Oversized loads

Load growth is coming, any way you slice it

Hi there,

Feeling green shoots of spring here in New York. Time to emerge from months of hibernating in my cave, writing and combing the internet to curate everything under the sun worth knowing about in climate and energy. Just kidding, there’s probably a lot more of that in my future still.

Lots in the news of late, including the SEC finally getting its new climate disclosures fit to print. We’ll round up all the headlines on Sunday, as always, and will have a piece out on Tuesday re: the SEC stuff for our premium subs (<3). Today, we’ll keep things foundational with perspective on the gargantuan growth in energy demand coming out to 2050.

The newsletter in 50 words: Across industries, energy demand will rise significantly in coming decades. Energy decarbonization doesn’t simply entail changing the ingredients of the existing pie. It requires growing the pie, too. Further, different applications will require different solutions.


This newsletter is simple. Often, however, simple challenges aren’t straightforward to solve. Demand for electricity globally is growing. It will continue to do so for decades. Sometimes dramatically. The rate of change of energy demand will accelerate more in certain industries than others as we move out to 2050, but the topline story is clear.

Countless things drive current and forecast growth in demand in the same way electricity is used for countless things. The news cycle fixates on data centers for AI and electric vehicles. These are real drivers, some of which we'll discuss more deeply today. If you zoom out globally, however, other technologies, like air conditioning, are just as salient. Air conditioning electricity use is forecast to triple by 2050. That's also a product of the fact that global temperatures are rising; the energy demand growth story is often inextricably linked with global warming itself. Global warming acts as its own demand driver.

Here's how Ramez Naam, a preeminence in climate and energy tech analysis, laid out the situation on a recent Keep Cool Podcast.

Let's do the math just for the U.S. If we electrify all ground transport, that entails a doubling of electricity consumption. If you electrify all ground transport and electrify all building heat, that's close to a tripling of total electricity consumption. This is a big deal.

Ramez Naam

Why does this matter? It matters for two reasons.

For one, the rate of change in electricity demand in the U.S. has been modest for decades. That's about to change. Time for a new paradigm.

Chart via Bloomberg

Secondly, as we concern ourselves with whether electricity comes from low-carbon sources or not, it's one thing to try to change the ingredients of a stable-sized energy pie. There's a whole 'nother layer of recipe complication if you start trying to grow the pie, too. Which it's evident we'll need to. Transitioning to low-carbon power sources while growing supply is doubly complicated.

To start with an off-the-beaten-path example of where demand will come from, we can think through how much additional demand things like electrifying heavy industry will introduce. I wouldn't blame you if you haven't thought about the continuous casting process, by which molten metals are converted to semi-finished steel products, recently. This process is highly energy-intensive. If you want to electrify steelmaking, you'll need a lot more electricity, as well as ways to turn low-carbon electricity into high heats.

What the continuous casting process looks like (Shutterstock)

China, the world’s largest producer of steel, knows this. Hence why China invests heavily in both new low-carbon and high-carbon (think coal) generation. Whether you look at renewables, nuclear, or coal, China is building more generation capacity than any other country. They know what’s coming. 

Hungry, hungry AI

Next, let's take a case study on a topic that's highly en vogue. The AI mania we're currently in (I think it's safe to call it that) has a lot of people thinking about the electricity that data centers will hoover up in the future. To be sure, the past decade, in which cloud computing has come to the fore and driven the continued ascendance of companies like Google, Amazon, and others, has already spiked electricity use for data centers. To date, growth in electricity demand for data centers has primarily been met by three things:

  1. Leveraging existing generation capacity: This one's relatively self-explanatory.

  2. Energy efficiency: Data center operators have spent the past decade fighting tooth and nail to squeeze every ounce of efficiency gains out of their operations. Future gains on this front are likely close to maxed out.

  3. Renewable energy: Data center owners and operators have also been busy inking PPAs with every bit of renewable energy capacity they can affordably acquire. However, there are limits to this as data centers get bigger and given the intermittency of renewables sans storage.

The above levers have been pulled. What's coming next will require new solutions. Bloomberg recently noted that:

Electricity consumption at US data centers alone is poised to triple from 2022 levels, to as much as 390 terawatt hours by the end of the decade, according to Boston Consulting Group. That's equal to about 7.5% of the nation's projected electricity demand.

As the AI boom heralds another exponential leg up in electricity demand, stakeholders are scrambling to figure out which generation technologies they can depend on. If you look at the landscape as a whole, the approach is scattershot. Sam Altman, the CEO of OpenAI, invests in both nuclear fission companies, especially ones designing small modular reactors, and fusion energy developers, like Helion. Google has a partnership with Fervo Energy to develop geothermal energy resources. In this sense, the evolution of data centers mimics the evolution of future grids as a whole. Everyone is trying to figure out which 'always on' generation sources, whether fission, fusion, geothermal, or other, will work, both in terms of cost, viability, and practicality.

To this end, it's worth calling out two things that happened this week:

  • Construction was completed on a small modular reactor in China. The reactor, Linglong One, is an ACP100 modular pressurized water reactor originally developed by China National Nuclear Corporation, rated at 125MWe, and being built in Hainan province. As much discussion as SMRs get, this is a world first.

  • AWS, Amazon's cloud computing unit, bought a 960 MW data center co-located with a nuclear power plant in Pennsylvania for ~$650M. Reading the tea leaves, we might infer that Amazon is placing its chips on the nuclear fission side of the table regarding what generation technology will meet the growing electricity demand for data centers.

If anything could catalyze a nuclear energy renaissance in the West, perhaps AI and data center electricity demand will. In the East, China, Russia, India, and other countries are already more actively building new nuclear reactors.

While the race to secure electricity for data centers offers corollaries for meeting growing energy demand in general, data centers differ from, say, industrial decarbonization, which we touched on earlier, in one fundamental way. Data centers spit off a lot of heat, whereas industrial processes often demand high heat. What 'works' for data centers might not always work for industry.

Ultimately, if you have enough power, you can do almost anything—whether it's heating, cooling, or something else you need. As long as we're in an energy-constrained environment however (see below graphic), specific technologies, like thermal batteries that store solar energy as heat, can cater to specific applications (like to service industry). Specific tech won’t work for all applications however; thermal batteries that store heat won’t be useful to data centers, where the opposite of heat is needed (i.e., air or liquid cooling).

Energy consumption growth has fallen off a previously exponential trajectory. Time to change that again.

*quick rant*

There are countless other applications and electricity hogs we could introduce into our analysis here. If the world wants to use green hydrogen to displace carbon-intensive processes like steam methane reforming to make fertilizer, that requires electricity. If the world wants to decarbonize shipping by using hydrogen, or ammonia, or methanol, the same axiom holds. As we noted earlier, if the world wants to use more electric arc furnaces in steelmaking and decarbonize other components of industrial production, that requires electricity. If the world wants to do large-scale carbon removal, that requires electricity. Heck, there are almost a billion people the world over who don't have access to electricity as is. Microgrids, a combination of renewable energy and storage, can undoubtedly service some of those folks. But you get the idea.

Building off these ideas, it's time for a small rant. Every iterative article arguing about whether something is or isn't a good use of energy is, to me, a bit silly. Sure, you may think cryptocurrencies like Bitcoin are dumb and that they shouldn't be allowed to consume whatever percentage of U.S. electricity generation they currently do. I won't argue with you on Bitcoin's merits or demerits, at least not here. Meet me at the bar for that.

What I’d offer instead is that the precedent of picking and choosing what's 'in' vs. 'out' in terms of electricity use could get quite slippery quite quickly. 

Our collective goal should be to create sufficient low-carbon power so that we can use it for all kinds of cool applications, whether for data centers, desalination, fancy EVs, or other. Whether or not those applications jive with your worldview is a separate conversation. The climate and energy battle won't be won by rationing ourselves into the bare necessities. At least not as long as the prevailing economic and political systems that dominate the world are in place. 

The idea that we can somehow unwind capitalism and all its attendant structures strikes me as sclerotic at this point. We either create an energy-abundant future powered by low-carbon generation sources, or we don't. Whether we do or don't has profound implications for our warming trajectory.

The net-net

Zooming all the way out, there’s a relatively simple expression of greenhouse gas emission as a product of four variables that’s useful to reflect on. The variables are population, GDP, the energy intensity of said GDP, and the carbon intensity of the energy. If you multiply the four variables together, you get a figure representing global carbon dioxide emissions.

The world has decoupled emissions from GDP to an extent. Still, population growth is doing its thing, marching right along. And GDP per capita is growing too, which is good if you care about people’s well-being, even if it isn’t ‘good’ for our emissions trajectory. The point is that modest reductions in average carbon intensity of power generation don’t meaningfully bend the carbon dioxide emissions curve as the global population grows and GDP per capita does, too.

Absent a low-carbon energy revolution, one that significantly reduces the carbon intensity of energy as use of energy growsdeep decarbonization will be hard to come by. The goal should be to expand energy access dramatically with very low-carbon power. No, I’m not saying we should ignore energy efficiency opportunities. Those are always welcome. But it’s time to welcome our oversized load overloads. They come in peace if we can rise to the occasion.


Are you a climate or energy tech company that’s growing like a weed? Looking to hire exceptional talent? Keep Cool is laser-focused on matching fantastic talent from within our ecosystem with roles at your company.

We’ve had excellent conversations with launch partners over the past week but have space for a few more partners (maybe you?). If you’re in, we’ll work intentionally and collaboratively with you to tailor our process to your needs and supercharge the quality of folks you’re interviewing.

See here for more and respond directly to this email to get in touch.

Go get outside!

– Nick