AI’s booming power appetite and the rising focus on consumer-led flexibility were among the themes that dominated the sessions, alongside growing concerns around data centre demand, trust in digital technologies, and the evolving role of the energy user. Here are three key themes observed by Marketing Manager, Ciara Inglis, at this year’s conference:

1. The growing energy impact of AI and data centres 

Data centres were a recurring topic throughout the day, with serious questions being raised about how the sector will scale without overwhelming the grid. Adrian Del Maestro, VP of Global Energy Advisory at AECOM, shared a detailed view of how surging demand for AI compute is pushing energy systems to their limits, particularly in the US, where some utilities are already experiencing power quality issues near major data centre hubs. He referenced Bloomberg’s research into the impact of rising data centre density on electricity networks, which found that areas with the most intense data centre activity are also seeing a measurable decline in power quality. The report highlighted how distorted waveforms in electricity supply, sometimes referred to as harmonics, are damaging household appliances, straining aging infrastructure, and exposing vulnerabilities in both urban and rural grids. 

Del Maestro shared that cooling and compute now account for more than 80% of total data centre energy use. With demand forecast to double in the coming years, operators are starting to look beyond traditional renewables. AWS, Microsoft, and Google are sourcing energy from nuclear, geothermal, and even fusion technologies to power new sites. The long-term goal remains low-carbon, but in the short term, natural gas is making a comeback, often paired with carbon capture systems and advanced cooling solutions like liquid immersion. 

Ireland was presented as a cautionary example. Its data centre electricity consumption is expected to double by 2026, with 14 facilities under construction and 40 more approved. This rapid expansion is putting pressure on grid infrastructure and community trust. In response, interest is growing in secondary European markets such as the Nordics, where abundant renewables and cooler climates make data centre deployment more viable. 

That said, it’s worth keeping the broader picture in mind. It should be noted that data centres are likely to contribute less than 10% of overall global power demand growth. Electrification of transport and space heating or cooling will have a far greater impact. Still, for energy professionals, understanding how and where data centre demand materialises – particularly in relation to grid access, emissions, water use, and permitting – is becoming an essential part of the landscape. 

2. Demand-side flexibility’s evolution into a commercial opportunity 

Dr. Alastair Martin, Founder and Chief Strategy Officer of Flexitricity (check out our case study with Flexitricity on MLOps for energy forecasting here), delivered a session on unlocking demand-side flexibility. At one point, Martin shared an image often used to misrepresent the concept – a family huddled around a table, wrapped in blankets and lit by candlelight – underscoring how public perception frequently distorts what flexibility actually involves. 

Martin reframed flexibility as something much more accessible and relevant to how businesses already operate. The term “consumer-led flexibility” is now emerging to better reflect this shift. 

The National Energy System Operator (NESO) has called for a fivefold increase in flexibility to help meet net zero targets. Martin highlighted how industrial and commercial users can contribute by making small operational adjustments that don’t compromise business continuity. Examples included flexing EV fleets, adjusting heating or cooling, or optimising behind-the-meter battery use. 

A key takeaway was the rise of ad hoc flexibility – assets not originally intended for flexibility, such as batteries installed for backup or load-shifting, which can now be made available to the grid when not in use. The session positioned flexibility as something that’s already within reach for many organisations, provided the data and systems are in place to support it. 

3. Why trust and validation matter in AI for energy 

It was surprising to see that, despite the central role software plays in reaching net zero, there was only one main stage session focused on digitalisation, AI, and data. The discussion brought attention to how the energy sector thinks about trust in digital systems, particularly AI. 

The panel made it clear that trust isn’t binary. AI doesn’t need to be perfect – it needs to be trustworthy enough for the use case at hand. One audience member asked what “good enough” looks like in AI. The consensus was to treat it the same way trust is built with people: start small, validate early outputs, and increase reliance as confidence grows. A human should remain in the loop, and some degree of healthy scepticism should always be maintained. Blind trust, particularly in high-stakes systems, is where risk can creep in. 

Capgemini’s Dr. Kerry-Ann Adamson noted that trust isn’t inherently digital – it’s shaped by how well people understand the systems they’re using. Graeme Booth from NZTC echoed this with a reminder that AI is only as effective as the data it’s fed. Poor inputs lead to poor outcomes, regardless of how advanced the model appears. 

The conversation also explored how AI bias can be understood similarly to how readers interpret news media: with an awareness of audience, context, and intent. It’s a useful way to highlight the importance of evaluating outputs critically, rather than assuming objectivity. This links back to the need to assess AI systems through the lens of each organisation’s specific risks and opportunities, rather than applying a one-size-fits-all approach. 

Finally, the session returned to the topic of energy use. With AI workloads pushing up data centre demand – on track to rival the electricity consumption of Japan by 2030 – energy leaders are now faced with a dual challenge: managing the power AI consumes, and ensuring AI is helping rather than hindering the path to grid resilience. In this context, building trust in data, systems, and decisions becomes not just a technical challenge, but a strategic one. 

The bigger picture 

All-Energy 2025 highlighted just how interconnected the energy transition has become. Whether it’s the strain of AI on the grid, the practical realities of consumer-led flexibility, or the importance of building trust in digital systems, the message was clear: energy, data, and technology strategies can no longer be developed in isolation. Get in touch with our team today to explore how to bridge the gap between tech and the transition.