Hosts: Rohan Kapoor & Amara Diallo
In this episode:
• Today we're tracking the hyperscaler data center power crunch that's completely reshaping grid planning, and how nuclear is making a surprising comeba...
• Rohan, let's start with this data center sit
Daily AI news for energy and sustainability professionals. Two hosts cover how AI is powering the clean energy transition, grid management, and climate solutions.
Rohan Kapoor: Welcome to Pivot Energy! I'm Rohan—
Amara Diallo: —and I'm Amara. Let's get into it.
Rohan Kapoor: Today we're tracking the hyperscaler data center power crunch that's completely reshaping grid planning, and how nuclear is making a surprising comeback as AI's clean power solution.
Amara Diallo: Rohan, let's start with this data center situation because the numbers here are genuinely staggering. We're talking about AI facilities that need as much power as entire cities, and utilities are scrambling to figure out how to deliver it.
Rohan Kapoor: The scale is what gets me. Microsoft alone is projecting they'll need 5 gigawatts of new capacity by 2028 — that's equivalent to powering 3.75 million homes. And they're just one player. When you add up Google, Amazon, Meta, and the others, we're looking at 30 to 40 gigawatts of new demand hitting the grid in the next five years.
Amara Diallo: And here's where it gets really interesting — this isn't just about finding the power. The entire supply chain is buckling. Steel manufacturers can't produce transformers fast enough, there's a two-year waitlist for high-voltage equipment, and transmission lines that normally take a decade to approve need to happen in half that time.
Rohan Kapoor: Yeah, and I'm seeing utilities completely rewrite their planning models. Dominion Energy in Virginia just tore up their 15-year forecast because data center growth obliterated every projection they had. They're now planning for 85% load growth over the next decade versus the 15% they expected just two years ago.
Amara Diallo: What's fascinating is how this is forcing unprecedented coordination. I'm hearing about weekly calls between hyperscalers and grid operators, joint task forces on interconnection queues, and utilities actually turning down data center projects because they simply can't guarantee the power.
Rohan Kapoor: That rejection rate is key — PJM just reported they're denying 40% of data center interconnection requests due to capacity constraints. That's up from basically zero denials three years ago. The grid literally cannot keep pace with AI compute demand.
Amara Diallo: Which brings us perfectly to our second story — nuclear power emerging as the solution to this mess. And Rohan, this isn't your grandfather's nuclear industry anymore.
Rohan Kapoor: The Three Mile Island restart is the headline grabber — Microsoft signing a 20-year deal to bring a shuttered reactor back online specifically for their data centers. But let's talk economics here. They're paying $100 per megawatt-hour, which is nearly double current wholesale prices. That tells you how desperate they are for reliable, clean baseload power.
Amara Diallo: And it's not just traditional reactors. The Air Force just approved deployment of microreactors at Eielson base in Alaska — these are shipping container-sized units producing 1 to 5 megawatts each. Perfect for edge computing facilities or backup power for critical AI infrastructure.
Rohan Kapoor: The regulatory piece is huge though. The Nuclear Regulatory Commission fast-tracked approvals for these micro designs, cutting review time from 42 months to 18. That's a sea change in how we think about nuclear deployment.
Amara Diallo: What really excites me is the fusion angle. Commonwealth Fusion Systems just announced they've presold their entire first decade of output to data center operators. We're talking about fusion plants that don't even exist yet, but hyperscalers are so power-hungry they're betting billions on unproven technology.
Rohan Kapoor: Though let's be realistic — their 'commercial operation by 2030' timeline has already slipped twice. And even if fusion works, we're talking about $150 per megawatt-hour minimum. That's premium pricing that only makes sense for AI workloads where compute power directly equals revenue.
Amara Diallo: True, but consider the alternative. Google's latest sustainability report showed their emissions increased 48% year-over-year, entirely due to data center growth. They either solve this with nuclear or abandon their net-zero commitments. There's no middle ground anymore.
Rohan Kapoor: And that tension is reshaping energy markets in real-time. Natural gas plants that were scheduled for retirement are getting life extensions because grid operators need dispatchable power to balance all these data centers.
Amara Diallo: The irony is thick — AI that could help solve climate change is driving emissions up in the short term. But Rohan, I genuinely believe this nuclear renaissance, especially these modular designs, could transform how we think about distributed clean energy.
Rohan Kapoor: Maybe, but the numbers have to work. Right now, hyperscalers are paying whatever it takes because AI revenue justifies it. The real test comes when that AI gold rush cools down and they're stuck with these premium power contracts.
Amara Diallo: Fair point, but I think we're past the point of no return. The infrastructure being built now — the transmission lines, the nuclear facilities, the grid upgrades — that's permanent change to our energy landscape.
Rohan Kapoor: That's your Pivot Energy briefing for April 24, 2026. I'm Rohan—
Amara Diallo: —and I'm Amara. See you tomorrow.