AI Race Demands 160% More Power by 2030

AI Boom Fuels Massive Energy Demand Surge

The artificial intelligence (AI) revolution is dramatically reshaping global energy needs. A new forecast from Goldman Sachs suggests AI could drive a staggering 160% increase in data center power demand by the year 2030. This surge is already influencing capital flows, with data centers becoming one of the world’s largest energy consumers.

Ameresco CEO Highlights Urgent Need for Energy Infrastructure

George Sakellaris, CEO of Ameresco, a leading energy efficiency and infrastructure company, recently discussed these critical issues in a Fox Business interview. He emphasized that to win the AI race, companies must build energy infrastructure independently, a concept he calls “inside the fence.” Waiting for existing utility companies to provide the necessary power is not a viable option, as they often lack the capacity.

Aging Infrastructure and Growing Demand

The United States faces a dual challenge: aging power infrastructure and rapidly increasing energy demand. According to Sakellaris, a significant portion of the nation’s power plants are nearing retirement. By 2025, about 50% of existing power plants are planned for retirement. This trend will continue, with another 35% expected to be retired by 2035. This means new power generation and distribution systems must be built to replace the old ones and meet future needs.

Compounding this issue is the push to electrify various facilities and technologies, further escalating power requirements. The cost of this new energy supply is often higher than what current utility rate bases can support. This gap is a primary reason consumers are experiencing rising electricity rates.

“Not in My Backyard” Dilemma

The rapid growth of data centers, essential for powering AI, is creating a “not in my backyard” (NIMBY) problem in many areas. States like Virginia and New Jersey are already grappling with the challenge of accommodating new data center developments due to their immense power needs. This trend highlights the urgent need for robust energy solutions that can be deployed quickly and efficiently.

Distributed Generation as a Solution

Sakellaris pointed to distributed generation as a key solution. He cited a startling statistic: by next year, the demand for national power to support AI could reach up to 68 gigawatts (GW). For comparison, California’s entire grid capacity is around 88 GW. Relying solely on the existing grid to meet this demand is unrealistic.

Instead, Sakellaris suggests a combination of renewable energy sources. For instance, obtaining 15 to 20 megawatts (MW) from wind power and another 15 to 20 MW from solar could provide a significant portion of the needed energy. This approach, known as distributed generation, allows for localized power production, reducing the strain on the national grid. Additionally, large industrial users and data centers themselves can install their own generation capacity, adding another 20 to 30 MW.

Innovative Energy Solutions

Companies are exploring innovative solutions, including micro-nuclear reactors and on-site power generation, to avoid overburdening local communities. Ameresco is investing in research and development for micro-scale nuclear technologies and micro-turbines, aiming to have them operational by 2028 or 2029. This move towards independent energy solutions is crucial for supporting the continued growth of AI and other power-intensive industries.

Market Impact and Investor Considerations

The AI boom presents significant opportunities and challenges for the energy sector. Investors should watch companies focused on energy infrastructure, renewable energy development, and energy efficiency solutions. The increasing demand for power means that utilities and energy providers that can adapt and expand their capacity quickly will be well-positioned.

For consumers, the short-term impact may include higher energy costs as infrastructure is upgraded and new power sources are developed. However, in the long term, increased investment in diverse and efficient energy solutions could lead to greater grid stability and potentially more predictable energy prices. The trend towards distributed generation also offers a path for greater energy independence and resilience for both businesses and communities.

“To win the AI race, we have to build what we call ‘inside the fence.’ Otherwise, they are power plants.” – George Sakellaris, CEO of Ameresco

Source: Ameresco CEO on AI race: We have to build INSIDE DEFENSE (YouTube)