The 2024 data center fantasy draft

Aaron Foyer is vice president of research and analytics at Orennia.

The data center boom is driving powerful and unexpected change across the entire energy industry, an impact usually saved only for global geopolitical events.

So (breaking the fourth wall here) my colleagues Jeff Taylor, Keaton Horner, and I sat down and picked who we believe will be the greatest beneficiaries of this once-in-a-decade event — and some who might wish we could go back to the era before ChatGPT.

In mirroring the approach long-embraced by fantasy sports fans, we set out to answer the big questions on everyone’s minds:

• First-round draft picks: Who looks primed to be the biggest money-makers of data center growth?
• Promising rookies: What emerging technologies could ultimately benefit from the rush?
• Aged vets: Are there existing energy sources on the path to obscurity?
• Potential busts: What could take a step backwards because of data center expansion?

Some pretext

Before jumping in, let’s clarify the terms of the draft.

First, this is limited to energy-related entities and technologies. As the old saying goes, during a gold rush, sell shovels. Nvidia and any startup that can share in the growing GPU chip market is going to do very well, as will anyone who can affordably improve the energy efficiency of data centers — but those players are outside of the discussion. 

As for how we’re defining “win” or “lose,” think in terms of how a player’s trajectory has changed with the infusion of data center growth into the mix. Winners are technologies or entities whose outlooks have improved, either directly or indirectly. For losers, the landscape has gotten worse; they’re getting outcompeted for finite resources or experiencing negative impacts from data center growth.

So, without further ado — and with apologies to those who don’t love sports analogies — welcome to the 2024 Data Center Fantasy Draft.

The top overall draft pick

Like any proper draft, the number one overall selection is coveted. In the case of data centers, it’s an easy choice: utilities. 

To meet the power requirements of new data centers, tens of gigawatts of new power generation will be needed by 2030 in the U.S. alone. Estimates of the exact total that will be needed to power the country’s data centers vary: McKinsey pegs it at 50 GW by 2030; Goldman Sachs has it closer to 47 GW; and BCG estimates it to be 28 GW.

At Orennia, we built a bottom-up forecast and our central case — with demand growth limited by both the number of planned-for facilities in each state and the power supply available — is for 27 GW of incremental power demand from new data centers being built by 2030. (In certain scenarios, though, we have found demand could top 40 GW.) 

Either way, Christmas came early this year for utilities. By the end of the decade, $30 to $50 billion in new power projects will be needed to meet additional power demand should this all come to pass.

Other first round draft picks

The list includes most types of power generators, gas producers and midstreamers. 

Generators: Almost every type of power generation will benefit from data centers, and those that are commercial, scalable and can connect to the grid quickly will benefit the most. With that in mind, the big winners are likely to be wind, solar, and gas.

For wind and solar, each has its merits. The higher capacity factor for wind power makes it more suitable for the low downtime needs of many data centers, though solar projects typically spend less time in development so can connect to the grid quicker.

Gas producers: According to a note published in April, Raymond James has a base case for cumulative natural gas demand growing by roughly 10 billion cubic feet per day by 2030. That’s just shy of the totality of U.S. liquefied natural gas export volumes. Lower estimates, such as one from Goldman Sachs, sees growth of more like 3 billion cubic feet daily. Regardless, the additional demand will help stymie low gas prices across the continent and warm the hearts of domestic gas producers.

Natural gas also stands to benefit from challenges in building electrical transmission. Bringing wind and solar into load centers often requires new or upgraded transmission lines. However, just 55 miles of high-voltage transmission lines were added in 2023, roughly the distance between San Francisco and San Jose. For data centers, it may be easier to connect to existing gas infrastructure, if even just for backup power. 

Midstreamers: Pipeline-building firms, meanwhile, will be needed to divert the large volumes of gas into these facilities, especially for the largest data centers, which are often hundreds of megawatts in size.

Alan Armstrong, CEO of pipeline giant Williams, said on an analysts call in August that the firm was staying busy with data centers. “We, frankly, are kind of overwhelmed with the number of [data center] requests that we’re dealing with, and we are trying to make sense of those projects,” Armstrong said.

Promising rookies

The Big Tech firms all have ambitious climate targets but are unlikely to be deterred from rapid AI development in the name of emissions. Consequently, they are starting to invest in emerging technologies that either provide low-carbon baseload power or remove carbon emissions from the atmosphere. Data center developers have not appeared to be particularly sensitive to electricity prices and seem willing to invest in earlier-stage technologies.

The following solutions, which have to date been slow to attract investors due to their high costs, will benefit from recent investments that may ultimately bring prices down. 

Next-generation geothermal: Baseload geothermal power has always been a holy grail of sorts, but new designs have not yet come down the cost curve. This may soon change as Google and its parent company, Alphabet, have been active in the space, investing in next-gen geothermal company Fervo as well as the home geothermal company Dandelion, itself a spinout from Alphabet X. Meta also recently announced funding for Sage Geosystems' first-of-its-kind plant, the first U.S. geothermal project east of the Rocky Mountains.

Nuclear: There are only a few technologies that can truly match the power needs of gigawatt-scale data centers, and nuclear is one of them. On the conventional nuclear front, Amazon acquiring the data center campus powered by the Susquehanna nuclear plant and Microsoft driving the resurrection of a Three Mile Island reactor both made headlines. 

The tech industry has also been active in the advanced nuclear space. Oracle announced a data center powered by three small modular reactors; and Oklo’s micronuclear reactor is getting attention from companies like Equinix and Wyoming Hyperscale.

Bioenergy with carbon capture and storage: More commonly abbreviated to just BECCS, this system produces electricity that is both baseload and carbon negative, so it checks multiple boxes for hyperscalers.

The power generation technology has long been viewed as one of the most expensive, but Microsoft signed deals with Ørsted and Stockholm Exergi paired with the current credit price of $160 per metric ton — a model that could spur more investment in the space. And in September, the British power generating firm Drax announced $12.5 billion of investment into U.S.-based BECCS projects, a decision the company’s president partially chalked up to data center growth.

Direct air capture: If U.S. data centers end up consuming 3 to 10 billion cubic feet per day of additional gas and increase their emission profiles accordingly, expect some atonement by tech companies in the form of new DAC projects.

These carbon removal facilities may have to be strategically located in regions where they’re not competing with data centers for power, as the cancellation of CarbonCapture Inc.’s Project Bison underscored. But multiple DAC startups have been getting attention from hyperscalers, including 1PointFive, Holocene and Climeworks. 

Green hydrogen: It’s tempting to throw any technology competing for green electrons into the potential bust category, and electrolysis-based hydrogen is particularly sensitive to high electricity prices and transmission and distribution costs. However, a behind-the-meter solution where both electricity and hydrogen are produced and used on-site may be the exception. 

Two electrolyzer startups, Sunfire and Electric Hydrogen, received investments from tech companies, and the data center startup ECL plans to build a 1-gigawatt off-grid data center powered by on-site green hydrogen. This would be another solution that could bypass the lengthy interconnection queues.

Other rookies that show promise: Renewable fuels could increasingly be used as backup for data centers. Many currently use diesel for backup generators, so renewable diesel could be swapped in to reduce emissions.

Similarly, with the jump in natural gas use, low-carbon-intensity renewable natural gas, or RNG, may be used to bring down carbon intensities, as highlighted by Microsoft’s deal with Enchanted Rock and U.S. Energy last year to supply RNG to back up a data center in San Jose.

Blue hydrogen probably also benefits in a round-about way. While green hydrogen may be used as an off-grid solution, data centers look like they’ll outcompete other loads looking for clean power on the grid. As a result, some regions will be less attractive for green hydrogen, which has the knock-on effect of making blue hydrogen look better by comparison.

Aged vets

Like Tom Brady, even long and prosperous careers eventually end. For the number of technologies that look well on their way to retirement, there is only one worth discussing.

Coal: Data centers may alter the path of U.S. coal retirements, but the likely ending remains the same. Of the roughly 180 GW of operating coal facilities in the U.S., Orennia still expects more than 90% to be retired by 2050. That said, the surge in new power demand from data centers may delay some of the expected retirements in the interim.

And why wouldn’t they? Prices from PJM’s 2025-26 capacity auction hit $269 per megawatt-day earlier this year, nine times higher than at last year’s auction. This was the result of both substantial plant retirements and increased power demand, including from data centers. These higher prices more than incentivize the continued use of existing coal plants that have modest operating costs.

How the Environmental Protection Agency finalizes the treatment of coal plant emissions will likely have a greater impact on the future of coal than data centers. The agency proposed new rules earlier this year that would force existing coal-fired power plants operating beyond 2039 to either install carbon capture or shut down by 2032.

Potential busts

The final category belongs to those whose futures are less rosy because of data center growth. 

E-fuels: All the power-to-X pathways use low-cost clean electricity to create some type of clean fuel. These include green hydrogen, e-ammonia, e-methanol, and e-fuels, all of which have green hydrogen as a foundation. Given how sensitive the production costs of these fuels are to power prices — and how insensitive data centers are — it’s expected the clean fuel facilities will either be outcompeted for finite green electrons or face incompatible power prices wherever data centers are vying to grow. 

Consumers: The Maryland Office of People’s Counsel estimated the impact to consumers of the aforementioned bombshell PJM capacity market auction to be a jump in monthly bills for homeowners and commercial businesses of between 2% and 24%. In Oregon, Portland General Electric is seeking to increase rates this year by 10.9% for consumers, higher than its estimate of a 7.4% hike earlier this year. The region is a hotbed for data centers and recently nearly doubled its 2030 power demand estimates for the power-hungry facilities.

No matter how you cut it, data centers look inflationary for consumer prices. 

And of course, emissions: In the short- and medium-term — with the expected delays in coal plant retirements and a surge of new gas demand that could rival the entire U.S. LNG export market — it’s no surprise that emissions related to data centers are expected to go up. For those keeping count, that’s a double whammy for consumers who will need to run their air conditioners a bit harder due to a warming climate and run them using higher-priced electricity. Electric vehicles may also get hurt by high power prices, which erode their lower operating cost advantage, often a key selling point. 

But there is hope. Perhaps the greatest long-term benefit we’ll see as a society is from the set of technologies labeled in this fantasy draft as “promising rookies.”

The amount of capital being put toward data centers and the energy to power them is staggering, with $30 to $50 billion being spent by the end of the decade solely on related U.S. energy projects. There is more than $400 billion of cash, cash equivalents, and marketable securities currently sitting on the collective balance sheets of Alphabet, Amazon, Apple, Meta, Microsoft and Oracle. Even a sliver of that could serve as a miniature Inflation Reduction Act focused on financing a wide range of emerging technologies and ultimately bring costs down for wider adoption.

So, will our short- and medium-term emissions go down because of data centers? Almost certainly not. But could global decarbonization be accelerated inadvertently as a byproduct of feeding power-hungry hyperscalers? It just might.

Revisiting the picks

Looking back on our board, here’s how the draft played out:

• Top overall draft pick: Utilities. They have a golden opportunity to build out large capital projects across the energy system. However, if they move too slowly, the large tech companies with their piles of cash could move in and figure out how to build it themselves.
• Other first round picks: Wind, solar and gas generation. The gas producers and midstreamers are no doubt enjoying a resurgence in domestic demand.
• Promising rookies: Next-generation geothermal, advanced nuclear, bioenergy plus carbon capture and storage, direct air capture and off-grid green hydrogen could all get the investment nudge they needed to start coming down the cost curve for wide scale adoption.
• Aged vets: Coal plants may see some delayed retirements, but they’re still likely destined for shutdown in the foreseeable future.
• Potential busts: Any clean fuel that requires cheap, abundant clean power faces fierce competition from data centers. Consumers will face higher power prices and increased emissions — but at least they’ve got some nifty apps and perfectly targeted ads to look forward to.

Aaron Foyer is vice president of research and analytics at Orennia. The opinions represented in this contributed article are solely those of the author, and do not reflect the views of Latitude Media or any of its staff.