Why data centers are driving investment in alternative energy
Key takeaways
Data centers are driving a surge in electricity demand, creating new investment opportunities while increasing the need for reliable, low-carbon energy.
- Global data center power demand is projected to grow 17% annually between 2022 and 2030, with electricity consumption expected to increase from 485 TWh in 2025 to 950 TWh by 2030.
- Grid constraints, long connection times, and reliability requirements are accelerating investment in clean energy, battery storage, fuel cells, and other alternative power solutions.
- Investment opportunities extend beyond data centers to transmission infrastructure, power equipment, engineering services, and emerging energy technologies.
Data centers sit at the heart of the artificial intelligence (AI) revolution, and their power needs are surging. Clean energy is the goal, but it comes with complexity.
As AI adoption accelerates, data centers are emerging as an important driver of a structural electricity demand super-cycle.
McKinsey projects 17% annual growth in data center power demand globally between 2022 and 2030. The IEA estimates that total data center consumption will double from 485 TWh in 2025 to 950 TWh by 2030, accounting for up to 50% of demand growth in some countries. (See Figure 1.)
The implications for the global energy transition will depend on how fast demand increases, and on the availability of low-carbon electricity.
Both are key questions for investors looking to understand the long-term outlook. However, projections are difficult to make with any certainty.
“Most of our clients are pretty confident about growth in the next three to five years, but further out the demand models are less predictable,” said Reid Tucker, Principal at KPMG. “There are other disruptors in the market, such as chip density or different data center technologies that could change the market.”
“Most data centers previously used video GPUs,” said Zap Cheng, CFA, Co-Founder of Hong Kong-based Ecovision, which helps clients produce disclosure reports, alongside carbon, energy and building audits. “Now they are often using application-specific integrated circuits (ASICs), which are best for stable, repetitive tasks.”
Unlike general purpose GPUs, ASICs are specifically designed for AI and cryptography work.
At the same time, power demand could get a further boost from other emerging technologies, such as robotics, with new players lining up to fund US data center campuses. “In the last six months I’ve been in discussions with investors that I have never seen in this market,” said Tucker.
The push for clean power
The biggest companies behind the data center build-out are intensely focused on clean energy. In 2019, Amazon co-founded The Climate Pledge with a commitment to achieve net zero by 2040. In 2020, Microsoft pledged to be carbon negative by 2030, and to remove historic emissions by 2050. Google also announced in 2020 that it would reach net zero in 2030, as did Meta.
Data center operators also need to be able to trust that power will be available. “If you are operating a data center, you need reliable power – you need it to be 24/7,” said Cheng. “If you look at the recent conflict in the Middle East, it reminds you of the risk of relying purely on fossil fuels. And then if you are using solar or wind, you have the issue of intermittency. Data centers may need technology to capture and store energy to ensure their consistency of supply.
Securing more clean power, however, is not easy.
“Power is the starting place,” said Matt Sallee, CFA, Senior Portfolio Manager and Head of Investments at Tortoise Capital. “In the early days of AI, everybody was focused on chips. Now it’s power, and in the near-term, it is very challenging to get that power.”
Virtual power purchase agreements (PPAs) have become a popular way for technology firms to invest in clean energy.
“They will invest in and pay to develop a very large wind or solar farm wherever it makes sense geographically to have it, and then they commit to buy the power through a PPA,” said Sallee. “It’s not directly serving their data center, but they are bringing clean power into the grid.”
But the grid itself is often the problem, given the speed at which the data center sector is expanding. Traditional grids are presenting hyperscalers with severe bottlenecks, with long wait times for grid connections.
“In the US, it’s three to five years to connect to the grid, depending on which part of the country you are in,” said Sallee. “I think that’s the number one issue here.”
As a result, data center developers are increasingly looking to build their own power plants.
“In the past, companies would rely on purchasing offsets or investing indirectly in clean energy, rather than running the data center full-time on clean energy,” said Tucker. “That’s the shift now, and in order to do that, they have to deploy those clean energy assets directly connected or very close to the data centers.”
Looking to alternatives
There are multiple hurdles to actually getting power into data centers.
“Anyone developing a data center wants to be connected to the grid- that’s the top priority here, because you have to have redundancy,” said Tucker. “The process to go through that, the interconnections, the grid constraints, those are the key bottlenecks to having your preferred energy supply right to your data center.”
New gas turbines have lead times of two to four years in the US, according to Sallee, depending on whether a customer wants a highly efficient Combined-Cycle Gas Turbine (CCGT) or a quicker, less efficient, behind-the-meter solution (in other words, connected directly to a customer rather than to the grid) using a simple cycle or reciprocating engine.
“Right now, the quickest and cheapest thing to build in the US is solar,” said Maurice Berns, Senior Partner and Leader of Energy & Climate at the BCG Institute. “If you go to a turbine manufacturer, they will probably tell you to wait five years and pay USD2,500 per kilowatt, up from about USD1,300 two years ago.”
This has opened up opportunities for companies that can offer an alternative. Bloom Energy, for example, can deploy onsite solid oxide fuel cells to data centers within two months. It has an agreement with Oracle to supply up to 2.8 GW of fuel cells.
“A solution like this isn’t 100% clean, but it is very low emission and is becoming cost-competitive,” said Sallee. “It’s scalable and has a big near-term opportunity.”
In the long term, most are planning to use nuclear power, particularly the emerging generation of Small Modular Reactors (SMRs). Retired reactors are also being brought back into service – the US is working to revive three sites and is considering others.
“SMRs are the holy grail, but they are five years out at best,” said Sallee. “We have a couple of investments in SMR companies, but they are obviously still speculative at this point.”
Another factor is having sufficient qualified labor for these projects. “It’s manufacturing and the technology that is used to do that, the welding techniques that have to change,” said Tucker. “There are all sorts of basic things that have to be developed to make new nuclear technologies commercially viable at scale.”
Data centers also have some fearsomely strict requirements. Chief among those is consistency and reliability. “Generally they need ‘five nines’ reliability,” said Sallee, referring to the common metric of 99.999% uptime. “That’s part of the issue with renewables for data centers, because of their intermittency.”
But within that, data centers in fact experience considerable variability in exact demand, as they process AI prompts that require more or less energy to satisfy. To balance that need smoothly is where battery storage plays a big role.
Berns said that data centers have more flex than is often realized. “They say they need 100% firm power 100% of the time,” he said. “But they don’t. They’ve got various tranches of demand, and there are customers who are happy to wait 10 seconds instead of one quarter of a millisecond.”
Investing in what’s adjacent
For Sallee, some of the most interesting ways to find investment opportunities in the data center industry are in areas that are adjacent to it. Tortoise Capital itself has a TCAI exchange-traded fund (ETF) that tracks much of this world.
The ecosystem includes sectors such as electrical transmission and distribution infrastructure construction, where the industry leader is Quanta Services, a Fortune 500 company with a market cap of over USD100 billion and which is heavily involved in projects to upgrade grids to cope with the demands of AI.
Others include generation equipment manufacturers, transformers and switch gear makers, or gas engine companies like Innio, which is backed by private equity investor Advent International as well as the Abu Dhabi Investment Authority. Innio held its IPO on Nasdaq in June 2026, valuing the company at about USD25 billion.
“With any investment, the question is what is your timeline for returns,” said Tucker. “For the near-term priorities, you are looking at engineering contractors and specialty trades.”
In the longer term, it depends what kind of investor you are. “Those with the deeper pockets are tending to be the ones investing in the SMRs, or the geothermal technology, or fuel cell batteries,” added Tucker. “The hyperscalers are certainly buying up and investing in the supply chain across everything at the moment.”
To learn more about energy transition investment trends, explore these stories:
How new tech is driving new investment in nuclear power
Grid and battery investment: Why it matters to the energy transition
Will AI help or hinder the energy transition?
What's driving the boom in grid-scale batteries?
You may also be interested in
11 Aug 2026
Will AI help or hinder the energy transition?
Explore our programs and certificates
CFA Institute offers a diverse range of programs and certificates designed to meet the needs of finance professionals across various career stages and specializations.
