Last year, my conversations with customers and partners revolved around the current state of the art, and the potential for AI to shape it. The tone was one of optimism and possibilities, despite the headwinds of global macro-economic pressures. There was a palpable sense that no one wanted to be left behind in the race for relevance in this new world.
The result, at least from the show floor I saw, was a lot of convergence. Many vendors seemed to be demonstrating variations of the same ideas. For example, several booths showed AI-assisted video search, where operators could find clips using natural-language prompts, while others demonstrated automated captioning, translation and metadata generation. The interfaces differed, but the underlying promise was often the same: use AI to make existing content easier to find, understand and repurpose.
There were plenty of impressive demoes. AI could do magic in seconds, but most of these capabilities lived in isolation. The demos worked because the boundaries were carefully controlled. The content was available, the metadata behaved, the network cooperated, and no production operator was standing nearby asking what would happen at 2 a.m. when one of the dependencies failed.
Real media pipelines are rarely that polite.
One short year later, I am gearing up for IBC 2026 with a different question. We already know AI can produce impressive demos. I’m walking into the conference looking forward to seeing how AI can earn a place inside production and distribution pipelines carrying real operational constraints.
This is a much harder challenge, and a much more meaningful test for video practitioners.
IBC 2026 takes place in Amsterdam from September 11–14. IBC expects more than 1,350 exhibitors, with participants from more than 170 countries. The programme spans content creation, distribution, monetization, Future Tech and the IBC Accelerator projects.
What interests me this year is whether emerging technology is ready to carry responsibility.
Here are the areas I will be watching.
1. AI will have to understand the workflow around it
A media workflow is a long chain of context.
What content is this? Who owns it? Where can it be distributed? Which language tracks are available? Is it live or on-demand? What happens when a rights window closes? Can a producer override the system? Can an operator explain why the AI made a decision?
An AI model working on a single clip can be useful. An AI system working inside production must understand the systems, metadata and people surrounding that clip.
The IBC 2026 show-floor agenda reflects this shift. Sessions cover agentic AI orchestration, media search and recommendations, near-live content creation, creative operations, federated retrieval, software-defined workflows and agentic AI for live sports. One session has a particularly candid title: “AI Won’t Make You Creative.”
That is a healthy direction.
The demonstrations I want to see will show where AI sits in the pipeline, what information it consumes, and how its output reaches the next system. I want to know what happens when metadata is incomplete, a recommendation is wrong, or a human needs to intervene during a live event.
A good demo shows what AI can do. A credible production system also shows what it does when AI gets things wrong.
2. Dynamic Media Facilities will meet day-two operations
Dynamic Media Facilities, or DMFs, are expected to be a prominent topic this year. The promise is attractive: production functions that once depended on fixed hardware can increasingly be instantiated as software, moved across compute environments, and reconfigured as production demands change.
That architectural flexibility can help teams respond to changing production needs. But flexibility moves complexity around. It rarely makes complexity disappear.
Who monitors a production running across on-premises infrastructure, cloud compute and remote contribution networks? Which team owns the incident when video is present in one part of the workflow and missing in another? How quickly can a production fail over? What does that failover cost when standby capacity is sitting idle?
These are day-two questions. Architecture diagrams often stop at day one.
TV NewsCheck’s IBC 2026 preview identifies AI and DMFs as major show themes, reporting broadcaster interest in operational efficiency, cloud workflows and connecting media-system data to AI models.
Sinclair CTO Mike Kralec captures the work ahead rather well: “It’s about data, context and integration, then us deriving value from that.”
Integration is where the real engineering begins.
3. Cloud and IP systems will need to coexist with what already works
I expect to see plenty of cloud compute, IP-based media systems and software-defined production at IBC. AMD’s IBC 2026 preview, for example, points toward software-defined, IP-based and AI-enabled media infrastructure, while also highlighting Accelerator work spanning immersive media, digital twins, 5G and next-generation live production.
Yet very few broadcasters have the luxury of replacing their entire production estate.
Cloud, IP, SDI and purpose-built appliances will continue to coexist. Sometimes that is a transition state. In many organizations, it may simply be the operating model for years.
The useful demoes will acknowledge this mixed environment. They will explain timing, synchronization, observability, security, transport, and the boundary between vendor systems. They will also explain the economics.
Cloud compute can make capacity available quickly, but live video has a habit of consuming network, storage and compute at the same time. A system can be technically elastic and still become financially difficult to operate.
I will be asking vendors a simple question: what happens to the bill when production capacity or distribution demand doubles unexpectedly?
4. Resilience and security will have to show up in the product showcases
Media systems used to have clearer boundaries. Production equipment lived in controlled facilities, contribution paths were specialized, and the number of people or services touching a signal was limited.
Software-defined and connected workflows change that boundary. More APIs, remote operators, cloud services, AI models and third-party integrations also create more places where a failure or security problem can enter.
So I expect resilience and security to move closer to the product demo itself.
Can the service continue when a cloud region becomes unavailable? Can an operator isolate a compromised component without taking down the production? Are AI-generated actions recorded and auditable? Can credentials and media rights be scoped to the minimum access required?
A security slide at the end of a presentation will not answer these questions. The design needs to show its failure paths.
For live sports, this matters deeply. Customers do not care which service failed behind the scenes. They care that they missed the final over, the winning goal, or the decisive play.
The infrastructure may be complex. But the customer’s verdict is simple; did the stream work?
5. MOQ has to prove where it belongs
Last year, Media over QUIC kept appearing in serious video-delivery conversations I had at IBC. That was my experience rather than a claim about the overall conference agenda.
This year, MOQ has moved into a more interesting part of the programme, with sessions on real-world deployments, low-latency delivery and the future of internet television.
This topic is personal for me. We have been building Oracle Video Edge with MOQ at its heart, demonstrating workflows spanning contribution ingest, relay and CDN distribution, browser playback, and transitions into CMAF and HLS. The ecosystem is exploring many of the same challenges: protected delivery, caching, authorization, observability and interoperability.
I am also interested in MOQ’s potential for interactive media. Its publish/subscribe model over QUIC or WebTransport supports low-latency exchange of media and data in both directions, enabling viewer input, timed metadata and alternate views to travel alongside video.
MOQT is still evolving, with related work continuing around media formats, authorization and security. That makes interoperability especially important: which draft is being used, how does MOQ fit existing workflows, and can vendors interoperate without coordinating every detail?
A protocol earns its place when teams can operate, troubleshoot and trust it under load. That is the evidence I hope to find in Amsterdam.
The test I will apply on the show floor
For every AI, cloud, DMF or media-transport demonstration, I plan to ask five questions:
- Where does this capability sit in the production or distribution workflow?
- What data, context and integration does it require?
- What happens when a dependency fails?
- Who owns the system and its operating cost?
- Can it coexist with the infrastructure customers already have?
IBC has always been a good place to see what the media industry wants to become. In 2025, I saw plenty of optimism and an understandable rush to participate in AI. Some ideas were promising. Many looked similar.
IBC 2026 has the opportunity to move the conversation forward.
The winners will show their technology carrying actual responsibility inside messy workflows, across mixed infrastructure, with security controls, human operators, unpredictable networks and customers waiting at the other end.
Demos create possibilities. Operations measure impact.
This year, I will be looking for both.