As searing heat waves sweep across the United States, our power grid faces unprecedented strain. These extreme conditions coupled with the advancement of artificial intelligence (AI) technologies, particularly large language models (LLMs), is creating an additional, less visible strain on our energy infrastructure. This convergence of energy challenges and technological progress presents a critical inflection point: the future of AI and the resilience of our power grid are now inextricably linked, demanding a new approach to energy security and cross-industry collaboration.
Chip Pickering, CEO of Incompas, spoke with Telecom Review about this future.
As countries recognize that AI is going to be critical to their economic productivity and national security, the energy demands are growing exponentially. Research suggests that by 2027, AI could consume as much as 85 terawatt-hours annually — equivalent to the power usage of a country like the Netherlands. As the technology becomes more ubiquitous, the increases in demand will be dramatic.
Our grid here at home, already vulnerable to climate-induced stresses, is ill-prepared for this surge. The North American Electric Reliability Corporation (NERC) warns that two-thirds of North America faces elevated risks of energy shortfalls during peak summer conditions. The Texas power crisis of 2021, which left millions without electricity and resulted in at least 246 deaths, remains a stark political and societal reminder of the consequences of grid failure.
The imperative for a sustainable and resilient grid is clear. AI companies require a reliable, clean energy supply to mitigate their environmental impact. Simultaneously, they need grid stability for the consistent development and deployment of AI technologies that increasingly underpin critical infrastructure and services — and our international competitiveness and security in an unstable world.
There can be no doubt that it requires a multifaceted approach and unprecedented collaboration between tech giants, energy providers, and policymakers. That’s the conversation we’re convening right now at the AI Competition Center.
Here are key areas that we believe will require immediate attention as our homegrown AI sector continues to boom:
- Renewable Energy Integration: While the shift to renewables is crucial and moving in the right direction, it presents its own challenges. The intermittent nature of solar and wind power necessitates advanced energy storage solutions and smart grid technologies. Nonetheless, The National Renewable Energy Laboratory (NREL) estimates that smart grid technologies could enable the integration of up to 55% variable renewable energy on the grid without compromising reliability.
- Nuclear Power Reconsideration: Nuclear energy, despite its controversies, offers a stable, carbon-free baseload power source. Advanced nuclear technologies, such as small modular reactors (SMRs) promise enhanced safety and flexibility. Including nuclear in our energy mix could provide the reliability needed to support AI’s growing energy demands. For instance, according to the U.S. Energy Information Administration, nuclear power plants operated at a capacity factor of 92% in 2021, significantly higher than other energy sources.
- AI-Driven Energy Efficiency: Ironically, AI itself may offer extraordinary solutions to the tension it is exacerbating. Google’s DeepMind, for example, has demonstrated AI’s potential in reducing data center cooling energy by up to 40%. Similar applications across the grid could significantly optimize energy use. Microsoft is on the scene, too. At their facility in Redmond, AI orchestrates the cooling systems with precision, optimizing efficiency and reducing energy waste.
- Smart Grid Development: The implementation of smart grid technologies, capable of real-time load balancing and predictive maintenance, is crucial. The bipartisan infrastructure law includes $73 billion for power infrastructure, a significant portion of which is earmarked for smart grid technology. This investment underscores the critical role smart grids will play in ensuring grid resiliency as AI continues to evolve and expand.
- Green AI Research: The AI community must prioritize the development of more energy-efficient algorithms and hardware; the types of products that can take advantage of this transformation and direct towards our most immediate and pressing needs. Recent work, such as the “Green Algorithms’’ project, aims to help researchers measure and reduce the carbon impact of their computations.
To move these solutions from concept to reality, we must convene a diverse coalition. The AI Competition Center is building a once-in-a-generation cohort that brings together tech companies, energy providers, grid operators, policymakers, and environmental experts. Some of the themes we’re actively exploring and investigating as a coalition are as follows:
- Establishing industry standards for AI energy efficiency
- Promoting R&D on how AI can itself transform our energy system
- Developing a roadmap for grid modernization that accounts for AI’s projected energy needs
- Creating incentives for AI companies to invest in renewable energy and grid infrastructure
- Formulating policy recommendations that balance innovation with sustainability
- Showcasing cybersecurity advancements that harness AI to protect our grid from adversaries and emerging threats
The path forward requires more than technological solutions; it demands a paradigm shift in how we approach energy and technology development. The conversations that may have existed in distinct silos up now, require consolidation and openness. For example, tech companies must recognize their responsibility in shaping a sustainable energy future and actively participate in grid enhancement efforts — that requires partnership and deep integration with a legacy sector.
This conversation is existentially critical — we must move urgently to propose economically and technically durable solutions for generations to come. By acting now to foster collaboration between AI innovators, energy providers, renewables builders, and policymakers, we can create a resilient, sustainable grid; one that is capable of powering the AI revolution and securing our international competitiveness, while mitigating its environmental impact. The future of AI — and indeed, our technological progress as a whole — depends on our ability to solve this energy equation. If you are working to answer these questions — from the energy side, tech, infrastructure, AI, or cyber — come join us.
