Dear Readers,

Happy Saturday!  Before we get into today's DeepDive, just a short announcement out of the Superintelligence lab: we are launching an AI tool on ProductHunt called Cracked.ai.  It uses agents to 100% automate your social media marketing; take a look and upvote us here: cracked.ai/upvote.  Thanks!

What if the real story behind the AGI race isn’t the models, but the power plants rising around them? Everywhere you look, the world’s biggest tech companies are pouring billions into solar fields, nuclear restarts, and massive GPU campuses that look more like industrial cities than server farms. In today’s issue, we dig into this new “concrete layer of AI”: the trillion-dollar data-center boom, the scramble for clean and reliable electricity, and the surprising bets shaping how Microsoft, Amazon, Google, OpenAI, xAI and China plan to power the next generation of intelligence. If you want to understand where AI is truly headed, this is the part of the map you can’t skip - let’s dive in.

All the best,

Kim Isenberg

If you want to understand how serious the AGI race has become, don’t look at the models - look at the concrete and copper. Behind every “GPT-5-level” headline sits an astonishing build-out of data centers and power plants. Analysts now expect up to 6.7 trillion dollars in global data-center investment by 2030, with about 5.2 trillion solely on AI-optimized facilities: GPU mega-farms built explicitly for training and running large models.

”Our research shows that by 2030, data centers are projected to require $6.7 trillion worldwide to keep pace with the demand for compute power. Data centers equipped to handle AI processing loads are projected to require $5.2 trillion in capital expenditures, while those powering traditional IT applications are projected to require $1.5 trillion in capital expenditures (see sidebar “What about non-AI workloads?”). Overall, that’s nearly $7 trillion in capital outlays needed by 2030—a staggering number by any measure.”

— McKinsey

At the same time, electricity demand from data centers is set to more than double this decade, driven overwhelmingly by AI workloads and the biggest bottleneck, as Satya Nadella recently said.

“The biggest issue we are now having is not a compute glut, but it's power," Nadella said. "It’s not a supply issue of chips. It’s actually the fact that I don’t have warm shells to plug into." The remarks referred to data centers that are incomplete or lack sufficient energy and cooling capacity.”

— Yahoo Finance

The race to AGI is increasingly a race to secure land, chips, and - most fundamentally - cheap, reliable, low-carbon power.

That raises a deceptively simple question: What is the “right” energy system for an AGI world? Is it solar roofs around every data center, offshore wind farms feeding hyper scaler campuses, a quiet renaissance of nuclear power - or even fusion reactors parked next to GPU clusters? As the United States, China and Europe, alongside companies like Microsoft, Amazon, Google, Meta, NVIDIA, OpenAI, Anthropic and xAI, pour unprecedented capital into infrastructure, they are implicitly betting on very concrete answers. This article looks at those bets: who is spending what, how fast electricity demand is rising, and which power mix is emerging as the front-runner to keep the lights on for AGI-scale compute.