For nearly two years, the artificial intelligence boom has largely centred on one trade: graphics processing units (GPUs). That made Nvidia one of the clearest winners of the AI race, as demand for chips surged across data centres, cloud providers, and large model developers.
Now, investor Cathie Wood says Wall Street may be overlooking another part of the AI buildout. Speaking about the next phase of artificial intelligence infrastructure, Wood pointed to CPUs and legacy technology companies that may benefit as AI workloads evolve beyond training models. Her argument comes as AI begins moving into inference, automation, and agentic systems.