Data Centers and Sustainability: The Uncomfortable Question

Almost every sustainable portfolio ends up holding something uncomfortable. For us, the recurring example is the data center.

The digital economy runs on buildings that draw enormous amounts of electricity, consume large volumes of water for cooling, and are being built at a pace that strains local grids. At the same time, the software, logistics, and services that run inside those buildings are among the more resource-efficient ways of delivering what people actually want — and the companies operating them are often the ones funding the build-out of renewable generation.

Both statements are true. That is the uncomfortable part, and it is worth working through rather than around.

Why the first glance is misleading

Judged on operational footprint alone, a data center looks like a poor sustainability holding. It is an energy-intensive physical asset, and its consumption is growing.

Judged on the service it delivers, the picture changes. Streaming a film uses far less energy than manufacturing, distributing, and playing a physical disc. Routing freight with better software reduces fuel use. Moving a workload to a hyperscale facility is usually more efficient than running the same workload in a small on-premises server room, because the large operators invest in cooling design, load balancing, and power purchasing at a scale nobody else can match.

Substitution matters. A company can have a large footprint and still be the lower-impact choice compared with what it replaces.

The three questions we ask

1. Where does the power come from, and who is paying for the transition? Operators that sign long-term contracts for renewable generation are effectively financing new capacity. Those that simply buy credits without changing the underlying supply are doing something different, and worth distinguishing.
2. How efficiently is the capacity used? Utilization is the metric that gets ignored. Idle capacity consumes power while producing nothing — the least defensible kind of environmental cost.
3. Who bears the local cost? Water stress and grid strain fall on specific communities. Whether the operator is a constructive participant there — disclosed, engaged, paying its share of infrastructure — is a governance question as much as an environmental one.

What this means for a portfolio

It means digital infrastructure is neither automatically excluded nor automatically included. It is held, if at all, with the trade-off stated plainly and the exposure sized accordingly.

It also means the exclusion decision is not always the sustainable one. Removing efficient digital infrastructure from the portfolio does not remove demand for it. It moves the ownership to someone who is not asking these questions — with no change to the environmental outcome.

The general principle

Very few businesses are unambiguously good or bad on sustainability. Most are somewhere in the middle, improving or deteriorating at some rate, and the useful work is measuring which.

A portfolio built on the assumption that clean categories exist is easy to market and hard to defend. A portfolio built on stated trade-offs, reviewed honestly, is harder to explain and much easier to live with. We think that is the better trade.