The world is shifting to meet the growing demand for computing power. In late October, KKR and Energy Capital Partners announced a $50 billion strategic partnership to invest in data-center and power-generation projects to support advancements in artificial intelligence. And in September, BlackRock and Microsoft announced a partnership to mobilize up to $100 billion in private equity capital and debt financing to invest in data centers and the energy infrastructure needed to power these facilities, with Microsoft announcing a deal to reopen a nuclear power plant to power its data centers shortly after. This follows a number of other joint ventures, large fundraises and debt financings in the data center space as institutional investors are taking notice and recent interest rate cuts are providing an additional tailwind.
These firms are right to recognize that investment in new data centers also requires investment in new energy infrastructure. As the AI era takes flight, the question remains: will AI and the growth of data centers cause us to run out of power? Some forecasts show data centers will represent 7.5% of all power demand by 2030, the amount of power used by a third of U.S. households.
This seemingly endless need for power poses several reliability challenges. Growing demand poses physical challenges to the grid infrastructure we have in place, and building more on our existing grid increases the risk of higher rates to customers. Additionally, regulatory and utility planning are often decentralized, inefficient, and built for the infrastructure we’re trying to move away from, not what we’re moving toward.
Overarching all of these challenges is a bigger one. To solve insatiable energy demands in the long term, we need to go beyond investing in simply new energy infrastructure and also invest in clean energy infrastructure to ensure this data center growth is sustainable — as we’ve seen with Microsoft’s latest power purchase deal. While some data center operators are already using renewable energy, not enough clean energy capacity exists to cover existing demand. And increasing demand can translate into a sharp increase in carbon emissions. Indeed, Microsoft announced in May that its 2023 carbon emissions increased by nearly 30% over 2020 levels.
Recognizing the emissions impact of their data centers, many major tech companies including Amazon, Apple, Google, Meta, and Microsoft have entered into long-term contracts to purchase renewable energy. These five companies have committed to having 100% renewable energy for their data centers, with contracts that represent over half of the total global corporate renewables market.
This shift in the market has created a major opportunity to revisit our fundamental grid architecture and invest in new infrastructure that is better designed to meet the needs of the future, including solutions that can deliver clean power directly to data center sites. The current electric grid architecture is still in the mainframe world, with large power plants pushing electricity in one direction to customers often hundreds of miles away. It is expensive and energy inefficient. Rather than build more gas power plants or wait for transmission lines to bring renewable power from long distances, new demand from data centers could help incentivize more clean power development closer to where the power is needed. Owners of data centers have already shown that they want to develop onsite renewable energy systems that adjust power output to match the site's electricity demand in real (or near real) time, using a combination of fuel cells, solar, storage and geothermal.
Developing these capabilities and additional innovative renewable energy solutions will require significant upfront investment but will make the grid more energy and financially efficient, including by reducing costs associated with transporting energy. In a more distributed system, data centers will act as anchors to other customers in the area. Since the power demands of the data center vary, there may be times when the data center can provide power to other customers and times when shifting demand from other customers could help offset the power requirements of the data center.
So what’s stopping progress? Renewable energy solutions have often been viewed as a riskier bet, with substantial checks required and little opportunities for short-term returns. And although there is growing confidence in clean energy as an asset class, there are still barriers to entry — both real and perceived — for large institutional investors from a risk and check size perspective. A greater mindset shift is needed, but there are reasons to be hopeful.
First, there seems to be growing recognition across the industry and investors of the immediate climate emergency we are facing and the need to invest across the value chain for digital infrastructure, including renewable power infrastructure. Second, the opportunity for risk-adjusted returns are high as we have a growing class of knowledgeable investors who are able to both fund infrastructure companies and projects as well as provide the nuanced advice and operational support for project development initiatives.
Form follows function. The rise of data centers stem from how IT infrastructure itself has evolved from a mainframe-only world of 40 years ago to a network of distributed devices (smart phones, PCs) connected to the mainframe and server “cloud.” Let’s use the increasing demand from data centers for computing power — and growing consensus around the need to also invest in the energy sources behind them — to finally invest in and build the much discussed “Grid of the Future” today while delivering strong, long-term returns to institutional investors.
Richard Kauffman is the chair of Generate Capital. He is based in New York. This content represents the views of the author. It was submitted and edited under Pensions & Investments guidelines but is not a product of P&I’s editorial team.