Citrix vs VMware Horizon: 2026 Comparison

The end-user computing (EUC) market changed shape in 2024 when Broadcom divested VMware EUC — including Horizon — to KKR, which rebranded the unit as Omnissa. That separation matters: Horizon is no longer a Broadcom product and is no longer subject to Broadcom's VCF bundle pricing or audit posture. It is, however, still a VMware-heritage product running in many of the same data centres, and customers comparing it to Citrix in 2026 face a meaningfully different decision than they did in 2023.

This article compares Citrix (now owned by Cloud Software Group / TIBCO) and Omnissa Horizon across the dimensions that actually move procurement decisions: licensing, cost-per-seat, feature parity, operational complexity, and the long-term roadmap. It is written for IT leaders evaluating a strategic EUC replacement, not for engineers comparing protocol latency.

Where each platform stands in 2026

Citrix has spent the last three years consolidating its product line, retiring the on-premises XenApp / XenDesktop split, and pushing customers toward Citrix DaaS (cloud-delivered control plane) plus Citrix Universal Subscription. The buying experience is now subscription-only for new customers; existing perpetual customers are being routed toward subscription at renewal. Citrix's strengths remain protocol performance (HDX), strong session brokering at scale, and a mature partner ecosystem.

Horizon, under Omnissa, has been re-positioned as the "anti-Broadcom" EUC option for VMware-heritage customers. Omnissa continues to develop the Horizon product line, has reintroduced a perpetual licence option for select customers, and is investing heavily in the Workspace ONE unified endpoint management integration. Horizon's strengths remain instant-clone provisioning at scale, vSphere-native integration where vSphere is still in place, and the App Volumes / Dynamic Environment Manager profile management stack.

Licensing models compared

Citrix Universal Subscription is sold per user-per-month, with the higher tiers including session recording, web app firewall, and analytics. Published list prices for Citrix DaaS Advanced Plus run around $25-$32 per user per month for enterprise commits; Premium Plus runs higher. Real enterprise pricing varies widely based on commit size and Cloud Software Group's appetite to displace Horizon.

Omnissa Horizon is now sold both as subscription (Horizon Universal Subscription, per named user) and — for existing customers — as a perpetual licence with annual support. Horizon Universal subscription typically lands in the $14-$22 per named user per month range for mid-size deals, with Workspace ONE bundled in higher tiers. The perpetual option is heavily restricted; new customers will generally not have access to it.

Cost per seat — realistic comparison

For a hypothetical 5,000-seat deployment over a five-year horizon, our 2026 engagement data shows total cost of ownership landing roughly as follows. These are TCO figures including licences, infrastructure, professional services, and operations — not just licence price.

• Citrix DaaS Advanced Plus, 5,000 named users — $1,650-$1,950 per seat over 5 years, mostly licence + control plane
• Omnissa Horizon Universal, 5,000 named users (on existing vSphere) — $1,300-$1,600 per seat over 5 years
• Omnissa Horizon Universal, 5,000 named users (on Nutanix AHV or Hyper-V) — $1,450-$1,750 per seat over 5 years (hypervisor cost varies)

The Horizon advantage shrinks when the hypervisor underneath Horizon is itself a paid-for VMware vSphere subscription. For customers running Horizon on Broadcom-licensed vSphere, the combined cost approaches or exceeds Citrix DaaS, which removes the strongest single argument for Horizon. Many Horizon customers are now evaluating a parallel hypervisor migration to break that dependency.

Feature parity

The two platforms have converged on the major capabilities. Both support persistent and non-persistent desktops, published applications, multi-session Windows 11, profile management, and GPU-accelerated workloads. The differentiators that remain are these:

• Protocol — Citrix HDX still has a measurable edge over Horizon Blast Extreme on high-latency or low-bandwidth networks, particularly for graphics-heavy workloads. The gap has narrowed in the last two years but has not closed.
• Image management — Horizon's instant-clone technology, when running on vSphere, remains best-in-class for rapid non-persistent desktop provisioning. Citrix MCS and PVS are competitive but operationally more complex.
• Unified endpoint management — Horizon's integration with Workspace ONE is tighter than Citrix's equivalent with Citrix Endpoint Management. For customers committed to a UEM strategy, this is meaningful.
• Cloud control plane — Citrix DaaS has been cloud-native longer and is more mature on the control plane side. Horizon Universal is catching up but is not at parity.
• Multi-cloud — Citrix is more polyglot on cloud hosting (Azure, AWS, GCP, on-prem all first-class). Horizon Universal supports Azure and on-prem; AWS and GCP support exists but is less mature.

Operational complexity

Citrix is the more operationally complex platform. The component set — StoreFront, Delivery Controllers, Provisioning Services, Profile Management, NetScaler, Cloud Connector — produces a larger surface area and a longer learning curve. Horizon's component set is smaller (Connection Server, Composer or Instant Clone, App Volumes, UAG) and the integration model is more cohesive.

This matters for teams that are short on EUC specialists. A small operations team is more likely to keep Horizon running cleanly than Citrix. A large operations team with dedicated Citrix expertise can extract more performance and capability from Citrix.

The audit dimension

One under-discussed factor in the Citrix vs Horizon decision is audit exposure. Citrix has historically been a measured auditor — soft audits, rare formal audits, and a generally commercial settlement posture. Horizon, while now under Omnissa rather than Broadcom, inherits VMware's auditing legacy: Broadcom retained the right to audit VMware-licensed infrastructure that underpins Horizon deployments. A customer running Horizon on Broadcom-licensed vSphere is therefore in scope for a Broadcom vSphere audit, even though Horizon itself is not Broadcom-owned.

This is the most-missed nuance in 2026 EUC procurement. Customers assume that moving to Horizon-on-Omnissa removes them from Broadcom's audit perimeter. It does not. Only a parallel hypervisor migration does.

Decision framework

For most enterprises in 2026 the decision falls one of three ways:

1. Existing Horizon customers with healthy vSphere relationships — stay on Horizon, renew with Omnissa, and revisit the question at next renewal. Disruption cost of moving to Citrix outweighs the marginal cost difference.
2. Existing Horizon customers facing painful Broadcom vSphere pricing — evaluate a parallel hypervisor migration (Nutanix AHV or Hyper-V) and either keep Horizon on the new hypervisor or, if a brownfield rebuild is acceptable, take the opportunity to move to Citrix.
3. Existing Citrix customers — stay on Citrix. The only reason to move would be a strategic Workspace ONE deployment, and that case is not strong enough on its own.

Greenfield EUC deployments in 2026 are split roughly evenly between Citrix and Horizon in our advisory data, with the choice driven mostly by existing team skills rather than product capability.

Bottom line

The Citrix vs Horizon decision in 2026 is less about features (which have converged) and more about three things: the cost of the hypervisor underneath, the operational maturity of the team, and the strategic direction of the EUC platform vendor itself. Broadcom no longer owns Horizon, but Broadcom can still audit the vSphere environment that supports it — and that fact reshapes the procurement decision more than any specification-sheet comparison.

Specific deployment patterns and which platform fits

Persistent desktops for knowledge workers

Both platforms handle this well. Horizon has a small edge on user experience due to mature instant-clone technology and tighter integration with Workspace ONE for profile delivery. Citrix has a small edge on protocol performance for users on poor networks. For most enterprises, either platform will deliver a high-quality persistent desktop experience — the choice should be made on operational fit, not technical capability.

Non-persistent desktops at scale

Horizon's instant-clone provisioning is the strongest single capability in this segment when running on vSphere. The non-persistent desktop boots in 15-30 seconds from a small set of golden images, with App Volumes layering applications on demand. Citrix MCS and PVS are competitive but operationally more complex and typically slower to provision at scale. For customers building a large non-persistent EUC environment from scratch in 2026, Horizon-on-vSphere remains the technical leader — subject to the cost-of-hypervisor question discussed above.

Published applications

Citrix has been the market leader in published-application delivery since the XenApp era and the lead is preserved in Citrix DaaS. The session-broker scaling, multi-session Windows support, and application-publishing tooling are more mature than Horizon's equivalents. For enterprises whose use case is primarily application publishing rather than full desktop delivery, Citrix is generally the better fit.

GPU-accelerated workloads (CAD, GIS, 3D)

Both platforms support GPU pass-through and GPU virtualisation, but the operational maturity differs. Horizon has historically been the platform of choice for high-fidelity GPU-accelerated workloads in design and engineering environments, with a substantial installed base and well-understood deployment patterns. Citrix has improved but is behind. For greenfield GPU-heavy EUC deployments, Horizon is the safer technical choice.

BYOD and contractor access

Citrix Workspace is more polyglot on client-device support and more mature on access-from-anywhere scenarios. For workforce models that emphasise BYOD, partner access, or short-term contractor access, Citrix is generally the better fit. Horizon Universal has improved on this dimension but is not at parity.

Migration cost between platforms

A migration from Citrix to Horizon, or from Horizon to Citrix, is a substantial project. Application packaging changes (App-V or App Volumes vs Citrix layering), profile management changes (DEM vs Citrix Profile Management), and user training all add cost. Typical migration cost runs $200-$450 per seat for a brownfield platform replacement, before counting the new platform's licence cost.

For most enterprises, that cost is too high to justify a platform change for marginal reasons. The migration case is only compelling when there is a strategic reason — the existing platform vendor has become commercially unworkable, the team skills favour the alternative, or the broader IT strategy (Workspace ONE adoption, multi-cloud commitment) tilts the balance.

Three-year outlook

Both Citrix and Omnissa are mid-cap private-equity-controlled software businesses. Both will continue to invest in their EUC products, but neither will operate as an open-ended R&D budget. Customers should expect both vendors to continue commercialising aggressively: subscription pricing pressure, bundle escalation, audit enforcement on perpetual customers, and progressive end-of-life of older product editions. The choice between them is less about which vendor is the "good guy" and more about which one provides the better technical fit for the customer's specific deployment.

One scenario worth monitoring: a future Broadcom re-engagement with the EUC space. Broadcom retains substantial EUC-adjacent IP through its VMware Cloud Foundation product set, and the divestiture of Horizon to Omnissa did not foreclose Broadcom's ability to re-enter the segment. A re-entry would change the competitive dynamics materially, and customers with long EUC procurement horizons should track this.

Frequently asked EUC procurement questions

If Broadcom doesn't own Horizon, why is Horizon licensing still relevant to Broadcom audit defence?

Because the hypervisor underneath Horizon is, in most existing deployments, Broadcom-licensed vSphere. The Horizon licence with Omnissa is one contract; the vSphere licence with Broadcom is another. A Broadcom audit on the vSphere environment sees the hosts that support the Horizon environment and applies the same audit posture as any other vSphere deployment. Customers who assume that "moving to Horizon under Omnissa" removes them from the Broadcom audit perimeter are mistaken.

Is now the right time to make a strategic EUC change?

For most customers, no. EUC platform changes are 18-30 month programmes with substantial change-management overhead. Unless there is a strategic driver (cost, capability, integration), the right posture is to stay on the current platform, evaluate at next renewal, and avoid making the EUC decision under duress. The exception is customers whose current platform vendor has become commercially unreasonable, where the displacement decision is essentially forced.

What about cloud-only EUC alternatives like AWS WorkSpaces or Azure Virtual Desktop?

Both are credible for specific use cases, particularly burst-capacity scenarios and SMB-scale deployments. For full-scale enterprise EUC with persistent desktops, GPU workloads, and granular control over the desktop environment, neither yet matches Citrix or Horizon at parity. Azure Virtual Desktop is the closer of the two but has gaps on profile management and application delivery. Customers with predominantly knowledge-worker, low-customisation desktops can credibly consider AVD as a third option.

What is the most-common mistake in EUC platform selection?

Optimising for unit licence price without modelling total cost of ownership including hypervisor cost, infrastructure cost, professional services cost, and ongoing operations cost. The licence is the visible cost; the hypervisor underneath and the operations team that runs the platform are the larger costs. A platform decision driven by per-seat licence comparison alone almost always produces a result that does not survive the second year of operations.