In an interview with The Information host Martin Peers, DataBank CEO Raul Martynek put Meta’s sweeping data center announcements in context and shared his candid assessment of where the industry boom could lead.
Martynek framesd Meta’s plans as a meaningful long-term signal rather than an immediate shift. Before the AI boom took hold around 2023, global capacity stood at roughly 35 GW, with about half of the data centers in the U.S. If hyperscalers are now talking about adding hundreds of gigawatts, that represents orders of magnitude beyond what exists today.
The bigger question, Martynek argued, is whether AI monetization can keep pace with the scale of investment. He draws a cautionary parallel to the dot-com era and the 1990s fiber overbuild, where infrastructure investment ran well ahead of the revenue needed to justify it. Supply-demand dynamics over the next three to five years bear watching.
On bottlenecks, Martynek pointsed to power access as the primary constraint. He also flagged an emerging counterforce: inference-only platforms that may require significantly less power per token than current GPU-based approaches, which could reshape demand assumptions across the industry.
Watch the full interview at The Information.
Share Article
Discover the DataBank Difference today:
Hybrid infrastructure solutions with boundless edge reach and a human touch.
Tell us about your infrastructure requirements and how to reach you, and one of team members will be in touch shortly.
"*" indicates required fields
Let us know which data center you'd like to visit and how to reach you, and one of team members will be in touch shortly.
"*" indicates required fields
"*" indicates required fields