Property Play: Data center CEO says, 'You can't build it fast enough'

Watch on YouTube ↗  |  January 13, 2026 at 13:23  |  23:07  |  CNBC
Speakers
Andy Power — President and CEO, Digital Realty
Diana Olick — CNBC Real Estate Correspondent

Summary

CNBC's Diana Olick interviews Andy Power, president and CEO of Digital Realty, about whether AI and data centers are in a bubble. Power argues data center demand is outpacing supply, customers are signing long-term 15-year leases, and efficiency gains from AI chips will expand rather than shrink infrastructure demand. He says Digital Realty is insulated by its diversified shared-infrastructure model and sees power/transmission as a near-term constraint, while downplaying space-based data centers as a scalable replacement.

  • Diana Olick interviews Digital Realty CEO Andy Power.
  • Power rejects the idea that data centers are in an AI bubble.
  • He says demand exceeds supply and leasing outpaces construction.
  • He cites Moore's Law and Jevons paradox to argue efficiency expands data center demand.
  • Digital Realty's shared-infrastructure model serves 5,000+ customers with long-term leases.
  • He downplays hyperscaler credit concerns, citing non-AI businesses.
  • Power says power/transmission is a near-term bottleneck; water use is managed.
  • He is skeptical space-based data centers can replace Earth infrastructure at scale.
Ideas
Andy Power President and CEO, Digital Realty 1:52
Digital Realty insulated by long-term demand.
Digital Realty is insulated from an AI bubble because its data centers are foundational hard assets in locations with barriers to new supply, demand is outpacing supply, and high-credit-quality customers are signing long-term 15-year leases with urgency. Its shared-infrastructure model serves over 5,000 customers across 50 metropolitan areas on six continents, so it benefits if technology broadly wins rather than depending on one hyperscaler or AI winner.
Andy Power President and CEO, Digital Realty 2:03
Efficiency gains expand, not shrink, data centers.
Data center demand should keep growing despite efficiency improvements in AI chips because Moore's Law and Jevons' paradox mean cheaper, more efficient compute creates new use cases and more infrastructure demand. He says leasing is outpacing the ability to construct, vacancies are at historic lows, and speculative builds cannot keep up with real customer demand.
Up Next

This CNBC video, published January 13, 2026, features Andy Power discussing DLR, DTCR. 2 trade ideas extracted by AI with direction and confidence scoring.

Speakers: Andy Power  · Tickers: DLR, DTCR