Andrew Obin on AI's Impact on Data Center Investment and Industrial Partnerships

AI-Driven Data Center Investment

• 💡 The current massive technology change is driven by companies like Nvidia, who are building a moat against competitors.
• 🚀 The core thesis is that major industrial players in HVAC and electrical sectors are partnering with data centers to facilitate this technological evolution.
• ⚡ The rapid pace of technological change, from GBX racks to Kyber Racks and new chip generations (Blackwell to Rubin to Fainmanship), necessitates significant upgrades to data center infrastructure.

Navigating Technological Noise

• 🎯 Data centers prioritize partners who can design, install, commission, and service their equipment, which is a key competitive advantage.
• ⚠️ While new chip technology might allow water cooling at higher temperatures (e.g., 45°C), the practicalities of cooling data centers in hot climates like Texas still require large water-cooled chillers.
• 📈 Although there might be less chiller content per megawatt in some iterations, this is offset by the need for more equipment on rooftops, creating overall complexity and volatility.

Long-Term Partnerships and Service Revenue

• 🗓️ Companies like Train are already sold out of water-cooled chillers for the next two years, indicating strong demand that may not immediately reflect in earnings.
• 🤝 Major players like JCI and Train maintain close relationships with hyperscalers (Meta, AWS, Alphabet), who have visibility into future technological roadmaps.
• ⏳ The service component of equipment installed today will become a significant revenue stream towards the end of the decade, offering more consistent income.
• 🚀 Nvidia provided a roadmap to hyperscalers 12-18 months ago, meaning current orders are for infrastructure to be built out in 2026-2027.