The race is on. Big Tech and startups alike are jostling to seize a generational market opportunity: artificial intelligence. Checkbooks are open to pay for the enormous computing capacity needed to scale AI—as well as the enormous amounts of electricity needed to support it.
Data centers, home to AI’s computing power, could consume more than 9% of U.S. electricity generation annually by the end of the decade—double or even triple today’s figure, according to a forecast from the Electric Power Research Institute.
The issue: We need power growth to accelerate in parallel, which means building more plants—ideally clean-energy ones—and routing electricity to where AI needs it.
That’s created a fascinating side effect of the AI era that may well have as large an impact as AI itself: rapidly scaling critical innovations in our power networks.
Never in my energy career have I seen a moment like this, when market forces are far outweighing regulatory guidance to substantially accelerate utility innovation. Utilities have long been conservative about embracing innovation, as introducing risk to the grid can undermine reliability—and, in turn, our entire economy. Moreover, the unique, regulated-monopoly structure of utilities means they don’t prioritize market demand; instead, they orient around regulatory requirements.
A recent report featured findings from a survey of leaders at utilities who are charged with surmounting structural obstacles to scale. About seven in ten (72%) of these leaders—those involved in innovation at their companies—admitted that in-house innovation is primarily driven by regulation or compliance. Moreover, just a quarter of those executives surveyed said they regularly identify fresh ways of thinking and operating from startups working in the utility industry. That’s a serious drawback, because startups play a vital role in creating tech breakthroughs that disrupt regulated markets and create new ones—look no further than Airbnb, Uber, and OpenAI. And that’s exactly what utilities need to meet this moment.
Utilities have been risk-averse for good reason, but it’s clear that in today’s fast-moving environment, slow-rolling innovation is far riskier—and could spark fierce competition from new entrants. OpenAI’s ChatGPT has upended legacy tech in just a few years—utilities could easily be up next.
However, AI is already changing utility business as usual–and promises to be just what the industry needs to move forward.
The AI acceleration
Winning the AI race is an existential challenge for tech companies that must innovate or vanish. They’re demanding more power for AI, and they want it to be clean. And they’re pushing utilities to move faster across every aspect of our business, from onboarding renewable resources to rewiring our infrastructure to working with regulators.
Consider the power of Big Tech’s net-zero goals. While many utilities—including mine—have committed to eliminating carbon emissions by 2050, tech giants including Apple, Google, Microsoft, and Meta aim to hit net zero (or even carbon negative) a full two decades earlier. That urgency means they need carbon-free energy throughout their supply chains now to power AI and everything else they do. Fossil fuels simply don’t cut it.
That’s why we’re seeing unprecedented interest from Big Tech in renewables as well as in nuclear facilities, where near-term costs are outweighed by massive potential benefits over the long term.
The rise of AI is also pushing utilities to accelerate infrastructure investments and startup partnerships. Because utilities must route enormous volumes of electricity from power plants to the growing galaxy of data centers, they’ll need far more transmission and distribution capacity. Utilities and their customers also will benefit by making power lines more efficient—adopting technologies from startups like LineVision, TS Conductor, and VEIR that can expand the amount of power transmitted over the same physical footprint by as much as tenfold.
More efficient AI means more energy demand, not less
The history of computing has shown us a clear pattern: When technology becomes more accessible, adoption scales exponentially—and so does energy consumption. We saw it when computing expanded from mainframes to personal computers, and we’re seeing it again with artificial intelligence. The rise of models like DeepSeek, which promise more efficient and democratized AI, won’t reduce energy demand. Instead, they will accelerate adoption across industries, widening the base of users and driving an even greater need for electricity.
Until recently, AI was largely the domain of hyperscalers and cutting-edge enterprises with the resources to afford energy-hungry compute infrastructure. But as AI models become more optimized and efficient, they will no longer be confined to high-end data centers. We are entering a post-DeepSeek world where AI will be pervasive across enterprise applications, consumer devices, and industrial operations.
Three major shifts will contribute to the net increase in energy demand:
- Data center expansion: Even as AI models become more efficient, overall compute demand will grow. More businesses will deploy AI, leading to a rapid buildout of hyperscale and edge data centers worldwide. These centers, in turn, will require greater power capacity and more renewable energy integration.
- Enterprise AI adoption: Companies that previously hesitated due to cost or complexity will now find AI accessible. From real-time analytics to AI-powered automation, businesses will integrate AI across supply chains, customer service, and decision-making processes—further amplifying compute requirements.
- Consumer AI explosion: Lighter AI models will enable local processing on personal devices, shifting some workloads from the cloud to edge computing. However, rather than reducing energy consumption, this decentralization will create new demands on distributed energy systems, powering a global network of AI-enabled devices.
History tells us that technological advancements in efficiency often lead to greater overall consumption—a phenomenon known as Jevons’ Paradox. AI will be no different. As AI-powered applications become more widespread, they will unlock new use cases, many of which will be compute-intensive. The result? The global energy footprint of AI will grow, not shrink.
For utilities, this means that AI isn’t just a disruptor—it’s a catalyst for accelerated innovation. To meet AI’s growing energy appetite, we must scale clean energy production, build more resilient power infrastructure, and integrate AI-driven efficiencies into grid operations. The utilities that embrace this challenge will be at the forefront of powering the next wave of digital transformation.
The AI revolution is not just about intelligence—it’s about energy. And the race to power it has only just begun.
Using AI to enable AI
Ironically, artificial intelligence itself can help us solve the challenges of powering the AI industry. Utilities are starting to deploy the technology everywhere, from power forecasting to planning and operations, helping to bring power supply to demand more efficiently than ever. Imagine harnessing AI to optimize solar energy production during sunny mornings, then effortlessly swapping over to battery resources during a rainstorm in the afternoon. Or driving down energy use in the evening with micro-targeted incentives and automation that utilities simply couldn’t manage before. The potential savings and carbon reductions can be dramatic—while freeing up more than enough clean power for AI.
Utilities need to get new power assets like solar up and running in months or weeks—not years, as is all too often the case. We could all learn something from Texas, which has quietly become a national clean energy leader thanks to rapid permitting. New platforms, many powered by AI, can dramatically expedite everything from helping developers check local permitting faster to automating design submissions.
AI is already at work in the grid today, improving everything from reliability to electric vehicle charging. Companies like EV.energy can sync EV charging for when overall grid demand is low, saving millions for drivers while optimizing the drain on the grid. And there are surprising applications of AI as well: AiDash uses AI to find trees most likely to fall on power lines and spark outages or fires. By trimming those trees first, National Grid has slashed power outages by 30% and slimmed the duration of blackouts by 55%.
By using AI across operations, leaders like Microsoft cofounder Bill Gates anticipate we will more than compensate for the rising energy requirements of AI itself. Early deployments confirm this opportunity, but we’ll need to deploy AI at scale.
A new utility
The AI race is forcing utilities to transform at speeds that are uncommon, even uncomfortable. The adoption of solar was a slow burn, comparatively, subject to government interventions and subsidies—territory that utilities understand well. AI is an overnight sensation powered by massive market forces and by technology leaders who expect rapid innovation. For utilities, this is a new game entirely.
Our energy system is too ubiquitous to sidestep this moment; it will have to change faster than ever. And that’s a good thing. AI is a critical catalyst to advance our industry into a thrilling new chapter where electricity is far cleaner, and the grid is more intelligent and efficient. Above all, this is a powerful opportunity for AI to deliver value across society—and I urge utility leaders to embrace this world-changing moment.
The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.
Read more:
- The most underreported and important story in AI right now is that pure scaling has failed to produce AGI
- AI ‘prompt and pray’ hasn’t cut it in the enterprise, but we’ve found the missing puzzle piece. Mass deployment is next
- IBM CEO: DeepSeek proved us right—AI is not about big, proprietary systems
- Honeywell CEO: AI will transform industry at scale beginning in 2025
