Artificial Intelligence is driving transformation in the energy industry at an unrelenting pace

The Transition-AI 2025 conference, held last Thursday at our Boston location, brought together industry leaders to discuss the impact of Artificial Intelligence (AI) on the energy sector. The event highlighted the need for adaptation and innovation as AI reshapes the landscape of energy production and consumption.

In a significant shift, hyperscalers like Google are moving away from traditional methods of procuring renewable energy. Instead, they are investing directly in capacity and flexibility to build a holistic energy strategy. This new approach values scalable storage, demand response, real-time arbitrage, and nuclear energy, as emphasised by Caroline Golin, a former global head of market development and innovation at Google.

The scale phase of AI is currently underway, posing challenges for developers, utilities, and grid operators. The industry must prepare for a constantly evolving landscape that values massive data centers today while planning for a more nimble, distributed, and flexible approach in the future.

Regulators, utility planners, data center and energy project developers, and customers will have to adapt to these changes. Reggie Singh from the US State Department noted that the administration has issued executive orders supporting nuclear and geothermal energy, but there will be no top-down requirement for clean firm power. This leaves room for regulators to protect ratepayers from paying for infrastructure devoted to data centers, and they may need to adopt new regulations or change the entire culture of regulation.

Mike Kramer, VP of data economy strategy at Constellation Energy, expressed intent to bring back the Three Mile Island nuclear power plant to its previous operating condition or better. Dawn Owens from Fervo Energy stated that clean firm power is valued between $15 and $20 per megawatt hour.

Infrastructure investors are combining or blending digital and energy teams and capital to approach AI infrastructure as a unique market challenge. Jigar Shah, former DOE Loan Programs Office director, stated that there is no capital problem in the market, and deals for 200 megawatts with batteries are getting done this week.

Speakers at the conference agreed that forecasting power demand from AI data centers is challenging, but it is expected to increase. Peter Nulsen from Generate Capital mentioned the complexity of building both energy projects and massive data centers simultaneously, with multiple entities involved.

The conference concluded with a hopeful consensus that the industry will have to change the way it has operated for the last 50 years to meet the demand for exponential growth. To stay updated on the latest news, podcasts, and analysis, subscribing to the Latitude Daily, The Latitude Weekly, and AI-Energy Nexus newsletters is recommended.

A version of this story was published in the AI-Energy Nexus newsletter on June 18. However, further research did not yield information about who published this specific report.