Q&A: Apptio's Greg Holmes on Energy in Financial Services

Electricity has become one of the biggest swing factors in the cost of running AI in financial services.

As institutions scale machine learning, real-time analytics and always-on digital infrastructure, the price and availability of power are starting to dictate what firms can afford to build and where they can afford to build it.

Nowhere is that more visible than in the UK, where businesses face some of the highest electricity costs in the developed world, adding a significant premium to any organisation trying to scale AI workloads.

Rising prices and grid volatility are pushing financial institutions to rethink cloud strategy altogether, from shifting workloads to cheaper regions to reconsidering private data centres for greater control over energy sourcing.

Greg Holmes, EMEA Field CTO at Apptio, an IBM company, works with financial institutions to connect these technology decisions with financial outcomes.

In this interview, he explores why energy now sits at the heart of digital strategy and what it means for organisations balancing innovation with cost, resilience and control.

How are rising energy costs reshaping the economics of AI in financial services?

Training and running models is already one of the fastest-growing areas of cloud spend in financial services. 

When electricity prices spike – as they have in the UK and across the globe recently – the same workload suddenly costs significantly more. This is pushing institutions to look harder at model efficiency and GPU utilisation. 

It’s also driving demand for more granular visibility, breaking down consumption hour by hour, spotting cost hotspots and predicting when certain AI tasks may become uneconomical. 

Energy volatility is becoming a real constraint on how quickly firms can scale AI.

It means for many queries, when we use an AI service we can optimise by choosing a model that is cheaper to run, without impacting the output we get.

Should energy efficiency now be a key factor when choosing cloud and AI providers?

In short, yes, and it’s quickly becoming a core part of procurement. 

Providers vary widely in how they source and manage energy and those differences feed directly into cost predictability, sustainability reporting and resilience. 

Some operate in regions with stable, low-carbon power, others rely on grids facing volatility. 

Energy-efficient infrastructure, transparent carbon accounting and the ability to schedule workloads intelligently are now meaningful differentiators. 

The practical point is this: choosing a provider without understanding their energy profile is an increasingly risky bet. 

Efficiency and transparency now belong alongside performance, security and compliance when evaluating cloud and AI partners.

Are energy prices forcing financial institutions to rethink their cloud strategies?

They are and in several ways. 

Some organisations are shifting workloads to regions with cheaper or more stable energy. Others are spreading workloads across multiple cloud providers to reduce exposure to volatility in any single market. 

Hybrid approaches are also gaining traction, with some firms revisiting private data centres to regain control over energy sourcing. 

FinOps teams are starting to model cloud decisions based on energy curves rather than just compute pricing.  

Rising costs are also accelerating interest in workload automation – letting non-urgent AI tasks run during off-peak hours when power is cheaper or moving workloads to regions with cheaper or more sustainable energy.

How might these costs and fluctuations affect the UK’s ambitions of being an AI superpower?

Realising the UK’s AI ambitions will depend in large part on affordable, reliable power and current energy prices make that harder.

With British businesses facing some of the highest electricity costs in the developed world, scaling AI carries a significant cost premium. That has implications for competitiveness, investment decisions and the UK’s ability to attract global AI workloads.

If volatility persists, some organisations may choose to run compute in cheaper markets, which would slow domestic adoption.

Currently, the UK has a large number of data centres, particularly with AI and high end cloud capabilities that are highly sought after. 

However, the cost of these is relatively high and affected by the ability to get additional electricity.

But this isn’t a lost cause. The UK remains well-placed to lead if it prioritises energy-efficient infrastructure, modernises the grid and supports greater transparency around consumption. 

What do you anticipate companies doing to improve energy security, sovereignty and their bottom line?

We’re seeing several shifts already. Organisations are investing in detailed energy visibility, understanding which workloads consume power, when and at what cost.

Many are adopting workload scheduling to move flexible compute to cheaper periods or more stable regions. 

Vendor vetting is becoming more rigorous, with firms assessing energy sourcing, carbon intensity and resilience to grid shocks. 

On top of this, some companies are building up their own private data centres to guarantee their own capacity and manage data sovereignty and independence. 

And FinOps is emerging as the framework that ties all of this together, helping leaders anticipate price spikes, avoid operational disruption and ensure AI growth doesn’t outpace the energy available to power it.