NAB is done. The stands are packed away, the jetlag long forgotten, and the industry has had another week of conversations about where video infrastructure is heading. 

Not everything that happens in Vegas stays in Vegas. Here is what those conversations confirmed for me. 

COTS is the foundation modern video infrastructure is built on. At NAB, the customers we spoke with are not questioning that. What they are asking now is how much useful work they can get out of it. Most deployments are leaving a significant amount of that value on the table. 

The new constraint is compute 

COTS removed much of the rigidity of fixed function hardware. It made video infrastructure more portable, more flexible, and easier to evolve over time. 

But it also exposed a harder truth. Not all COTS deployments perform equally. 

In live video environments, inefficiency shows up quickly. More servers are required. Power consumption rises. Operational complexity grows. Costs rise with it. 

The real question is no longer whether a platform runs on standard hardware. It is how much useful work each server can deliver. 

Why MediaKind standardised on AMD 

Modern video workloads are highly parallel by nature. Multiple streams, formats, bitrates, and outputs must run continuously, reliably, and often under strict latency constraints. 

That makes processor choice a strategic decision, not a background component choice. 

Encoding is not just compute bound. It is data movement bound. Memory bandwidth and input/output throughput matter because video pipelines depend on moving large volumes of data continuously, without stalls or contention. 

This is one of the reasons AMD aligns so well with how these workloads behave in production. High core density, strong memory bandwidth, and high I/O throughput are not abstract specifications. They determine how many streams you can run per server, how stable performance is under load, and what your infrastructure actually costs to operate at scale. 

We went into the full technical detail in a piece published with AMD. The short version is that the consolidation scenarios we modelled show meaningful reductions in physical headend footprint. What once required racks of single-purpose appliances can run on a much smaller number of high-density servers without compromising performance or control. 

What this changes for MK.IO Beam 

In MK.IO Beam, that processor decision shapes the entire deployment model. Contribution, encoding, and processing run on a unified COTS foundation. There are no isolated hardware silos to manage, no separate refresh cycles, no single-purpose boxes sitting idle between events. The infrastructure does more, with less, and stays simpler to operate as it scales. 

Efficiency compounds 

In video infrastructure, efficiency is never a one-time gain. Fewer servers mean lower power and cooling demand, less rack space, fewer failure points, and less operational overhead. Over time, those gains compound into a meaningfully lower cost base and a simpler operating model. 

Just as importantly, engineering teams spend less time managing infrastructure and more time improving the service. 

Performance is the differentiator 

COTS is now table stakes. The next phase of infrastructure evolution is about extracting maximum value from that shared foundation. That comes down to compute efficiency, platform design, and how much performance can be delivered from every server deployed. 

For MediaKind, that is why processor choice matters. AMD gives MK.IO Beam the density, efficiency, and throughput needed to run modern video workflows at scale. The operators who figure that out earliest will spend less, scale faster, and have engineering teams focused on the service rather than the infrastructure underneath it. 

Read AMD’s blog
The technical case behind our processor choice.

Discover MK.IO Beam
Learn how MK.IO Beam supports compliant, high-quality live video workflows for global sports distribution.