Semiconductor stocks have experienced notable volatility in recent weeks, with the Philadelphia Semiconductor Index underperforming global equities (MSCI AC World) over recent trading sessions.
This volatility has been caused by investor concerns about future capital expenditure (capex) trends and the sustainability of AI-driven demand. Media reports that Meta Platforms may launch a cloud infrastructure business, selling access to AI computing power and models, sparked the latest wave of worries.
More broadly, mega-cap tech companies have seen share price declines, raising questions about whether they will maintain their aggressive AI capex buildout. Concerns are growing over whether firms will slow capex by using lower-cost, open-source AI models, or lease out excess compute capacity to external customers to improve cash flows.
However, CIO remains constructive on the wider AI value chain, and we see roughly 10% further potential gains over the next 12 months, for three main reasons:
1. Strong and growing demand for AI workloads. We believe the medium-term demand story for AI remains intact. Advances in lower-cost models may inject short-term uncertainty, but lower token costs will ultimately drive greater adoption and demand for agentic AI workloads, in our view. Cloud revenue growth at major platforms is accelerating, with 40% growth reported in the first quarter of 2026, and this trend is expected to continue through the rest of the year. As more businesses adopt AI and embed it in how they work, demand visibility, pricing power, and earnings momentum for the “picks and shovels” of the AI buildout should remain strong.
2. Supply bottlenecks remain. Supply constraints in key segments of the AI hardware supply chain have yet to ease. This supports pricing power and earnings for semiconductor companies, even as questions arise about capex sustainability. Better pricing can also underpin monetization prospects, as AI workloads are expected to accelerate further, especially as cloud platforms ramp up their offerings and as regulatory clarity improves.
3. Diversification rather than divesting makes more sense. We favor diversifying both within the AI complex and across other structural themes. Investors seeking more defensive exposure can consider data center operators and select payment companies, as they may offer resilience amid volatility. Our broader order of preference prioritizes the “picks and shovels”—semiconductor equipment due to shortages, processor providers, and firms manufacturing memory components. These companies benefit from strong demand visibility, pricing power, and earnings momentum.
Beyond AI, CIO highlights the Power and resources and Longevity ideas as areas for capital allocation. These themes stand to benefit from structural trends like the energy transition, resource scarcity, and demographic shifts.
How to invest?
The purest way to access our top picks is through vehicles following CIO’s AI selection, including those with capital preservation features that can limit potential losses in case of a correction.
Investors can also consider building block approaches as part of a wider, diversified asset allocation. Other options include actively managed approaches that invest in fewer stocks, look beyond the largest companies, and take larger, high-conviction positions.
We also see potential ideas in private markets. We believe value may start spreading beyond frontier AI labs and semiconductors. Vehicles invested in private companies operating in resilient software-as-a-service businesses, AI infrastructure, data centers, and energy may benefit, while generic legacy software remains vulnerable. Selectivity, underwriting discipline, and operational execution will be increasingly important amid rising dispersion across sectors, business models, and asset classes.
