AI is perfect partner for both renewables and gas says Mitsubishi chief Cavada

The advent of artificial intelligence and its huge demand for power has upended our assumptions about energy, and with it our expectations for how the energy transition will progress.

But the advent of AI is not a zero-sum game for clean energy: its rise is not the transition’s wider loss.

While it’s true that AI is driving global electricity demand – with the expectation that it will more than double by 2030 according to the International Energy Agency – it is also driving huge progress in how we manage power demand.

If deployed wisely, AI may be one of the greatest enablers of the energy transition, with the ability to reconcile rising demand with decarbonisation in ways that were unimaginable even a decade ago.

The most immediate example lies in the power grid. For decades, power networks were built for a world where energy flowed in one direction, from large central plants to predictable consumer demand.

That world has vanished. The rise of distributed solar, wind farms, the ‘electrification of everything’ and intermittent generation has created a very complex system. And now, AI is proving indispensable in helping operators manage it.

Advanced models ingest massive real-time data, from weather forecasts and smart meters to asset health metrics and convert them into actionable decisions. AI can forecast demand, detect faults, and balance supply dynamically.

For example, according to analysis from the Center for Strategic and International Studies (CSIS), such tools enable the grid to maintain stability, schedule reserves, and recover from outages more reliably.

Have you read?
Grid challenges demand revolution not evolution from digital superheroes 

A good example is this year’s widespread power blackout in Spain and Portugal which is now largely attributed to excessive renewable energy input overwhelming grid stability and causing grid inertia to drop sharply.

This reduced its ability to respond to sudden frequency changes. Renewables lack the stabilising mass of traditional gas turbines, for example. AI, however, has the ability to predict and forestall such blackouts and optimise flows of renewable and gas generation, with a level of precision that was once out of reach.

And without this stability in the grid, we cannot integrate the renewables that the energy transition demands.

Put simply, AI shifts grids from reactive to anticipatory, and that is essential when you have intermittent renewables and volatile loads. Machine learning models are now forecasting fluctuations hours, or days, in advance, allowing operators to switch to natural gas or dispatch storage systems more effectively and avoid costly imbalances. Digital twins are also increasingly used to fine-tune solar farms, wind turbines and battery facilities.

The financial benefits are equally significant. One of the barriers to faster decarbonisation has been the cost of upgrading grid infrastructure. AI offers a way to stretch existing assets further, identifying bottlenecks and redistributing load in real time, potentially reducing or even delaying the need for expensive network expansion.

For emerging markets, where infrastructure budgets are tight, this is transformative. Instead of waiting years for new lines or substations, utilities can get more out of what they already have.

Read more:
Ember research highlights Spain’s gas reliance for grid services post-blackout

Today, natural gas remains a crucial element in ensuring reliability, especially for baseload. Here, AI is a harmoniser. By combining market data, weather forecasts and demand signals, AI systems can determine the most efficient times to operate gas turbines.

This reduces fuel consumption, lowers emissions, and cuts maintenance costs. Advanced models are now detecting faults before they cause outages, allowing predictive interventions that extend equipment life.

Around the world, we see utilities adopting AI not as an experiment, but as an operational necessity. In North America, partnerships between technology companies and grid operators are already using AI to speed interconnection decisions and ease network congestion.

For example, some large (and power-hungry) tech companies have paired with grid operators, deploying AI to speed interconnection decisions and ease blockages. At the same time, demand-response programs deployed in tandem with utilities allow AI workloads to throttle in response to grid signals, turning a ‘demand monster’ into a flexible load asset. 

Watch: How intelligent flexibility can make renewables more profitable 

These examples show that AI need not be constrained by grid limits: it can become a partner with both renewables and natural gas in balancing, flexibility, and reliability.

But none of this is to downplay the challenges. AI’s appetite for electricity is real, and its volatility adds stress to already stretched grids. Data centres can increase consumption tenfold within seconds and new demand is causing delays to power and electricity supply chains as utilities race to keep up.

These concerns are valid and demand regulation, planning and investment. However, they do not negate the larger point. The solution is not to resist AI, but to govern it wisely, channel its power, and ensure innovation is matched by infrastructure.

As such, the conversation around AI and energy needs to change. The focus should not be on whether AI consumes electricity, it does, and it will, but on how much electricity it can save, how much emissions it can prevent, and how much resilience it can inject into systems under strain.

If we continue to frame AI only as a consumer of power, we risk overlooking its far greater role as an amplifier of efficiency and enabler of decarbonisation.