AI is certainly having its moment – but there’s a downside. The rapid advancement and integration of artificial intelligence has led to a big spike in electric power demand. AI, especially at a larger scale, requires substantial computational resources, which in turn necessitate more data center capacity. Pair this with a broader electrification trend (think electric vehicles, greening buildings, etc.) and now we have a power demand surge, one that outweighs supply. So where does that leave us? And what technologies do we have to fill the gap?
By 2030, AI will draw as much power as an entire country
The relationship between AI and power demand is undeniable. By 2030, AI could account for 3% to 4% of total global power demand. That’s enough consumption to make AI’s electricity use equal to that of countries like Russia or Japan. Largely as a result, power demand from data centers is expected to increase dramatically. In the U.S., that increase will be 80% between 2023 and 2030, reaching about 35 GW total. This is an enormous increase in electricity demand, with no plan yet in place to make up the difference.
$600 billion for digital infrastructure
Rising power demand has led to a series of investments in digital infrastructure. By 2035, investments are expected to reach over $600 billion, roughly double of what has currently been invested. The expanding market is largely driven by the rising demand for data centers. Take I Squared Capital‘s recent launch of a $2.15 billion digital infrastructure business as an example. Or investment firms like IPI Partners or DigitalBridge, both of which are singularly focused on funding digital infrastructure projects. Momentum for digital infrastructure investments is here and growing.
Batteries and nuclear could save the day
An increase in data center capacity requires an increase in readily available energy. Who will be the big winners that ensure the power keeps flowing? I have my bets on batteries and nuclear. Batteries facilitate the storage of renewable energy, while nuclear offers a large-scale, dispatchable, and low-emission power source. Collectively, these solutions could support the digital infrastructure critical for AI operations and broader electrification trends.
On the battery front, companies like ZincFive, Inc., headed by CEO Tim Hysell, are making huge strides with their battery storage systems tailored for the demands of data centers. Big data centers require high capacity, reliable, and safe energy storage systems.
When it comes to nuclear, companies are also taking notice. Microsoft is exploring Small Modular Reactors (SMR) to fuel its energy-intensive AI goals. These models are cheaper and easier to build, making them an attractive option compared to traditional nuclear (which is often late and over budget).
Keeping both AI and carbon emissions on track
The combo of AI-driven power demand and broader electrification is reshaping the electricity landscape. Investments are steering towards innovative power solutions and digital infrastructure to support rising demand. With the right tools, we can keep carbon emissions low and still sustain the digital momentum fueled by AI.