Software-defined production isn’t a “maybe” anymore. Teams are under pressure to deliver more content, faster and across more platforms than ever before – all while reducing operating costs – the industry has crossed a threshold.

The real question now is simpler and harder: how to scale software-defined production, without reintroducing the complexity it was meant to eliminate?

Fortunately, two initiatives are emerging as practical answers:

  • JT-DMF: a reference architecture for how dynamic facilities should be structured, governed, and operated across on-prem, cloud, and hybrid environments.
  • MXL: a high-performance media exchange approach that enables software applications to truly collaborate, not just connect.

Together, they point to what the industry needs next: a shared set of principles that enables predictable interoperability, clear rules of the road for scaling software-defined production.

Importantly, this isn’t a vendor-only approach. Both JT-DMF and MXL are being shaped with customer advisory participation, supported through organizations including EBU, NABA, AMWA, and leadership from CBC/Radio Canada to ensure the outcomes reflects the real operational needs, not just technical ideals.

Why common principles matter now

Over the last decade, the industry has progressed from turnkey hardware to lift-and-shift virtualization, then to containers and platforms. Through this, the drive for best-of-breed has continued to be a key driver  – quite rightly – but it has also reinforced patterns that don’t scale well: fragmented orchestration, single-environment approaches, and media handoffs that still mirror traditional system-to-system transport, even while applications run on shared compute.

The result is unfortunately predictable: more overhead, wasted resources, and otherwise avoidable latency.

The point of common principles isn’t to reduce choice, or differentiation. It’s the opposite: standardizing the foundational architecture, so that best-of-breed is easier to deploy, operate, and evolve, without turning every deployment into a bespoke integration project.

The next phase isn’t just software-defined execution. It’s about collaboration, at scale, and that requires alignment across the industry on architecture and operational principles.

JT-DMF: a blueprint for scalable, repeatable architectures

JT-DMF is best thought of as a blueprint for building and operating scalable software-defined production, especially in hybrid environments. It’s intended to focus on the architectural and operational foundations that let dynamic facilities grow without becoming fragile or bespoke:

  • Workload orchestration and responsibilities
  • Identity and access management
  • Lifecycle management
  • Deployment models – that work across on-prem, cloud, and hybrid

It’s early days with formal kick-offs only having taken place in late 2025, but the intent isn’t to constrain innovation. It’s to make delivery repeatable, so that scaling doesn’t mean rebuilding the same plumbing in every deployment.

Just as importantly, JT-DMF provides a shared frame for vendors to collaborate alongside customers. A common structure for partnership that reduces ambiguity, accelerates deployment, and improves long-term operability.

MXL: the enabler for true application collaboration

Where JT-DMF defines the direction, MXL is a key enabler.

SDI, and much of IP media, was built on linear, predictable pacing and transport across historically closed or tightly controlled environments. That model works well for connecting systems, but it assumes media is something you move from one box to another.

COTS compute doesn’t behave like that. CPUs and GPUs work in bursts, queues, and pipelines, optimizing throughput via asynchronous effort. To scale SDP, the media exchange has to work with the compute, not against it: asynchronous, parallel, and copy-avoiding. Every copy, handoff, and duplication burns cycles, increases memory pressure, adds latency, and reduces headroom.

That’s where MXL represents a step-change. Rather than transport between systems, it enables applications to share media flows within the same (or combined) compute fabric. Multiple applications, across one or many vendors, can operate on the same content concurrently with extremely low latency, while keeping video, audio, ANC, and timing aligned without repeatedly duplicating and transporting media – keeping the data local and accessible, avoiding costly handoffs.

AMPP: proven today, evolving in line with industry direction

None of this matters of course unless it holds up in a real production. Live events, real operations, real expectations. AMPP is evidence of this model, demonstrating that software-defined production can deliver at scale with centralized control, shared operational approaches, and efficient use of compute – a true platform.

Within AMPP, we’ve now released MXL support on managed nodes, and we’re working with partners to bring native MXL applications to release by NAB.

As JT-DMF and MXL mature, AMPP will evolve in step, incorporating architectural principles and interoperability approaches as they become established.

Hybrid is real, and it includes hardware

This future isn’t software-only though. Hybrid is the path that most organizations are taking, and hybrid isn’t just a term for cloud + COTS. It also includes purpose-built hardware where it delivers performance, or economic benefits, especially at the edge.

Hardware and software need to operate as one coherent, orchestrated environment, with consistent control, predictable performance, and efficient movement (or non-movement) of media across the edge from hardware to cots. The opportunity is to bring high-density baseband, IP, and compute together under the same operational environment, rather than treating them as separate worlds stitched together with gateways.

A practical example is the release of Grass Valley’s ACE-3901 – a modular card designed to bring high-density baseband I/O into MXL software defined workflows. The point isn’t hardware versus software, it’s about reducing friction at the edge so that systems can scale without creating new bottlenecks.

What this means and what comes next

Software-defined production (SDP) is already here and JT-DMF and MXL are direct efforts to industrialize it – through shared approaches and repeatable interoperability.

The outcomes are practical:

  • Faster and cleaner integration
  • Lower operational overhead
  • Better utilization of compute resources
  • More freedom to adopt new capabilities without reworking the foundation every time
  • A healthier multi-vendor ecosystem where differentiation happens in applications, not the plumbing

AMPP has direct alignment to these concepts and the DMF reference architecture. It shows that software-defined production can run at scale with shared operations, centralized control, and efficient use of compute – all while continuing to evolve in step with JT-DMF and MXL as the industry alignment continues.

What comes next is where the opportunity compounds: repeatable deployment models and interoperable interfaces – leading to increased partnership – so that SDP across vendors and ecosystems become easier to manage, operate, and faster to iterate.