By Jon Haley, Chief Revenue Officer, Scalstrm
In streaming, carbon emissions don’t come from video. They come from work that never needed to be done in the first place.
For years, the industry has accepted pre-transcoding as a given. Encode everything upfront. Store every variant. Keep servers running permanently, just in case demand arrives. That approach was convenient, but it was never efficient, and today, it is one of the biggest hidden drivers of power consumption in streaming operations.
At Scalstrm, we took a different view: if content is not being watched, processed, or delivered, why should infrastructure be burning electricity for it?
This is where Just-In-Time transcoding fundamentally changes the sustainability equation.
Instead of pre-processing content and keeping vast transcoding farms alive around the clock, Scalstrm processes streams only when they are actually requested. No speculative encoding. No idle compute. No silent power drain waiting for traffic that may never come.
The environmental impact of this shift is immediate and measurable, not because we set out to “be green,” but because we eliminated unnecessary work.
When pre-transcoding disappears, server footprints collapse. When server footprints collapse, power consumption drops. When power consumption drops, cooling requirements fall with it. Carbon reduction is not an extra step in this chain, it is the natural outcome.
This is why with Scalstrm’s Just-In-Time transcoding you see power consumption reduced by around eighty percent. It is not the result of marginal tuning or hardware swaps. It is the result of removing entire layers of always-on processing from the architecture.
The same logic applies to server count. Traditional workflows require scale through multiplication: more channels mean more servers. Just-In-Time workflows scale through efficiency: more channels are handled by the same infrastructure because work only happens when demand exists. That is how server footprints shrink by three quarters or more without compromising reliability.
What often gets missed in sustainability discussions is that software architecture decisions have a far greater environmental impact than energy sourcing alone. Running inefficient workflows on renewable power is still inefficient. Reducing the work itself is where the real gains are.
Streaming does not need to be power-hungry by default. It became that way because the industry optimised for predictability rather than efficiency. In a world of dynamic demand, personalised streams, FAST channels, and bursty traffic patterns, that assumption no longer holds.
Just-In-Time transcoding aligns infrastructure behaviour with actual viewer behaviour.
When no one is watching, systems stay quiet. When demand arrives, capacity scales instantly. Power is consumed only when value is being delivered.
That is what sustainable streaming looks like in practice. Simply doing less unnecessary work, and letting power, cost, and carbon reductions follow.
