The Shift to Infrastructure
Financial professionals are increasingly viewing semiconductor and infrastructure stocks as the most reliable way to capitalize on the expanding artificial intelligence market. While consumer attention and market sentiment frequently shift between competing AI agent products, the underlying demand for computing power remains consistent.
Paisley Nardini, head of investment strategy at Tema ETFs, described the current leadership in the AI agent space as a "game of hot potato." She noted that while Meta Platforms Inc. has recently gained momentum, investors are already anticipating a shift in focus within weeks. This volatility in application-level leadership has prompted many to look further down the supply chain.
Why Chips Are the Safer Bet
Rick Lear, chief investment officer at Lear Investment Management, explained that rather than predicting which company will build the best AI agent, his firm focuses on the foundational elements required for all agents to function. This includes chips, memory, and power infrastructure.
"Tech leadership keeps going back and forth, but if agents take off overall, we will need a lot more chips, more memory, more power, more infrastructure," Lear said. His portfolio includes holdings in Nvidia Corp., Micron Technology Inc., and Williams Cos Inc., reflecting this infrastructure-first strategy.
AI agents, which handle complex multi-step tasks such as managing finances or making reservations, require significantly more computational resources than previous generation chatbots. This structural shift ensures that demand for hardware will persist even as software preferences change.
Market Performance and Revisions
The market has responded strongly to this thesis. The Philadelphia Stock Exchange Semiconductor Index has jumped 81% this year, fueled by strong performance from companies like Micron, Marvell Technology Inc., Intel Corp., and Advanced Micro Devices Inc. Nine of the top 10 gainers in the S&P 500 Index currently have exposure to AI infrastructure themes.
Analysts have also raised their expectations for the sector. According to Bloomberg Intelligence, the AI infrastructure sector is now expected to post net income growth of 63% in 2027, driven by a 54% increase in revenue. This represents a substantial upward revision from estimates made at the end of July, which projected profit growth of less than 48% on a 32% revenue rise.
The Volatility of AI Leaders
The recent surge in Meta’s stock highlights the unpredictable nature of the AI application market. Following the debut of its Muse personal AI assistant last month, Meta’s shares rose 27% in September, its best monthly performance in nearly four years. However, earlier in the year, the company faced scrutiny over its massive AI spending with limited visible returns.
Similarly, Alphabet Inc. has seen its status as an AI leader fluctuate, moving from early-year praise to recent scrutiny. Other players, including OpenAI with its Dots product and SpaceXAI’s Grok Bot, are also competing for market share. This constant rotation makes betting on a single application winner risky.
Long-Term Outlook
Despite the short-term churn in AI product popularity, the long-term outlook for hardware remains robust. Citigroup analyst Atif Malik estimates that the total addressable market for CPU chips will grow at a 60% compound annual rate through 2030. Different types of semiconductors, including processors, memory, and networking chips, are all expected to benefit from the development of more sophisticated AI agents.
As Nardini noted, while leadership in the AI agent space may change in six months, the fundamental demand for the chips that power these systems is unlikely to diminish.