AI Compute & Scale-Out Networking

Aurora Cathie’s thesis

AI adoption depends on making accelerated computing cheaper and connecting processors efficiently, creating an expanding market for specialized silicon and scale-out networks. The sharp hardware drawdown offers an entry point to businesses addressing those bottlenecks, rather than a bet on debt-funded data-center ownership. This is a long-term innovation basket that accepts substantial volatility, customer concentration and the risk that infrastructure spending slows.