Why data centres should accelerate clean power, not gas
247Solar
Written by 247Solar
The rush to build AI data centres has turned into a convenient rationale for a new wave of gas plants. Utilities and energy suppliers are pointing to “unstoppable” AI demand to justify trillions of dollars in gas, nuclear, and even coal investments, arguing that only these resources can keep the lights on in an AI era. But underneath the rhetoric, the numbers tell a different story: building fossil-fuel infrastructure to power data centers is shortsighted when renewables are already the cheaper, faster choice.
A risky bet
As Amory B. Lovins and Justin Locke write in Canary Media, energy suppliers don’t yet know how many of those facilities will eventually materialize, yet they are using AI’s ravenous appetite for electrons to justify vast new investments in long‑lived fossil assets. This is locking in decisions that will expose them to volatile power prices for decades, even though the demand forecasts behind them are highly uncertain. There are several risks to this approach:
- Overbuild risk: If even a portion of today’s AI demand projections fail to materialise, ratepayers and investors are left paying for plants that are neither fully used nor easily repurposed.
- Fuel‑price risk: Gas plants embed long‑term exposure to fuel price volatility, while AI customers are demanding stable, predictable cost structures.
- Carbon‑policy risk: As climate policy tightens, highly emissive assets built today are more likely to face higher compliance costs, curtailed run times, or accelerated write‑downs.
By contrast, modular clean resources — wind, solar, storage, and other firm clean options — can be staged over time and financed with far less exposure to future fuel prices or demand shortfalls.
Gas is often slower and more expensive in practice
The argument that “only gas can scale fast enough” ignores real‑world constraints that are already biting. Turbine supply chains, long development timelines, and mounting local opposition are slowing conventional projects. At the same time, renewables and storage have become the dominant source of new capacity because they can often be sited and built faster, especially at the scales that individual data centre campuses require.
Clean portfolios also avoid the compounding effect of fuel and carbon costs over a project’s life. Once built, wind, solar, and many forms of storage have effectively zero marginal fuel cost, whereas gas plants commit owners to decades of commodity risk. For hyperscalers and AI developers under pressure to deliver both growth and cost predictability, that distinction matters as much as emissions.
Smarter ways to meet AI‑era load growth
The good news is that utilities and large loads are not limited to “gas or nothing.” A growing set of policies and incentives are emerging to make big new loads take responsibility for their own power while protecting other rate-payers:
- Large‑load tariffs and cost‑shielding structures: Several states are experimenting with dedicated tariffs for very large customers that explicitly allocate grid upgrade and resource costs to those who are driving the demand, rather than socialising them across all ratepayers.
- Bring‑Your‑Own (BYO) and Clean Transition Tariffs: Under BYO structures, large loads pay directly to add new power resources, often clean resources, that serve their needs without burdening other customers. When designed as Clean Transition tariffs, eligibility can be restricted to carbon‑free options, further aligning with policy goals.
- On‑site and “on‑campus” clean generation: Data centres can build or contract for renewables and storage on or near their sites, easing transmission constraints and speeding deployment compared with distant, central‑station plants.
These approaches are already moving from theory to practice. Utilities in multiple states are exploring structures that require data centres to pay most or all of the cost of new capacity built on their behalf, while some large tech firms are co‑developing clean power projects and associated tariffs to cover their incremental load.
What this means for developers
For data centre operators and AI developers, the choice is increasingly clear. They can accept utility plans that saddle them with long‑term exposure to fuel and policy shocks, or they can embrace cleaner, more flexible alternatives that support their growth trajectories while mitigating public concerns. The second path is not only better for the climate; it is also the more rational business decision in a world where power, not compute, is becoming the real bottleneck for AI infrastructure.
Author:
Source: https://247solar.com/zero-carbon-why-data-centers-should-accelerate-clean-power-not-gas/#data