Executive Summary
- For AI-era data centres, equating “energy storage” with lithium is a costly category error.
- The firmest storage we have is generation itself — and some of the best of it is already turning waste into round-the-clock power.
- Here, Andy Ling, Co-Founder & CEO, Beyond Grid Energy explains why we’re solving the wrong problem.
When the data centre industry talks about energy storage, it almost always means one thing: lithium-ion batteries. Racks of them, sized to ride through a grid dip or shave a peak. Batteries are extraordinary pieces of engineering and they have a real place.
But if we let “storage” and “batteries” become synonyms, we miss the more important question the AI build-out is forcing on us: how do you store energy at the scale, duration and reliability that continuous compute actually needs?
Batteries answer a narrow version of that question. They exist, fundamentally, to paper over intermittency — to hold solar and wind for the minutes and hours when the sun sets or the wind drops. That is useful. But it is expensive, it degrades, and it is short-duration.
A battery buys you minutes; it does not buy you a firm, year-round supply. Behind every big battery is still an assumption that something, somewhere, is generating the power in the first place — and for a 24/7 AI training cluster, “something intermittent plus a battery” is a workaround, not a foundation.
Let’s talk biology
There is a form of energy storage the industry consistently overlooks, and it isn’t electrical. It’s biological.
Anaerobic digestion — AD — is, at its heart, an energy-storage technology. Organic material such as farm waste, food waste and slurry is broken down by microbes to produce biogas, which is stored and converted on demand into electricity.
Think about what that actually is: energy captured from the sun by crops and waste streams, banked as chemical energy in a feedstock, and released as power precisely when you want it. It’s a battery whose cells are measured in tonnes of feedstock rather than kilowatt-hours of lithium — and it discharges not for four hours, but continuously, for as long as you keep feeding it.
That distinction matters enormously for data centres. AD delivers firm, dispatchable baseload. It doesn’t care whether the wind is blowing. It runs at high availability around the clock, which is exactly the load profile an AI data hall demands.
Where solar-plus-battery is a generation source that needs storage bolted on to become useful, AD is generation and storage in a single asset. The storage sits upstream, in the feedstock and the gas — not downstream in a hall full of cells you have to buy, cool and replace every decade.
And it is genuinely low-carbon — often better than that. Because AD captures methane that would otherwise escape from decomposing waste, on a lifecycle basis it can approach carbon-neutral or below, while diverting waste from landfill.
For operators under mounting scrutiny on Scope 2 and embodied emissions, that’s a firm, traceable, behind-the-meter electron a grid tariff or an offsite certificate simply can’t match.
The broader subject
AD is not the whole answer on its own, and I’m not going to pretend it is. The shift I want the industry to make is broader: to widen the definition of “energy storage” from a single chemistry to the full palette of firm, dispatchable renewables — and to recognise that the firmer your generation, the less storage you have to buy to prop it up.
Hydropower is the example everyone forgets. Pumped hydro is the original grid-scale energy store; we’ve been banking energy as water behind dams for a century. Tidal is another — not yet mainstream, but uniquely valuable because it’s predictable to the minute, decades out. You can schedule compute around a tide table in a way you never can around the weather.
Geothermal offers true baseload where the geology allows, and sustainably-sourced biomass plays a storage-like role similar to AD. None of these is a silver bullet, but together they change the framing. Instead of asking “how big a battery do I need to make intermittent power reliable?”, you ask “how much of my supply can I make firm at source?” Every megawatt of firm renewable generation is a megawatt you don’t have to underwrite with storage.
Architecture of supply
This is where resilience and efficiency actually come from. The conventional data centre resilience stack is grid supply, backed by diesel, backed increasingly by batteries — three systems, two of them idle most of their lives, one of them burning fossil fuel at the exact moment it’s needed.
Firm renewable generation behind the meter reconfigures that. If your primary supply is already dispatchable and on-site, you’re not oversizing a battery array to compensate for a weather-dependent source, and you’re leaning far less on diesel. Resilience comes from the architecture of the supply, not from stacking ever more back-up on top of a fragile foundation.
Which brings me to the constraint that trumps all of this — and the reason the storage debate can feel academic to anyone actually trying to build. None of it matters if you can’t connect.
Join the queue
Grid connection queues in the UK now stretch a decade or more; some data centre projects are being handed connection dates in the 2030s. You can have the cleverest storage strategy in the world and still be years from a live site. The unlock is to stop treating the grid as the only route to power — to wire firm renewable generation directly into compute on a private wire, behind the meter, and skip the queue entirely.
Do that, and energy storage stops being a component you bolt on and becomes a property of how the whole site is designed.
I spent years building grid-edge infrastructure before this — including a 12km private wire connecting a 50MW battery to EV charging in Oxford — and the lesson that stuck is that the winning designs treat generation, storage and delivery as one system, not three procurement line items.
Solving the wrong problem
So by all means, keep the batteries. They earn their place. But if the industry’s answer to “how do we store energy for AI” begins and ends with a lithium rack, we’re solving the wrong problem.
The most powerful storage available to us is firm, dispatchable, renewable generation — and some of the best of it has been quietly turning waste into round-the-clock power on British farms for years, waiting for the data centre industry to notice.
About the author: Andy Ling is Co-Founder and CEO of Beyond Grid Energy, which builds private-wire renewable power infrastructure connecting firm generation directly to modular data centres.
Editor’s note: This article is taken from the Fall issue of DCi’s Frontiers magazine, and you can read the full, digital publication here.