Baillie Gifford warns data center investors of a hybrid power shift

The prevailing wisdom suggests a future where data centers, powered by renewable energy, seamlessly support the digital economy. Yet, a counter-narrative is taking hold among power investors: the escalating electricity consumption of AI is disrupting this vision, demanding a pivot from exclusive reliance on renewables to diversified energy portfolios that ensure unwavering, dispatchable power. This shift directly influences where capital flows in critical infrastructure.

What Baillie Gifford actually says about demand and the transition The note frames the energy transition as "increasingly complex, moving away from a binary choice between fossil fuels and renewables toward a system that requires both," and highlights rising electricity demand driven by AI infrastructure and industrial growth as a key pressure on supply. That language comes from Baillie Gifford's Official Research and positions demand growth as the central variable that will determine how grids and asset owners choose generation sources.

No one in the reported packet is on the record.

The concrete mechanism: why AI compute stresses the baseload Baillie Gifford's central technical point is straightforward: AI workloads create concentrated, high-density electricity demand footprints in hyperscale data center clusters that favor reliable, low-variance power delivery. Intermittent renewables by themselves increase variability on the supply side; for mission-critical compute, variability is a risk vector that shows up as higher operational complexity and potentially higher unplanned costs.

Framed as an investor decision, that risk increases the value of contracted, firm generation or hybrid procurement arrangements that can guarantee capacity when needed.

Where the common narrative fails: the limits of a renewables-only procurement thesis The consensus narrative among many sustainability advocates and some corporate buyers is that commitments to 100% renewables — via PPAs, renewable credits, or onsite generation — are the path for data centers. Baillie Gifford rejects that as a universal solution, arguing instead for pragmatic mixes.

The mechanism is economic: reliable baseload reduces the total cost of ownership for latency- and reliability-sensitive workloads, and as compute intensity grows, the premium placed on that reliability can outweigh the price advantage of purely renewable profiles. This is a margin-structure shift argument: operators and their investors will accept higher up-front or contracted costs for predictable power rather than pursue renewables-only deals that carry intermittency risk.

What Baillie Gifford leaves out and why it matters for procurement The note is an investment house perspective and omits operational procurement detail: it does not model how PPAs, capacity markets, grid-scale batteries, demand-response contracts, or merchant firm gas/nuclear contracts would be structured across regions to meet this demand profile. That omission matters because the pathway from a thesis about 'hybrid mixes' to an executable procurement strategy is where margins and lock-in are decided.

Without that level of detail, the research is a directional signal for investors, not a blueprint for data center operators or utilities.

The skeptic case: renewables plus storage could close the gap A natural counter-read — one Baillie Gifford does not fully address — is that faster deployment of grid-scale storage, smarter demand-side management, and optimized hourly PPAs could enable renewable-dominant footprints even for high-intensity AI clusters. Critics will point to falling battery costs, improved grid integration, and corporate offsite PPA portfolios as mechanisms to flatten variability.

That objection highlights the missing financial engineering discussion in the Baillie Gifford note: it claims hybrid mixes will be favored but does not quantify the cost or timeline at which storage and market design would tilt procurement back toward renewables-only.

Who benefits, who loses, and the unnoticed middle If Baillie Gifford is right, investors in firm-generation assets — existing gas peakers refitted with emissions controls, nuclear capacity holders, and grid-scale hydro where available — become more valuable as sources of contracted reliability. Data center developers with flexible site selection and strong utility partnerships gain leverage in pricing and resilience.

The exposed middle is the merchant renewable developer who lacks storage or firming contracts; their projects may require richer off-take terms to remain attractive. This is a procurement and margin story: higher-priced, firmed capacity becomes an acceptable line-item for enterprises that run revenue-critical AI workloads.

Short, falsifiable signals to watch over the next six to twelve months Watch for three observable developments: major hyperscalers publicly committing new data center builds that are explicitly powered exclusively by renewable sources; regulatory or grid operator reports showing a downward trend in natural gas or nuclear generation for data center clusters in key AI hubs; and commercial data center developers announcing new facilities with 100% renewables without relying on offsets. Any of those would materially weaken the Baillie Gifford thesis that hybrid, firmed capacity will command a margin premium for AI-era compute.

Baillie Gifford's note is a sober reminder to investors that demand growth matters as much as supply-side decarbonization. It is a marketing_blog–tier analysis that sketches the direction of travel but stops short of operational procurement mechanics — the real place where data center margins and investor returns are re-priced. Investors should treat the conclusion as a hypothesis to be stress-tested with concrete PPA, storage, and capacity-market scenarios rather than as an immediate fact.