Multi-Vendor AV Stack Ownership: Why Integrators Pay for the Gaps

You're on site at go-live. The displays are stable. The DSP is clean. The control processor is talking to the cloud licence server. Then, a codec firmware update breaks the meeting platform's camera handoff, the cloud room-management dashboard shows ten rooms offline even though nine of them are fine, and the client's IT lead wants to know why the signage CMS can't SSO into their new identity provider.

None of this is in your contract. All of it will take you a week to sort out. When the client writes a stiff Friday-afternoon email, your name is at the top — not the codec vendor's, not the cloud provider's, not the identity platform's.

This is what the AV stack actually looks like in 2026. Hardware, software, cloud services, identity, networks, and licensing portals. Pieces from half a dozen vendors, bound together by integrator labour and goodwill. Multi-vendor is the default. Ownership is the gap. And this gap is where projects die.

ISE 2026 confirmed the direction of travel. Florian Rotberg of invidis wrote in his ISE 2026 industry analysis that platform and operating system fragmentation is "as bad as ever — maybe worse." The halls in Barcelona were full of specialist vendors selling narrower, software-driven slices of what used to be an integrated system. That story plays out in your next project, on your invoice, and in your support queue.

Pro AV used to ship as a room. A rack of hardware from two or three brands, a control program written on-site, a commissioning sheet, done. The integrator owned the room, the hardware vendors owned their boxes, and the IT team owned the network drop. Everyone knew their lane.

That model is gone for any serious deployment. Rooms now depend on cloud codec services, subscription-licensed DSPs, SaaS signage platforms, API-linked room scheduling, cloud device management portals, and identity platforms the client's security team controls. The "room" is a node in a distributed system running across vendors, regions, and billing cycles.

Best-of-breed buying accelerated this. Enterprise clients no longer accept single-vendor stacks because they want camera X with codec Y with scheduling Z, and they write the spec that way. The supply side responded by building narrower products that assume they plug into everything else.

One ISE 2026 launch is a clean signal of where that's heading: a middleware platform whose entire business is abstracting away operating-system and hardware fragmentation in signage networks. When a company can raise money purely on gluing other people's products together, you know the stack has broken apart for real.

The result is a stack that nobody can draw a box around.

Multi-vendor works on the happy path. Every vendor assumes their product is the centre of the universe. Each writes documentation and support pathways for their own slice. The integrator is expected to stitch it together and carry the operational risk when the stitches show.

AVIXA Xchange's piece on converging AV, IT, and IoT for intelligent spaces flags this directly: organisational alignment, lifecycle management, and coordinated software updates are now the hard part of smart-environment design. Not the hardware. Not the protocols. The boring coordination work nobody sells a product for.

A UC room device stops joining meetings reliably after a firmware push. The root cause could be the platform, the device vendor's firmware, the identity provider's conditional access rules, the network team's proxy config, or the room's Bluetooth beacon. Five vendors. One problem. And the client's AV lead is phoning you, not them.

AVIXA's Recommended Practices for Security in Networked Audiovisual Systems already frames security as shared across end user, integrator, consultant, and manufacturer. That shared-responsibility model assumes everyone shows up. In fragmented stacks, most don't. Contracts are rarely written to force them.

Four failure zones now dominate post-handover support, and none of them are hardware.

1. Identity and access. SSO federations drift. Conditional access policies change. A client swaps identity providers, and the room scheduling platform stops authenticating. You didn't sell the identity platform, but you sold the rooms, so you're on the call.

2. Cloud service dependencies. A vendor's management portal has a regional outage, your monitoring goes blind, and your managed service SLA is on the hook even though your code is running fine. The client signed a contract with you, not with the vendor whose cloud fell over.

3. Licensing and entitlements. Subscription licences silently expire, software features disappear mid-meeting, and the client asks why the room "broke." The licence owner is sometimes you, sometimes the client, sometimes procurement. Nobody agrees until there's a failure.

4. Platform integration drift. Meeting platforms, signage CMS, and device management tools update on their own cadence. A feature you specced works on day one and breaks on day 200 because a vendor deprecated an API. You didn't do anything. You still own the fix.

ISE's own security advisor Pere Ferrer put it bluntly in his ISE 2026 interview: "cybersecurity is not an additional layer — it must be integral to the design." The same logic applies to ownership. You cannot bolt accountability onto a fragmented stack after the fact. You either design it in, or you end up carrying it.

European procurement has spent five years driving fragmentation deeper. The regulatory pressure is now closing around the integrator from the other direction.

Best-of-breed clauses are standard in UK public sector AV tenders, German enterprise frameworks, and Nordic estate-wide procurements. Buyers want vendor choice and the freedom to swap components. That means more interfaces and more surface area for things to drift.

The law is catching up — badly, for integrators. Germany's NIS2 national implementation entered force in December 2025. It makes regulated organisations responsible for supplier cybersecurity risk. You are now that supplier. The EU's revised Product Liability Directive 2024/2853, in force since December 2024 with transposition by December 2026, goes further. Integrators who substantially configure a networked product can be treated as manufacturers for liability purposes.

Buyers get fragmentation. You get accountability. That is not a sustainable commercial model.

AVIXA Xchange's ISE 2026 coverage on cybersecurity and AV-IT convergence caught the same split. One of its featured conversations — with Tara Dunning and Phil Langley — argued that services-led support is now the only sensible response to global deployments where AV, IT, security, wireless, and IoT collide on every project. Managed services is not an upsell. It is the business model that fragmentation forces.

DSS ISE 2026 treated this as its own dedicated panel topic: "Managed Signage: The Future of Digital Signage and ProAV." One panellist made the underlying point sharper: a modern tech stack is now a competitive asset, not a delivery detail. The platform companies that understand this are scaling. The integrators that don't are subsidising everyone else's SLAs.

Start with the contractual scope. If you're not the manufacturer for each product in the stack, say so in writing. Define which components you supply, which you integrate, and which the client procures directly. Write the RACI matrix into the statement of work: who is Responsible, Accountable, Consulted, and Informed for each layer — hardware, software, cloud services, identity, network, and content. Without that, you inherit everything by default.

Then separate the commissioning scope from the ongoing support scope. A handed-over system is not the same as a supported system. If the client wants you on the phone when a vendor's cloud dashboard misbehaves six months after sign-off, that's a managed service with a fee and an SLA, not an unpriced favour.

Price cross-vendor integration honestly. API mapping, identity federation, SSO testing, licence reconciliation, and platform drift management all take time. If you bury them in a fixed-price install, you're gifting the client a subsidy every time a vendor updates something. Break them out as visible line items tied to the deliverables they support.

Build a vendor risk exposure list and refuse the worst of it. Some manufacturers ship products that create obvious integration debt: opaque APIs, no EU legal entity, unpublished roadmaps, no vulnerability disclosure. Specifying them loads risk onto you. Cheaper boxes cost more in support hours. Price that correctly and walk away from the ones that can't be defended.

Finally, document the stack at handover like an asset register, not an installation report. Versions, licences, integration points, identity rules, update windows, and known drift risks. Clients facing NIS2 and PLD obligations now need that artefact. Supply it, and you become harder to replace. Skip it, and you become easier to blame.

Multi-vendor AV is not a passing phase. Buyers want it, tenders demand it, and the product economics reward vendors who build narrow, software-driven tools that plug in elsewhere. The fragmentation is permanent. ISE 2026 confirmed it. New middleware businesses are being built on it. Regulators are writing laws around it.

What's optional is whether you keep absorbing the cost of it. Integrators who treat the stack as a single system they're somehow meant to guarantee — with no contractual authority over the vendors in it, no margin priced for the integration labour, and no exit when a vendor behaves badly — are running a support charity funded by their own project margin.

Write the RACI matrix. Price the integration work. Refuse vendors who don’t meet basic supply-chain hygiene. Sell managed services as a priced product, not a relationship tax. And when a client wants a best-of-breed stack, show them the ownership map before you show them the quote.

The AV stack has broken apart because buying forces pulled it apart. The bill lands on you only if you let it.