ARK Invest Says AI Data Centers Will Thrive Amid Rising Energy Needs

Overview: ARK Invest, led by Cathie Wood, claims that AI data centers will continue to expand and remain profitable, even with growing energy demands and increasing costs.

Key Insights: A recent blog by Sam Korus, ARK’s Director of Research for Autonomous Technology & Robotics, highlights a potential solution: AI firms might operate partially off-grid by generating their own power. For instance, Elon Musk, CEO of Tesla and SpaceX, has utilized generators to run xAI’s Memphis data center, reducing reliance on the main power grid.

ARK anticipates that AI data centers will cause global electricity demand to rise by 3.2% annually, compounded through 2030. This projection contrasts with the 2.7% average growth rate in electricity production seen globally over the past five years.

Related News: Apple’s Smart Camera Expected in 2026, Aiming for ‘Tens of Millions’ in Annual Shipments, Analyst Reports.

ARK’s analysis indicates that the pace of building new electricity generation and distribution will not be a limiting factor. Currently, electricity accounts for about 9% of total costs for AI data centers, providing ample opportunity for these companies to invest in alternative power solutions without significantly impacting their operations.

The research forecasts that an extra $235 billion will be needed by 2030 to meet the increased electricity demand, which amounts to approximately 6% of the projected spending on AI hardware that year.

Importance of the Findings: The growing electricity demands of AI have sparked concern, highlighted by President-elect Donald Trump, who expressed surprise at AI’s substantial power needs in a discussion with Musk. Additionally, Jacob DeWitte, CEO of Oklo Inc., referred to AI’s power demands as “mind-blowing.” He noted that Alphabet Inc. (the parent company of Google) partnering with Kairos Power marks a significant step for nuclear startups eager to collaborate with major tech companies.

A report from April 2024 warned that without improvements in energy efficiency, AI data centers might consume up to a quarter of U.S. electricity by the end of the decade.

Next Steps:

Disclaimer: This content was partially produced with the help of Benzinga Neuro and was reviewed and published by Benzinga editors.