Skeptical investors questioning the return on the historic levels of capital expenditures chasing AI leadership positions triggered a sharp downturn in the data-center, semiconductor, and hardware trade in July. Many equities were effectively cut in half. The selloff accelerated when a multi-billion-dollar hedge fund — heavily levered long hardware and short software — blew up and was forced by its prime brokers to liquidate margined positions, deepening the drawdown.
Yet as earnings season unfolded, the narrative flipped. Leading hyperscalers reported exceptional demand for their AI and cloud offerings, prompting a rapid recovery in the very same equities that had been punished weeks earlier. Microsoft and Amazon both cited accelerating demand for their cloud platforms, with revenue growth now scaling faster than capital expenditures — a key inflection point. Management teams suggested margin improvement was imminent. These statements not only propelled their stocks higher but restored confidence across the entire AI trade, reinforcing the notion that meaningful returns on investment are approaching.
So where do we go from here? Investor schizophrenia notwithstanding, a clear-eyed assessment suggests the following: hardware stabilizes, infrastructure buildout continues at full speed, enterprise adoption enters the early trial phase, select software names begin to work, new AI-native companies start to participate, and agentic enterprise adoption emerges in late 2026 into 2027. Taken together, these dynamics point to a broad continuation of the bull market. Extreme volatility will persist, and macro events and geopolitical risks will periodically take center stage — but the secular trend remains intact.
Inflation fears will, in the short run, dictate interest-rate direction. Ironically, a rise in short-term rates could flatten the yield curve and push longer-term rates lower, as bond vigilantes conclude the Fed is serious about fighting inflation. Equally ironic, the inflationary spiral associated with rising semiconductor unit costs may be ending as supply and demand rebalance — paving the way for the broader deflationary cycle long promised by AI-driven productivity gains.
Investors will have a great deal to sort through. But disciplined stock selection — focused on companies with real adoption curves, durable demand, and credible monetization pathways — should lead to outsized gains.
Michael P. DeSantis
Managing Member
Bruce M. Lupatkin
Managing Member
