The Trump adminstration talks a lot about "energy abundance." But its actual policies are creating energy scarcity and the higher prices that come with it.

That will boost the utility's peak load from 9 GW to 11 GW. Esmeralda 7 could have covered that growth and then some. But as a result of the adminstration's policies, Nevadan families and businesses will pay more for electricity than they otherwise would have.

The people and the economy of Nevada—a purple state—are likely to feel the most pain as a result of this project cancellation. The utility in Nevada, NV Energy, is projecting power consumption from data centers to grow by 373% by 2030.

As Jael Holzman at Heatmap writes, Esmeralda 7 was one of the many projects caught up in the Trump adminstration's attacks on clean energy. Earlier this year, the adminstration froze all federal permits for solar projects. https://t.co/hf0ig0QIJm

BLM just quietly updated its website to say that one of the largest planned clean energy projects in the world has been cancelled. The Nevada project would have delivered 6.2 GW of power—enough electricity to supply millions of homes.

Learn why these women are choosing Waymo over traditional ride-hailing services and say they feel much safer.

A similar dynamic led to some famously wrong forecasts about solar and wind's growth over the last decade. I wouldn't be surprised if we saw something similar happen with battery storage.

Even more so than in the past, there seems to be a disconnect between what the models say should happen and what the folks building this stuff think will happen.

I've been looking at the data on battery storage for a few years now. Forecasters keep assuming the growth will stop. And they continue to be wrong.

All of these challenges have led many energy modelers to forecast a decline in storage additions next year. But I think there's a good chance these pessimistic forecasts will be wrong.

$RGBP $RGBPP Regen BioPharma is presenting at the Emerging Growth Conference on October 23, 2025 at 4:10 pm Eastern. We will cover Orphan Drug status and HemaXellerate Clinical Phase I status. Also planning on answering questions from the audience.

The storage market faces a lot of challenges that didn't exist just 6 months ago. Tariffs threaten large cost increases. The Trump adminstration keeps passing bills aimed at curbing renewable energy's growth. And the adminstration's FEOC rules will soon add new red tape.

Over the last 12 months, the US has added more than 13 GW of utility-scale battery capacity. That's 50% more than just a year ago.

But for now, the boom continues—and with it so does the insatiable thirst for electricity that comes with it.

Over the last month, it's been impossible to avoid articles arguing that this is all a massive bubble. The similarities between this infrastructure build out and previous bubbles are certainly there.

Add all that up and OpenAI is saying that it wants to use $800 billion worth of AI infrastructure over the next few years. (Much of the upfront capex will be spent by the company's partners like Oracle and private equity firms; then the capacity will be leased to OpenAI).

These chips—and the infrastructure needed to host them—aren't cheap. A single NVIDIA chip costs about $60,000. NVIDIA CEO Jensen Huang estimates that every 1 GW of data center capacity costs $50 billion—most of it flowing to the trillion dollar chip giant.

So over the next few years, OpenAI is trying to deploy 16 GW of chip capacity, which is about as much as every data center in the US uses today to serve up Netflix shows, memes, and T Swift albums.

Today data centers in the US—the largest data center market in the world by far—use about 20 GW of power. Last month, OpenAI announced a deal with NVIDIA to buy 10 GW of the company's chips. Today, the company announced its 6 GW deal with AMD.

OpenAI just announced a new deal to buy 6 GW of AMD's newest chips. Here's one way to put that number in context: OpenAI's chips could soon use as much power as every data center in America uses today.

There's been much discussion about electricity demand—largely driven by AI data centers—in recent years. And yet, somehow, I think we're all mostly underestimating just how consequential all of this will be.

But it's increasingly clear that *much* of this demand will materialize. ERCOT has signed contracts with 26.7 GW (~23%) of the 116.9 GW that is expected to come online by 2030.