2026 Engineering Playbook for Broadcast & Media: ST 2110, Latency Budgets, and Live Workflow Validation

Vadim Shilov

Head of Broadcasting & Telecom at Promwad

As broadcast and ProAV ecosystems move further into IP, cloud, and AI-assisted workflows, the competitive advantage for OEMs is shifting. It’s no longer enough to ship “a box that supports 4K.” Winning platforms in 2026 are defined by interoperability (ST 2110 + NMOS), measurable latency budgets, and software-defined video pipelines that can evolve after deployment. 

Ahead of ISE 2026 at Fira de Barcelona Gran Via, I want to share seven engineering trends I expect to dominate customer requirements—and to outline what they mean for broadcast equipment development roadmaps. 

At Promwad, we support broadcasting OEMs and operators with full-cycle embedded systems development—from architecture and prototyping to verification and production support—backed by 20+ years of experience and a 100+ engineer team.  

1. Cloud-native broadcasting & REMI → your devices must behave like managed IP nodes, not isolated appliances.
2. ST 2110 + NMOS by default → interoperability becomes a product feature (and a commissioning risk if you ignore it).
3. Ultra-low latency as a system requirement → latency budgets, not “best effort,” especially for live and interactive use cases.
4. AI-first production → automation shifts from “nice-to-have” to default workflow layer.
5. Vertical video for live & VOD → multi-aspect pipelines and auto-reframing move into professional production.
6. OTT/FAST keeps growing → ad-ready delivery + analytics hooks become mandatory across devices and apps.
7. Trusted media & deepfake protection → provenance and security move into broadcast engineering requirements.

1. Cloud-native + IP production (REMI is a consequence, not a feature)

Cloud playout, distributed editing, remote monitoring, and hybrid production aren’t separate “cloud projects” anymore—they are the operating model. For OEMs, this changes what customers expect from your hardware and software:

• Your device becomes a managed component in a larger IP system (observability, remote lifecycle, predictable updates).
• Resilience is not optional: failover behavior and real-time visibility matter as much as raw throughput.
• Interoperability drives purchasing decisions: integrators want commissioning predictability, not vendor lock-in.

From an engineering standpoint, this is where SMPTE ST 2110 integration becomes the backbone. SDI-to-IP migration is rarely a “lift-and-shift”; it is a controlled transformation where you must maintain on-air stability while transitioning signal transport, timing, and device discovery/control.

What this means for OEM roadmaps in 2026:

• Build an SDI→IP strategy that supports coexistence, not only “final state.”
• Treat NMOS as a first-class capability for discovery, connection management, and predictable integration. 
• Validate end-to-end behavior under realistic load—timing, jitter, packet loss, and failover—not just lab demos.

How we can help you drive this trend: seamless SDI-to-IP migration, ST 2110 + NMOS integration, performance validation, and commissioning-focused interoperability engineering.

Case in Brief: High-Speed OpenGear SDI-to-IP Сards (Multi-Camera)

For a European streaming operator, we developed FPGA-based OpenGear cards converting 8× 4K120 SDI streams to IP under SMPTE 2110, plus a high-bandwidth 10Gb NAS subsystem for real-time multi-camera workflows. This is the kind of “IP + performance + storage” architecture OEMs are increasingly asked to deliver as an integrated platform, not as separate boxes.

Read full case study: High-Speed OpenGear SDI-to-IP Cards for Multi-Camera Broadcasting

2. Ultra-low latency becomes product-critical (live, sports, interactive)

Latency used to be “a number” on the spec sheet. Now it’s a customer experience requirement—and a differentiator for OEMs building contribution, production, monitoring, and live distribution systems.

The trap: trying to fix latency late in the program. In modern IP pipelines, latency is a system property driven by choices across the entire chain:

• packet I/O and buffering strategy
• CPU/GPU/FPGA pipeline topology
• synchronization and timing boundaries
• protocol choices and acceleration layers

This is why we start with a defined latency budget and verify it end-to-end—under realistic load—before optimizing selectively. In practice, performance layers like DPDK or NVIDIA Rivermax can be the difference between “works in the lab” and “works in production at scale.” 

A practical OEM playbook for 2026:

• Measure first: instrument latency per stage (ingest → processing → transport → output).
• Design for zero surprises: avoid hidden buffering; document what adds frames.
• Pick the right acceleration point: not everything needs “maximum performance,” but your bottleneck must be intentional.
• Prove resilience: latency under stress (packet loss, failover, noisy networks) matters more than ideal-case numbers.

Where 5G fits in this story: 5G broadcast and ultra-low latency delivery will keep moving from pilots toward commercialization. For OEMs, the important takeaway is less the radio standard and more the engineering discipline: latency, power, and transport efficiency will be evaluated together—especially for mass events and live sports.

How we can help you drive this trend: ensure ultra-low-latency IP transport for live broadcast with Nvidia Rivermax and DPDK.

3. AI-first production & AI video analytics 

AI in broadcast is quickly moving from “experimental features” to the default production layer—especially in analytics, compliance, and quality control. For OEMs, the strategic shift is clear: customers want systems that reduce manual workload without compromising reliability.

The engineering question isn’t “Do we use AI?” It’s:

• Where does AI run? Edge/on-prem/cloud depending on latency, cost, and privacy.
• How do we trust AI output? Confidence scoring, explainability, human-in-the-loop controls.
• How do we operationalize it? Monitoring, retraining pipelines, and versioning.

This is where AI video analytics becomes a practical, shippable capability: automated scene/object/quality analysis that reduces operator routine across mixed-vendor ecosystems—from contribution to playout and monitoring.

How we can help you drive this trend: AI-powered content analytics embedded directly into your video pipeline—covering media content categorisation, advertisement filtering and personalisation, precise content segment retrieval and analysis, and sports analytics.

Case in Brief: Next-Gen Portable Live Streaming Device 

We upgraded a professional portable broadcast device with four LTE modems, additional HDMI/USB interfaces, a touchscreen UI, smart framing, and OTA software updates. 

A new battery increased runtime to 8 hours, supporting real-world live coverage needs. The client planned the first batch of 1,000 units—a good example of how AI-assisted features and updateability must be engineered into mass-produced devices, not bolted on later. 

Read full case study: Next-Gen Portable Live Streaming Device for Professional Broadcasting

4 more trends you can’t ignore

Vertical video for live & VOD (professional workflows go multi-aspect)

Vertical is no longer “social-only.” Live sports, concerts, and events increasingly require multi-aspect outputs with automatic reframing and synchronized multi-camera workflows. 

For OEMs, the product requirement is clear: support for multi-aspect encoding, metadata-driven framing, and predictable real-time performance—especially in portable and field-ready devices. 

Our portable streaming device upgrade mentioned above (smart framing + OTA + multi-modem connectivity) is a good example of how these requirements show up in real products.
 

OTT & FAST dominate viewing (monetization stacks become engineering requirements)

For OEMs building gateways, STBs, CTV devices, or enabling SDKs, OTT is no longer “someone else’s app.” Ad-ready delivery, analytics, and consistent playback behavior become part of the platform contract.

Case in Brief: WildTV Ecosystem (Android TV + Mobile Apps)

We delivered a white-label ecosystem (Android TV, iOS, Android) and the result improved monetization: 35,000+ downloads and 20% trial-to-paid conversion. For OEMs, this translates into practical requirements: observability, DRM-ready integration points, and performance stability at scale. 

Read full case study: Android TV Platform & Mobile App Development for Wildlife Content Producer
 

Modular content & automated versioning 

Content is increasingly assembled from reusable blocks—video segments, audio, captions, graphics—then re-versioned across CTV, mobile, and social outputs. OEM impact: you’ll be asked to support metadata-first pipelines, version-aware processing, and automation-friendly integration points. 
 

Trusted media & deepfake protection 

As AI-generated content becomes easier to produce and harder to verify, the demand for provenance, watermarking readiness, and real-time integrity checks will rise—especially in news, sports, and premium content. This trend is also reflected in broader ISE discussions about AI and cybersecurity in 2026.

OEM takeaway: design your pipeline so trust controls can be added without redesigning the entire platform.

OEM programs can fail for predictable reasons: unclear performance budgets, integration surprises, and late-stage rework. Our typical engagement model is built to de-risk delivery:

• Architecture & constraints: define throughput, latency, timing, interoperability, and update strategy.
• PoC / pilot: validate the hardest parts early (ST 2110/NMOS behavior, latency under load, AI module integration).
• Implementation: FPGA/video processing, embedded Linux/firmware, IP transport optimization, cloud modules where needed.
• Verification & production support: system-level testing, manufacturing readiness, and post-launch evolution.

The goal is simple: help OEM teams ship platforms where hardware, embedded software, and real-time networking work as one.

If you want a practical starting point (without rewriting your whole product line), here’s what I recommend doing in the next 90 days:

• Define your latency budget end-to-end and decide how you will measure it under load.
• Set interoperability targets: which ST 2110 profiles, which NMOS behaviors, and what “done” means for commissioning. 
• Decide where AI lives (edge/on-prem/cloud) for your top 1–2 automation use cases (analytics, QC, captions, compliance).
• Plan OTA + observability: product lifecycle is part of the value proposition now, not an afterthought. 
• Pick one PoC that de-risks your roadmap: SDI→IP migration slice, low-latency transport path, or an AI-assisted module.

Book a 24h Expert Call / Architecture Review and we’ll map the shortest path to interoperability, lower latency, and AI-ready pipelines.