DigitalBridge: Industrialisation of AI will need 500GW of compute over next decade

Digital infrastructure development is shifting toward physical asset optimization as AI demand accelerates. Marc Ganzi, CEO of DigitalBridge, emphasizes that long-term value lies in controlling power and cooling rather than the rapidly evolving GPU stack. By implementing microgrids and sophisticated cooling systems, operators can achieve significant power arbitrage and operational efficiency, effectively insulating themselves from grid constraints. The industry is currently transitioning from large language model training to the industrialization of AI, with robotics and autonomous systems emerging as the next major growth drivers. Success in this sector requires rigorous contract discipline and deep technical expertise, as the divide between experienced operators and new market entrants becomes increasingly apparent in the face of widespread project delays and complex delivery requirements. These infrastructure-first strategies are essential for meeting the massive compute demands of future AI factories.