Bank of America equity research analysts identified a significant alpha opportunity in overlooked artificial intelligence infrastructure stocks, according to a report issued on July 19, 2026. The analysis argues that the next wave of AI winners exists beyond the crowded semiconductor and mega-cap software trades, pinpointing companies providing critical power, cooling, and networking solutions with an average upside potential of 89%.
Context — [why this matters now]
The AI investment theme has concentrated capital into a narrow cohort of semiconductor designers and cloud infrastructure giants. This crowding created valuation dislocations exceeding those seen during the 2000 dot-com bubble for select stocks. The current macro backdrop features a Fed funds rate of 4.25% and 10-year Treasury yields stabilizing near 4.1%, providing a clearer cost of capital framework for long-duration growth projects.
A catalyst for the rotation is the physical limitation of existing data centers. AI model training and inference require exponentially more power density than traditional computing, overloading electrical grids and cooling systems. The report states global data center power consumption will reach 1,200 TWh by 2027, exceeding Japan's entire annual electricity demand. This bottleneck forces hyperscalers to invest heavily in next-generation infrastructure, creating a tangible revenue event for suppliers.
Data — [what the numbers show]
Bank of America's screening model evaluated 50 companies across the AI value chain based on projected revenue exposure growth and current valuation gaps. The top picks show a collective implied upside of 89% based on sum-of-the-parts analysis, compared to a 15% average upside for the widely held semiconductor basket.
A comparison of forward price-to-earnings ratios reveals the opportunity. The selected infrastructure providers trade at an average 18x FY2027 earnings, while the semiconductor leaders command an average 35x multiple. This disparity exists despite the infrastructure group's projected three-year compound annual growth rate of 32% versus 22% for the semis.
Revenue exposure to AI-driven demand is a key metric. The identified companies derive between 40% and 75% of future revenue from products directly servicing AI data center build-outs. This includes advanced power management systems, direct liquid cooling solutions, and high-speed fabric connectors. One stock, a cooling systems specialist, is projected to increase its AI-related revenue from $200 million in 2025 to $1.8 billion by 2028.
Analysis — [what it means for markets / sectors / tickers]
The primary second-order effect is capital rotation from mega-cap tech into mid-cap industrial and technology hardware names. This shift pressures the Nasdaq 100 index, which is market-cap weighted and overly exposed to the former winners. Exchange-traded funds tracking the industrial sector, such as XLI, and the semiconductor equipment sector, such as SMH, are direct beneficiaries of this flow.
Specific tickers highlighted for gain include Vertiv Holdings (VRT), which provides power and cooling solutions, and Astronics (ATRO), which manufactures advanced electrical systems for data centers. The report assigns a $95 price target for VRT, representing a 60% upside from current levels, and a $32 target for ATRO, representing a 120% potential gain.
The analysis acknowledges the counter-argument that a broader economic slowdown could delay capital expenditure plans from cloud providers, deferring the projected revenue. However, the report contends that AI infrastructure is a defensive capex category because it directly enables cost savings through improved computational efficiency. Institutional positioning data shows hedge funds are already increasing exposure to these names, with net long interest rising 22% over the last quarter.
Outlook — [what to watch next]
The next significant catalyst for the theme is earnings season starting July 24, 2026. Guidance updates from key players like Vertiv, NVIDIA, and Super Micro Computer will validate or contradict the demand projections for AI infrastructure components. Specifically, commentary on order backlogs and capacity expansion plans will be critical.
Key levels to watch include the relative performance ratio of the Industrial Select Sector SPDR Fund (XLI) against the Technology Select Sector SPDR Fund (XLK). A breakout above the 0.58 resistance level would confirm the sector rotation thesis. On an absolute basis, the analyzed basket must hold its 200-day moving average to maintain the bullish technical structure.
The U.S. Presidential election on November 5, 2026, represents a macro risk event that could impact infrastructure investment tax credits. Any proposed policy changes affecting data center energy efficiency standards would directly influence the long-term profitability of these companies.
Frequently Asked Questions
What does AI infrastructure mean for utility stocks?
AI data center demand is a net positive for regulated utilities with exposure to growth markets like Virginia, Texas, and Arizona. However, the capital intensity of grid upgrades may pressure near-term dividend growth rates. Analysts project utilities servicing data center hubs will see a 4-6% uplift in annual rate base growth, but this requires approval from public utility commissions which can be a slow process.
How does this AI infrastructure build-out compare to the cloud boom of 2015-2020?
The current build-out is more concentrated and capital intensive. Cloud infrastructure required broad-based leasing of existing data center space. AI infrastructure requires ground-up construction of specialized facilities with 50-100 megawatt power capacities, a scale not seen since the build-out of hyperscale cloud regions. Investment per watt of computing power is approximately 3.2x higher for AI-optimized data centers versus traditional ones.
What is the biggest risk to the AI infrastructure investment thesis?
The largest risk is a technological breakthrough in AI computing efficiency that drastically reduces power requirements. Quantum computing or new neuromorphic chip architectures could theoretically collapse energy demand per computation by an order of magnitude. While not a near-term threat, any credible research demonstrating such efficiency gains would immediately devalue traditional power and cooling infrastructure plays.
Bottom Line
Bank of America's analysis identifies a high-conviction alpha opportunity in AI infrastructure providers trading at a significant discount to semiconductor beneficiaries.
Disclaimer: This article is for informational purposes only and does not constitute investment advice. CFD trading carries high risk of capital loss.
