Morgan Stanley is telling investors that the AI infrastructure buildout has entered a new phase, one where the bottleneck is no longer just graphics processing units (GPUs) In a new episode of the bank’s Thoughts on the Market podcast, Shawn Kim, Morgan Stanley’s head of Europe and Asia Technology Research, says the AI use case is transitioning from chatbots to agents, shifting hardware requirements from GPUs to central processing units (CPUs). According to Kim, CPUs allow AI agents to remember instructions, understand user preferences, work across digital tools, plan workflows and adapt as circumstances change. “Agentic AI also depends on three stacks: the brain, or the large language model, orchestration, where the CPU manages the doing, and knowledge, which is memory. Memory may be the most important layer. An agent that knows your preferences doesn’t tone, and task history becomes more useful over time. That creates a context flywheel.…