Sustainability

Browse all sustainability tools – pick the topic that matches your question.

🌐
Digital Impact

All digital-carbon planning tools in one place – streaming, cloud, AI queries, video calls. Pick the calculator that matches your question.

1 Tool

What Sustainability Looks Like When You Can't See It

A car commute and a plastic bag are sustainability stories told in the right places — on the road, in the bin, on the news. They have weight, they have packaging, they leave a trace someone can point at. A four-hour video call doesn't. Neither does a month of cloud backups, an evening of streaming, or a workday of AI prompts. The energy moves the same way and the resources get drawn down the same way, but the trail goes to a building you've never seen and the bill is split across millions of users. None of it appears on your power statement. None of it shows up in your trash. And yet, by the International Energy Agency's 2024 electricity tracking, data centers alone could approach the annual electricity use of a mid-sized country before the end of the decade.

Sustainability is the topic that covers both halves of that picture — what you can see and what you can't. The visible half has thirty years of public conversation behind it. The invisible half is most of what's growing.

The Two Sides of Environmental Impact

Most environmental thinking still runs on the visible side. The numbers people talk about in everyday life — miles driven, single-use cups, household recycling rates — describe the parts of life where the resource flow is physical and the impact is in your hand. They matter, and they are also far from the whole story. They leave out a category that has grown into one of the fastest-rising shares of global energy use in fifteen years.

The split is roughly this: the visible side is what you buy, drive, eat, and throw away. The invisible side is what runs on your behalf — the cloud services holding your photos, the streaming platforms decoding what you're watching, the AI systems answering what you typed, the always-on infrastructure that powers all of it. The first half is regulated, labeled, and personal. The second half is mostly outsourced, mostly unlabeled, and mostly counted by sector rather than by person. Both belong in the same conversation; only one usually shows up.

Why Digital Is the New Invisible Half

The growth curve explains the attention. Data-center electricity use has roughly doubled in the past decade and is projected to rise further as generative AI scales — the IEA's 2024 Electricity report puts the data-center share of global electricity at around 1.3% today, with credible 2030 paths between 2% and 3%. Streaming long ago tipped from a fraction of internet traffic to its dominant share; Sandvine's Global Internet Phenomena reports have consistently placed video at well over half of downstream bandwidth. AI training and inference now run continuously across most large cloud regions, with workloads that didn't exist as a category five years ago.

None of those are a user's direct emissions in the way a commute is. They are a user's share of an infrastructure that exists to serve the request. The Green Software Foundation's framing puts it plainly: the carbon intensity of software is a property of the system itself, the energy mix it runs on, and the demand it serves. All three matter, and the user has partial influence over each. Pretending otherwise — on either the optimistic or the pessimistic side — produces bad planning.

Why Sustainability Decisions Are Hard

Three things make sustainability questions harder than they look. First, lifecycle effects sit out of sight: the manufacture of a phone or a battery often outweighs the operational footprint of using it, and the supply chain reaches into materials extracted continents away. Second, resource consumption is indirect: a query, a stream, a backup all draw on servers, networks, and cooling systems that don't appear on any bill the user sees. Third, assumptions vary widely: two credible estimates of the same activity routinely differ by an order of magnitude depending on what's counted, what's allocated, and where the boundary sits. The GHG Protocol exists precisely because the boundary question is hard, not because it's been settled.

Technology also moves. A streaming hour in 2018 is not a streaming hour in 2026; the grid has changed underneath it, the codecs have changed, the data centers have changed. Numbers that were broadly right five years ago need re-checking, and any single figure stripped of its assumptions tends to mislead more than it informs. Honest sustainability writing names its boundary, its baseline year, and the things it deliberately left out.

Where the Numbers Get Specific

Awareness without a number stays abstract. The digital sustainability resources on this site sit one layer deeper than the framing on this page — they take the broad categories above (streaming, cloud, AI, infrastructure) and translate them into the kind of estimate someone can actually plan around. The point of the deeper layer isn't precision for its own sake; it's having a directional figure that can be compared, repeated, and improved over time.

If the question is the everyday side — which choices change the digital share of a household or a team's environmental load — the technology impact tools are where the framing on this page becomes a number. The conceptual layer sets the context; the next layer does the math. Both are useful for different decisions, and the order matters: a number without context is harder to act on than a context without a number.

Choices That Move the Needle

Five categories of decision shift the result more than the rest.

Default settings. Auto-play, auto-sync, auto-update, default 4K, background refresh. These carry most of the load on the digital side, and most people never open the menu. The difference between a moderate digital footprint and a heavy one is usually written into the defaults rather than into the headline choice of which app to use.

Hardware lifespan. Manufacturing a smartphone or a laptop typically dominates its lifetime footprint — the German Environment Agency and other national bodies have published consistently on this. Keeping a device a year or two longer is usually the single biggest environmental decision a user makes about it, ahead of every operational efficiency tweak combined.

Energy mix. A server in a region running on coal and a server in a region running on wind do not have the same impact for the same workload. For organizations, region selection is a real leverage point. For individuals, the analogue is when you charge — the grid's cleanest hours are usually mid-day with renewables on, not overnight on baseload coal.

Efficiency gains. Codec efficiency, hardware efficiency, model efficiency — each compounds. Each also tends to be offset by demand growth, which is why efficiency alone has rarely reduced total resource use over the long run. The trade-off is real and worth naming honestly rather than ignoring.

Scope choices. The sharpest environmental decisions are usually about what's in scope at all: which subscriptions to keep, which features to default off, which devices to retire and which to repair. At the individual scale, sustainability is mostly subtraction. At the organizational scale, it's mostly accounting boundaries.

Common Questions About Sustainability

What is sustainability?
Sustainability is the practice of meeting today's needs without using up the resources future generations will need. In environmental terms, that splits into resource use (water, materials, energy), emissions (greenhouse gases, pollutants), and ecosystem effects (biodiversity, land use). The United Nations' Brundtland definition is the most widely cited starting point, and most national environmental frameworks build on it.
Why are environmental impacts difficult to measure?
Because most impacts are spread across systems the user doesn't see directly. A device used at home draws on a power grid mixed from many sources, a network of servers the request passes through, and a supply chain that built the hardware in the first place. Each of those carries an environmental cost, and assigning a share of it to one user requires assumptions about allocation, boundaries, and time horizons that experts disagree on in good faith.
How does technology affect sustainability?
In two directions at once. Technology enables efficiency — remote work avoiding flights, video calls avoiding commutes, sensors cutting waste in supply chains. It also creates new demand: streaming, cloud services, AI inference, and the manufacturing footprint of the devices that access them. Whether the net is positive or negative depends on the specific use case and the time frame; both effects are real, and the conversation gets worse when only one side is named.
Why do sustainability estimates vary so much?
Mostly because the boundary varies. A "footprint" can count direct energy only, or it can include manufacturing, supply chain, end-of-life, and a share of shared infrastructure. The GHG Protocol's distinction between Scope 1, 2, and 3 emissions is the standardized way to talk about this, and the same activity can show very different numbers under each scope. A useful estimate names its boundary up front.
What role does efficiency play in environmental planning?
Efficiency lowers the resource cost of a given activity, but it doesn't automatically lower total resource use — Jevons' paradox has played out repeatedly with energy, computing, and transport. Planning has to combine efficiency improvements with explicit demand decisions; otherwise gains in the first column get eaten by growth in the second. Efficiency is necessary; on its own it's rarely sufficient.
Why are some environmental impacts so hard to see directly?
Because the trail leads to infrastructure that no individual owns, controls, or is billed for separately. A streaming hour runs servers somewhere, cooled by something, fed by a grid mixed from many sources, and arriving over networks built over decades. None of that appears in the home bill or the device interface. Sector-level disclosures from cloud providers, regulators, and bodies like the IEA and the ITU are how the picture gets assembled at all.

Adjacent Sections