Both True · edition
AI is revolutionizing industries, but its hunger for compute power is putting unprecedented strain on energy systems. Today, we weigh the growing energy demand from data centers, projected to quadruple by 2035.
This isn’t just about faster algorithms or smarter systems—it’s about whether the electrical grid can keep up, who gets left behind, and what the planet can bear. Let’s dig in.
The lead
STRAIGHT TALK
Data centers are expected to use one-fifth of the electricity generated in the U.S. by 2035, driven by AI compute demands, according to BloombergNEF. Nearly half of the projected 200 gigawatts of data center capacity will be devoted to AI training and inference. Grid operators like PJM Interconnection and ERCOT face significant strain as electricity demand surges. (Source: TechCrunch) →
The case for
AI-driven data centers represent transformative potential for industries spanning healthcare, transportation, and climate modeling. Nearly half of the projected 200 gigawatts of capacity will power AI training and inference, enabling breakthroughs like early cancer detection, autonomous vehicles, and faster drug discovery. The U.S., hosting 64% of AI chip power demand by 2033, could cement its leadership in global AI development. This growth also drives innovation in renewable energy, as solar and wind projects rush to meet surging electricity needs. For instance, solar power is already setting records, with developers planning an additional 43.4 GW of utility-scale capacity in 2026 alone. If managed strategically, this demand could catalyze a greener grid, pushing utilities to modernize infrastructure and expand renewable adoption. In short, while this energy spike is daunting, it also represents an opportunity to rethink energy systems for a high-tech, low-carbon future.
The cost
The projected quadrupling of data center energy demand comes with enormous challenges and risks. In the PJM Interconnection region, electricity prices have already risen 76% in one year, driven by congestion and capacity strain. The grid operator had to pause new connection applications for four years, delaying critical infrastructure projects. This isn’t just a regional issue—global data center energy demand could reach nearly 1,935 terawatt-hours by 2033, almost matching India’s annual consumption. Utilities like American Electric Power are threatening to exit PJM due to mounting grid congestion, highlighting the fragility of the system. Building renewable energy capacity fast enough to meet demand is another hurdle, as supply chain constraints and permitting delays persist. Meanwhile, communities near these data centers often bear the brunt of higher energy costs and environmental impacts, such as water use for cooling. The AI boom risks exacerbating inequities, leaving some regions and populations behind as others reap the benefits.
Terms, plainly
Context
AI’s energy demands aren’t coming out of nowhere. Over the past decade, the rise of cloud computing and big data has steadily increased data center capacity worldwide. Now, AI accelerates this trend, especially with models like GPT-4 and beyond requiring vast amounts of compute for training. Grid operators like PJM and ERCOT are already struggling with capacity and connection issues, highlighting the need for faster grid modernization. Meanwhile, renewable energy is expanding, but not without challenges like supply chain bottlenecks and policy uncertainties. Looking ahead, the balance between meeting AI’s needs and ensuring grid reliability will shape both the tech and energy sectors. Watch for breakthroughs in energy efficiency and grid management technologies.
Both true
The AI revolution is a double-edged sword here: it drives incredible innovation but tests the limits of our energy systems. Renewable energy can help, but only if the grid expands quickly enough to keep up. This isn’t a story of tech versus the planet; it’s a story of whether we can align growth with sustainability. For now, the tension remains unresolved, and the stakes couldn’t be higher.
Also today
HUMAN IMPACT
AI is accelerating cancer detection and drug discovery, according to a Johns Hopkins surgeon. Tools for early detection of pancreatic cancer show promise in improving outcomes for one of the deadliest cancers. (Source: Fox Business) →
Why it mattersAI-driven healthcare breakthroughs could save lives and redefine early disease detection.
TECH & PLANET
American solar farms produced their highest monthly output ever, up 21% from last year. Developers are on track to add 43.4 GW of new utility-scale solar capacity in 2026, while AI data center demand drives electricity use to record highs. (Source: 24/7 Wall St.) →
Why it mattersRenewables are rising to meet growing energy needs, reshaping the U.S. grid in real time.
FRONTIER
Emory physicists received U.S. Genesis Mission awards to study plasma dynamics and deep space structures like supernova remnants. The research leverages AI to accelerate discoveries about phenomena like the Crab Nebula. (Source: Emory News) →
Why it mattersAI is pushing the boundaries of fundamental science, from deep space to fluid dynamics.
My analysis
The bigger picture
AI’s energy appetite is reshaping the grid at every level, from record demand to record renewable output. The stakes are clear: we need tech-driven solutions that work for both innovation and sustainability.
What it means for you
AI is driving the future—and forcing us to rethink how we power it.