November 5, 2026
Last week I was thinking about how many companies proudly wave around their cloud sustainability dashboards like they have cracked the code. Green graphs. Nice percentages. Executive smiles. Job done.
Not quite.
The awkward truth is this: a lot of cloud carbon accounting is built on abstraction stacked on abstraction, with a sprinkle of accounting wizardry on top. It can look precise while hiding the actual maths underneath. And if you are hiring teams to drive operational sustainability, that matters more than most leaders realise.
This is not just a tech issue. It is a leadership issue. Because when the numbers are fuzzy, decisions get fuzzy too.
Computer Weekly recently explored this in 'The invisible maths behind cloud carbon accounting', pointing out how hypervisor abstraction and market-based accounting can conceal real emissions, and why raw telemetry, real-time grid data and hardware lifecycle tracking matter if you want something closer to the truth. Source: Computer Weekly, https://www.computerweekly.com/news/366653518/The-invisible-maths-behind-cloud-carbon-accounting
Here is the first problem. Most business leaders see a cloud emissions figure and assume it is like a bank balance. Concrete. Traceable. Hard edged.
It is often nothing of the sort.
Cloud environments are abstract by design. That is the whole point. You buy compute without hugging the server. You consume storage without meeting the hardware. You spin workloads up and down without seeing what is going on underneath.
Brilliant for scale. Less brilliant for carbon clarity.
Hypervisors, shared infrastructure, multi-tenant environments and proprietary allocation models mean the number you get is often an estimate of your share of a complex system. Useful? Potentially. Precise? Steady on.
That would be manageable if organisations treated these figures as directional. But many treat them as if Moses brought them down from the mountain on stone tablets.
He did not. A provider dashboard is a model, not gospel.
This is where things get especially slippery.
Market-based accounting allows companies to report emissions based on contractual instruments such as renewable energy certificates or power purchase agreements, rather than the actual local grid electricity being used at the time a workload runs.
Translation: your workload may be running in a region powered by a fairly grubby grid at 3pm, but the reporting may still look cleaner because of how energy purchasing is accounted for.
Now, I am not saying these mechanisms are useless. They can support renewable investment. Fine. But they can also create a very polished story that hides operational reality.
And if you are trying to reduce actual emissions, not just produce a sustainability report that gives everyone a little serotonin hit, then operational reality is the game.
That means asking a simple, annoying and necessary question:
Are we reducing reported emissions, or real-world emissions?
Those are not always the same thing. Pretending otherwise is how clever companies end up doing sustainability karaoke.
If you want meaningful cloud carbon accounting, you need better visibility than most organisations currently have.
The Computer Weekly piece makes the point clearly: operational sustainability requires raw telemetry, real-time grid data and hardware lifecycle tracking. That is the grown-up stuff.
Why? Because emissions are shaped by more than usage alone.
You need to understand:
This is where a lot of teams hit the wall. They have cloud ops. They have FinOps. They may even have a sustainability lead. But nobody owns the whole picture.
So the business ends up with three partial truths:
And the emissions truth sits somewhere in the gaps, quietly mocking everyone.
This is the bit I care about most as a recruiter.
Many firms say they want greener infrastructure, better reporting and more responsible operations. Great. Then they hire like it is 2019.
They look for a conventional cloud engineer.
Or a generic sustainability manager.
Or a data person with no operational influence.
None of those hires are necessarily wrong. But on their own, they often miss the real challenge, which sits at the intersection of infrastructure, data, commercial trade-offs and sustainability.
The companies getting serious are not hiring job titles. They are hiring capability clusters.
They want people who can:
That is not your standard unicorn job spec. Nor should it be. Usually, it means being smarter about team design rather than dumping everything into one heroic role and hoping for miracles.
If you are building in this space, ask yourself whether your team can actually answer tough questions like:
If nobody can answer those, you do not have a measurement system. You have a confidence costume.
Let us make this practical.
If I were advising a leadership team that genuinely wants to improve cloud carbon accounting, I would push for four moves.
Do not settle for top-line provider summaries alone. Push for workload-level telemetry where possible. Understand what is estimated, what is measured and what is simply inferred through opaque models.
Carbon is not just about how much compute you use. It is about where and when that compute happens. Real-time or near-real-time grid intensity data gives you a far better operational signal than annualised averages alone.
Operational emissions matter. So does embodied carbon in the kit itself. If your sustainability model ignores hardware manufacturing, utilisation and retirement, you are only telling half the story.
This cannot live in one silo. You need infrastructure, data, finance and sustainability people in the same conversation. Preferably without everyone using different definitions and nodding politely while secretly meaning different things.
Here is a quick filter you can use internally. I call it Measure, Map, Move, Monitor.
Measure
What do we actually know versus what is modelled?
Map
Which workloads, regions and suppliers drive the biggest emissions impact?
Move
What changes can we make to architecture, timing or provider choices?
Monitor
How do we continuously track improvement using better data, not prettier slides?
Simple beats flashy. Especially when flashy is hiding the problem.
This is not only about carbon.
Teams that get serious about granular cloud visibility tend to get better at a few other useful things too:
Funny how truth has side benefits.
That is why I think the smartest leaders should stop treating cloud carbon accounting as a niche ESG issue and start seeing it as an operational intelligence issue. Because it is.
And if you are hiring into cloud, data or sustainability right now, this is where the market is quietly shifting. The valuable people are not just technical specialists or policy translators. They are the ones who can navigate the messy middle, where systems, numbers and commercial reality collide.
It is a pattern we see with greentech employers building teams that can interrogate the numbers, not just present them.
Cloud carbon accounting today is a bit like calorie counts on restaurant menus. Useful as a rough guide. Questionable if you inspect the kitchen too closely.
The leaders who stand out will not be the ones with the prettiest dashboard. They will be the ones willing to interrogate the assumptions, invest in better visibility and hire people who can connect infrastructure to impact.
Because if your carbon story depends on abstraction, accounting tricks and blind faith in vendor numbers, you are not managing emissions. You are managing appearances.
And appearances, as we all know, do not reduce a single tonne.