First published on SVG Europe, 9 July 2026.
For years, almost everything the streaming industry said about its environmental impact was based on models and estimates. Top-down calculations; reasonable assumptions; papers citing papers — with remarkably few actual measurements underneath any of them.
I know because I spent two years reading those papers. And then I watched a room full of engineers go quiet when, for the first time during one of our hackathons set up by [Greening of Streaming board member] Simon Jones, live power data appeared on screen. Not modelled or estimated — actual watts drawn by a video server and half a dozen TVs doing actual work.
That moment captures the transformation that Greening of Streaming (GoS) has undergone — and that I’ve spent the last two years trying to operationalise. As SVG Europe reported, GoS has formalised its governance, established a board, and launched a structured research model. Those are the facts. This is the story behind them.
Someone else’s baby
GoS was founded by Dom Robinson. Pulling together engineers, broadcasters, CDN operators, and tech vendors around streaming’s energy footprint was something no one else was doing when he started during the COVID lockdowns.
But Dom hit a wall that anyone who’s run an industry initiative will recognise. As the founder and public face of an organisation that depends on members’ goodwill, you can’t always say what needs to be said. You can’t call out greenwashing without risking the very companies you need in the room. After four years, that tension was wearing him down.
We spent months discussing whether I’d take over, and more months discussing how. Barbara Lange mentored us through some of those conversations. Dom had built something valuable, and handing it to someone else meant trusting it would survive the transition.
The scariest spreadsheet in France
Restructuring GoS from a UK limited company into a French non-profit association under the Loi 1901 terrified me. Becoming an association meant GoS would be governed by a board, not by a single person. Decisions would be slower. The nimbleness of one person making the call would give way to collective deliberation.
I did it anyway, because an organisation whose credibility rests on independence needed governance to match. If GoS was going to produce data that CTOs and policymakers could rely on, it couldn’t be run as a commercial entity on one person’s judgment.
From working groups to labs
We also reorganised how GoS works — moving from broad working groups into focused labs, each designed to produce published outputs.
WattLab, which Simon and I run, is our measurement engine. It uses REM — our Remote Energy Measurement platform — with smart plugs on real devices capturing wall-power data at one-second intervals. No models, no proxies: a device, a codec, a piece of content, and a power reading. In our first end-to-end hackathon, we discovered that some TVs consume significantly more power when fed HD than 4K — the opposite of what most people assume. It challenges the widespread industry belief that lower resolution always means lower energy, and it’s the kind of finding you only get by measuring. We’re also developing OWL (Online WattLab), a server-side platform running standardised workloads continuously across video transcoding and AI inference, which we expect to become the backbone of our research programme.
Language Lab produces our research papers. The most recent, AI in Streaming Media Sustainability: Distinguishing Impact from Innovation, examines where machine learning actually appears in streaming workflows and what it means for energy. The core finding is that AI is neither inherently a sustainability solution nor a threat — it depends on where it sits in the workflow and how much energy it adds versus avoids.
Policy & Guidelines Lab translates measurement work into evidence-based input for policymakers and standards bodies.
Protocol Lab handles our technical engagement with the standards community. In March, we co-organised a workshop with 3GPP and 5G-MAG on measuring media energy consumption — a sign that our approach is gaining real traction.
Marketing Lab programmes professional development sessions for members and ensures our message reaches the wider industry.
Audio Lab and Academia Lab are still in development. All labs are designed to produce outputs, not just discussion.
We’re not there yet — and that’s the point
GoS now has a proper board, a lab structure, over 20 member organisations, and the beginnings of a credible measurement capability. But we’re at an inflection point, not a finish line.
None of this would exist without our members. They bring the devices, the expertise, and the engineering hours that make measurement possible. Membership means access to pre-publication data, influence over how methodologies are designed, a voice in the guidelines we feed into policy discussions, and participation in hackathons that generate primary data no one else produces. Members don’t just fund the work, they shape it.
We need to grow to sustain the breadth of our activities without spreading the jam too thin. The infrastructure is built, the methodology works, the question is who wants to help shape what comes next.
Dom saw this problem five years ago. He built the community that made it possible to even ask the question. What we’ve added is the governance, the structure, and the measurement infrastructure to start answering it. This journey wouldn’t have been possible without the rest of the board — Mike Mattera, Stan Moote, and, more recently, Veronika Marfina, Tania Pouli, and Marisol Palmero.
We’re not eco-warriors, we’re just people who dislike waste. And after five years, we can finally measure some of it, which means the industry can finally start making decisions based on evidence rather than assumptions.