which is the last thing anyone wants to hear when budgeting a six-figure renewal. This is the core frustration running through people's comments. Broadcom's paperwork still lists the 16-core definition, but people dealing with quotes and renewals say they're living in a different world entirely — one where the floor is set far higher, and nobody from Broadcom is exactly eager to write that number down. ## Where 72 Became the New Normal The 72-core minimum didn't come out of nowhere. Broadcom ditched VMware's traditional a-la-carte licensing model and replaced it with **subscription bundles**. That move alone forced thousands of customers to relearn how VMware pricing works. But hidden inside those bundles is a key assumption: Broadcom doesn't want tiny installations. Their sales model is optimized around customers who run VMware in bulk. That's why people in the thread said things like: - "72 Core is the minimum purchase." - "It is 72c minimum even if all you got is 24 cores." - "You are too small for them, so they are making sure it is worth their time to sell to you." That last line captures the new vibe perfectly. VMware used to scale down gracefully. Small two-node shops, test clusters, edge deployments — they weren't exactly Broadcom's dream accounts, but VMware let them exist. Now the sentiment is different. If your footprint is under 72 cores, the message is essentially, *you can buy 72 cores or you can go somewhere else.* As one person noted, they were able to get the minimum waived during a renewal, which means the number is negotiable — but only if you're a customer Broadcom wants to keep, and only if your rep is motivated to make the exception. That alone tells you how much the model has changed. ## Then Came the 96-Core Surprise Just when the community was starting to understand the 72-core minimum, another number appeared: **96 cores.** One commenter said: - "I thought it was 96 core minimum as of version 8?" - "Yes. So VMware is 72 minimum but Broadcom now is making it 96. Just ran into this 3 weeks ago with new VMware stack on version 8." That's the kind of detail that makes admins pause. If the minimum jumps from 72 to 96 depending on whether you're running vSphere 7 or 8, that's not just a policy change — that's a licensing cliff. The difference affects: - Small-scale on-prem clusters - DR sites - Remote offices - Customers who deliberately choose CPUs with lower core counts - Edge deployments that don't need huge compute footprints And because Broadcom hasn't clearly published anything confirming the 96-core minimum, people are left trading field reports like they're decoding a mystery. It's not an exaggeration to say Broadcom has created a situation where unofficial, crowdsourced knowledge is now more detailed and accurate than official documentation. That's never a good sign. ## The EEA Exception That Only Makes Things Messier A surprising twist in the discussion came from someone pointing out that **the 72-core minimum doesn't apply everywhere.** One user said: - "If you are in the EER, then no. For the rest of the world - yes." Another clarified: - "EER or more commonly EEA is the European economic area… They have different price books for EEA, EMEA NON EEA, China, and rest of world due to legal differences." This matters because it introduces a new layer of complexity: **where your business operates can determine how painful your VMware bill becomes.** If you're in the EEA, local regulations around software licensing and fair-competition practices appear to limit Broadcom's ability to enforce the 72-core minimum. So a customer in France or Germany might get a smaller minimum than someone in the US — not for technical reasons, but because regulatory frameworks are different. That's the kind of inconsistency that drives IT teams up the wall. Two companies with identical hardware can end up with wildly different licensing bills simply because of geography. ## The Customer Perspective: Death by Uncertainty What stands out most as you read through people's comments isn't anger — it's exhaustion. IT teams depend on predictability. They need stable licensing rules to plan hardware refresh cycles, budget for renewals, and justify decisions to executives who don't care about core counts but do care about cost. Broadcom's model has replaced that predictability with ambiguity. People aren't saying the new pricing is high — they're saying they don't even know what the real pricing *is*. They can quote a Broadcom doc that says 16 cores per socket. They can quote a salesperson saying 72 cores per cluster purchase. Now they can quote someone else saying 96 cores minimum for vSphere 8. These numbers aren't just confusing. They shape real decisions: - Whether to delay a hardware refresh - Whether to stay on vSphere 7 to avoid new minimums - Whether to jump to Proxmox, Nutanix, or Hyper-V - Whether to push workloads to the cloud to dodge on-prem licensing entirely - Whether to drop VMware slowly, cluster by cluster, before renewal day arrives When a pricing model becomes unpredictable, companies stop expanding. They freeze. They strategize. They look for exit paths. That's exactly what's happening here. ## Why the Core Minimums Hit Harder Than Broadcom Expected Most enterprise vendors raise prices. It's part of the game. But the way they do it matters. Broadcom didn't just increase costs — they rebuilt the entire structure. Subscriptions replaced perpetuals. Bundles replaced standalone products. Sales minimums replaced granular licensing. And core minimums were set high enough that small and mid-sized VMware shops suddenly found themselves priced out of a platform they'd been using for years. The message, intentional or not, reads like: *"We want bigger customers, and we're fine losing the smaller ones."* From a pure revenue perspective, maybe that makes sense. From a community perspective, it creates a rift. VMware spent two decades becoming the default virtualization layer for everyone — from tiny two-node deployments to sprawling enterprise clouds. Broadcom seems comfortable walking away from the bottom 30% of that market. But here's the thing: those smaller customers aren't quiet. They talk. They share experiences. They influence peers, friends, and partners. And their stories travel fast. Too many of those stories right now sound like this: - "We can't get a straight answer." - "The documentation doesn't match what our rep told us." - "We're not big enough for them to care." - "We're preparing to migrate off VMware." That's not the kind of energy you want swirling around your flagship virtualization platform. ## Where This All Goes Next The licensing minimums probably aren't going away. Broadcom is clearly optimizing VMware for larger, more predictable, higher-revenue customers. But the lack of clarity is becoming its own problem. Companies can handle change. What they can't handle is ambiguity. If Broadcom wants to stabilize the customer base, they'll need to publish clear, global documentation for: - Minimum core purchases - How those minimums vary by region - Which versions (v7 vs v8) are tied to which minimums - Whether waivers exist and who qualifies - How small deployments are supposed to operate under the new model Until that happens, customers will keep relying on each other to figure out what's real. And that's how we end up with a world where "16 cores," "72 cores," and "96 cores" are all simultaneously correct, depending on who you talk to. ## The Bottom Line What started as a simple question — *Is there still a 72-core purchase minimum?* — reveals something deeper. VMware admins aren't just asking about numbers. They're trying to understand whether VMware still fits into their future. Right now, the answers feel murky. Between the move to subscriptions, the aggressive core minimums, the regional exceptions, and the conflicting stories from the field, VMware's licensing model has become a puzzle people resent having to solve. And when a platform's licensing is harder to understand than the technology itself, customers start weighing their options. That's the real tension here. Not the 72 cores. Not the 96. But the sense that VMware is shifting into a different era — one where the company decides how much you can buy, not the other way around. And customers are deciding how long they're willing to play along.