A growing share of AI-capable research universities are treating data center capacity as the new differentiator for faculty recruitment, graduate enrollment, sponsored research growth, and indirect cost recovery. The reporting frames compute infrastructure—kilowatt draw, cooling design, and liquid cooling readiness—as a leading factor in whether researchers can run AI workloads. As racks for AI training can draw far more power than traditional server configurations, campuses face a facilities constraint that affects timelines and capital planning. The article argues the competition is beginning to separate research institutions into distinct tiers based on physical computing environment readiness. For higher-ed leaders, the practical takeaway is that compute planning is now tied to revenue modeling and talent acquisition, meaning infrastructure decisions must be made early—before losing key investigators or graduate cohorts to better-prepared competitors.