The Infrastructure Investor Podcast KKR: Parts of AI market running ‘a bit hot’ in one of tech’s fastest capex cycles
Dec 4, 2025
Waldemar Szlezak, KKR's Global Head of Digital Infrastructure, brings deep expertise in data centers and AI-related investments. He discusses the real demand for AI infrastructure versus potential market bubbles, highlighting overheating risks during a fast capex cycle. Szlezak explains the lifespan of data centers compared to rapidly aging GPUs, the concept of 'coordination tax,' and KKR's integrated strategies. He also shares insights on construction efficiencies and the industry's challenges, from regulatory hurdles to evolving hyperscaler needs.
AI Snips
Chapters
Transcript
Episode notes
AI Demand Is Real, But Parts Run Hot
- AI demand is structural and long-term, not a pure bubble, but some market segments are overheating due to an exceptionally fast CAPEX cycle.
- Investors should ask where enthusiasm outpaces discipline rather than ask if AI is a bubble.
AGI Timing Should Drive Build Strategy
- The timing to AGI determines how aggressively to build infrastructure today and whether flexibility matters more than raw scale.
- If AGI is years away, design for adaptable shells and systems; if it's imminent, current build rates may be insufficient.
Core Hubs Tight; Remote Giga Builds Risky
- Core cloud hubs (e.g., Northern Virginia, Dallas, London) are capacity-constrained with low vacancies and long queues, so those markets are not overheated.
- Overheating appears in remote giga-scale training builds where residual value and latency risks are higher.