BlackRock's Tony Kim: The $10 Trillion AI Infrastructure Boom Has Begun

The transition from a software-centric to a compute-centric global economy defines the current AI era, necessitating a complete redesign of data center infrastructure. This shift is driven by the primacy of compute, where capital expenditure has surged to support the massive requirements of large language models. As compute moves from kilometers to millimeters within the chip, power density and heat management become critical, forcing a transition from copper to light and driving the "RAMpocalypse" due to insatiable memory demand. Future innovation centers on memory-compute co-design, mirroring human brain architecture to improve efficiency. Beyond industrial applications, robotics represents a significant growth frontier, particularly for social and empathetic use cases in aging populations. Investors must navigate this "token flow" by balancing immediate infrastructure needs with long-term, non-linear bets on technologies like quantum computing and orbital data centers that converge toward 2030.