Every VCF 9 core you buy from Broadcom now carries 1 TiB of vSAN capacity entitlement, pooled across your clusters. That single licensing detail has quietly rewritten enterprise storage strategy: vSAN is no longer a product you evaluate — it’s a sunk entitlement you either consume or waste. I’ve sat in three renewal meetings this quarter where the storage decision and the hypervisor decision collapsed into one line item, and teams discovered they were arguing about both at once without realizing it.
This brief maps the realistic vSAN alternatives in 2026: staying HCI on vSAN or Nutanix, or decoupling storage onto external arrays from Pure Storage, NetApp, or Dell — arrays that, as of this year, finally speak more than one hypervisor’s language. The core argument: pick storage for data gravity and exit optionality, not for whichever hypervisor you happen to run this year.
What changed
Broadcom’s VCF 9 bundling made vSAN capacity a per-core entitlement rather than a separate purchase. Each VCF core includes 1 TiB of vSAN entitlement; vSphere Foundation (VVF) includes 0.25 TiB per core. The entitlement pools across clusters, and you buy add-on TiBs only above the included amount. Note that entitlement TiBs are raw licensing capacity — usable capacity after FTT and RAID overhead lands meaningfully lower, so budget accordingly.
Meanwhile, the external-storage vendors stopped being VMware-only citizens. Nutanix shipped its Pure Storage FlashArray integration to general availability in December 2025 and extended it to more FlashArray models at .NEXT 2026, announced an NFS-based alliance with NetApp for AFF and select FAS systems, and already supported Dell PowerFlex for compute-only nodes. NetApp and Pure have both pushed integrations toward Proxmox and OpenShift as well. External arrays are now multi-hypervisor assets in practice, not just on the roadmap slide. And the clock is real: vSphere 8’s 2027 end of support means most shops must land somewhere new within roughly 15 months.
Why it matters
Storage is where the switching costs actually live. Compute is stateless — you can vMotion, rebuild, or re-image your way off a hypervisor in a quarter. Petabytes are not. If your data sits inside vSAN datastores, your data follows your VMware licensing decision whether you like it or not. If it sits on an external array with multi-hypervisor support, the hypervisor becomes a two-year decision instead of a ten-year one.
The entitlement math cuts both ways. If you’re staying on VCF at your current core count, the included vSAN TiBs make external arrays look expensive — you’d be paying twice for capacity you already own. If you’re shrinking your VMware footprint, every TiB stranded in vSAN is a hostage in the renewal negotiation. The takeaway a VP can repeat: vSAN’s price is now a function of your Broadcom relationship, not of storage economics.
Path 1: Stay HCI — vSAN under VCF, or Nutanix
vSAN ESA remains a genuinely good storage product — strong NVMe performance, mature operations, and effectively “free” up to your pooled entitlement if you’re committed to VCF anyway. For a shop that has made peace with Broadcom pricing and runs a stable, VMware-standardized estate, consuming the entitlement is the rational move. The risk isn’t technical; it’s that your storage exit cost compounds every year you deepen the dependency.
Nutanix is the other HCI answer, and 2026’s twist is that it’s no longer HCI-only. AHV plus AOS gives you a vSAN-like operational model with a different licensing counterparty — and Nutanix now lets you attach external storage rather than forcing everything into its own distributed fabric. We analyzed that shift in depth in our brief on Nutanix external storage as a VMware exit ramp. Nutanix fits shops that want the HCI operating model without the Broadcom invoice; it does not fit shops with heavy vSphere-specific tooling they aren’t ready to rewrite, and its migration still consumes real project quarters.
Path 2: Decouple onto external arrays
The decoupling case rests on one 2026 fact: the major array vendors now integrate with the hypervisors people are actually migrating to.
Pure Storage (now Everpure) FlashArray is the most aggressive mover. The Nutanix integration — GA since December 2025 — maps every AHV virtual disk to its own array volume, so per-VM snapshots, QoS, and replication work at array speed, and .NEXT 2026 extended support beyond the //X and //XL models. Evergreen subscriptions mean the array outlives multiple hypervisor generations. Strong fit: shops that want one storage platform under vSphere today and AHV or OpenShift tomorrow. Weak spot: file services lag NetApp’s, and FlashArray pricing sits at the premium end — list pricing varies, but you’re paying for the operational model.
NetApp AFF and ASA counter with breadth. ONTAP speaks NFS, iSCSI, NVMe-oF, and SMB from one system, the new Nutanix alliance runs over NFS to AFF A-series and select FAS, and NetApp’s Proxmox and OpenShift tooling is credible. If your estate mixes VM storage with large file workloads, NetApp consolidates what Pure would split across products. The cost is complexity: ONTAP’s flexibility comes with a learning curve that flat-out annoys teams accustomed to FlashArray’s minimalism. Our Pure vs NetApp comparison breaks down where each wins.
Dell PowerFlex is the scale-out option — software-defined block storage that disaggregates compute from capacity and was the first external storage Nutanix certified for compute-only nodes. It shines at large scale (hundreds of nodes, mixed hypervisors, database-heavy estates) and inside Dell-standardized shops. Below roughly a half-petabyte, its operational overhead outweighs the benefit, and Dell’s broader portfolio (PowerStore, PowerMax) can make the sizing conversation feel like a menu with no prices.
Side-by-side: the three storage paths
| vSAN under VCF | Nutanix AOS | External arrays (Pure, NetApp, Dell) | |
|---|---|---|---|
| Best fit | Committed VCF shops consuming the TiB entitlement | HCI operating model without Broadcom | Data gravity, mixed or changing hypervisors |
| Hypervisor coupling | Total — vSAN is vSphere-only | High, but AHV is the destination for many | Low — vSphere, AHV, Proxmox, OpenShift |
| Licensing model | 1 TiB/core included in VCF; add-on above | Per-core/per-node subscription | Capex or storage-as-a-service subscription |
| Exit optionality | Weakest — data moves when you leave | Moderate — external storage support helps | Strongest — array outlives the hypervisor |
| Watch out for | Entitlement TiBs are raw, not usable | Migration effort, vSphere tooling rewrite | Paying for capacity VCF already entitles |
The honest downsides of each path
Staying on vSAN means accepting that your storage exit cost grows with every TiB you land there, and that Broadcom — not you — controls the future price of that capacity. It also means your DR and data-protection tooling stays vSphere-shaped.
Decoupling isn’t free either. If you hold a VCF subscription, external arrays mean paying for storage twice until the next renewal. Array-based estates need SAN or dedicated Ethernet fabric skills that many HCI-era teams let atrophy. And multi-hypervisor integrations are young: the Nutanix–NetApp path is NFS-only today, and the FlashArray–AHV integration, while GA, has fewer production miles than a decade of vSphere plugins. Nutanix as a middle path carries its own migration tax. Nobody exits this decision clean; you’re choosing which costs you can live with.
What to do about it
Three rules of thumb I’d defend in any architecture review:
- If you’re renewing VCF at flat or growing core counts: consume the vSAN entitlement, but keep new data-heavy workloads (databases, file shares, analytics) off vSAN so the hostage doesn’t grow.
- If you’re shrinking VMware by 30% or more: move storage first. Land data on an external array with AHV and Proxmox support before you migrate compute — data gravity is the exit blocker, not the hypervisor.
- If you’re undecided: refuse any storage purchase that locks to a single hypervisor. As of mid-2026, multi-hypervisor support costs little to demand and preserves the negotiating leverage you’ll want at the next Broadcom renewal.
The one-liner for the steering committee: storage decisions now outlive hypervisor decisions, so make the storage call for the estate you’ll run in 2030, not the one you licensed in 2024.
Frequently asked questions
Is vSAN included with VCF 9?
Yes. VCF 9 includes 1 TiB of vSAN capacity entitlement per licensed core, pooled across clusters and VCF Operations instances. VVF includes 0.25 TiB per core. Capacity beyond the entitlement is a paid add-on, and entitlement TiBs are raw — usable capacity depends on FTT and RAID choices.
What is the best alternative to vSAN in 2026?
It depends on your exit posture. Committed VCF shops should usually consume the entitlement. Shops leaving VMware lean toward Nutanix AOS for a like-for-like HCI model, or external arrays — Pure FlashArray, NetApp AFF/ASA, Dell PowerFlex — when data gravity and multi-hypervisor flexibility matter more than operational simplicity.
Can Nutanix use external storage instead of its own HCI storage?
Yes, as of late 2025. Nutanix supports Pure Storage FlashArray (GA December 2025, expanded at .NEXT 2026), NetApp AFF and select FAS over NFS, and Dell PowerFlex for compute-only nodes. That breaks the old rule that choosing Nutanix meant choosing Nutanix storage.
Is HCI still cheaper than a SAN?
Not reliably, and the question has changed. With vSAN bundled into VCF cores, its cost is a licensing artifact rather than a market price. External arrays cost more upfront but decouple storage refresh cycles from hypervisor licensing — which is where the money actually moves over a five-to-seven-year horizon.
Enterprise Techie publishes vendor-honest analysis like this daily — get the brief by email, free.