AI Data Centers To Drive 160% Surge In Power Demand By 2030: Goldman Sachs
Energy-intensive artificial intelligence data centers are projected to drive a massive 160% increase in global power consumption by 2030, according to a new Goldman Sachs report that signals a fundamental shift in energy markets after nearly a decade of stagnant demand.
Key Takeaways
- Data center power usage expected to surge 160% by 2030
- 60% of demand growth requires new power capacity
- Natural gas dominates current supply but faces development bottlenecks
- Tech giants exploring nuclear, renewables for sustainable AI operations
Transmission Emerges as Critical Bottleneck
The report highlights that power generation represents only part of the challenge, with transmission infrastructure creating significant bottlenecks for bringing new plants online. In the United States, where most data centers currently rely on natural gas, permitting and supply chain issues have stretched development timelines to 5-7 years for new gas plants.
Energy Mix for Future Capacity
Goldman Sachs estimates the new capacity needed to meet data center demand will come from:
- 30% natural gas combined cycle gas turbines (CCGT)
- 30% natural gas peakers
- 27.5% solar power
- 12.5% wind energy
Tech Giants Adopt Multi-Pronged Energy Strategy
While natural gas remains crucial, renewables are gaining importance due to their faster deployment capabilities. Hyperscale companies are pursuing hybrid approaches—mixing power sources for immediate needs while cautiously investing in long-term solutions like nuclear energy.
Major technology firms are avoiding direct development risks through innovative strategies like forward-start power purchase agreements (PPAs). The report cites Alphabet’s recent agreement with Elementl Power to secure three sites for advanced nuclear energy as indicative of this trend toward securing sustainable, reliable power for future AI operations.
