March 2026 Dan Ives Power 30 Research Report
The AI Buildout Hits Its Physical Bottleneck, and Energy Emerges as a Key Beneficiary
Dan Ives and the Wedbush research team have launched the Dan Ives Power 30 Research Report (the “AI Power Report”), a framework of 30 stocks built around the physical infrastructure powering the AI Revolution. The March 18, 2026 report reflects Wedbush’s view that after two years of the AI conversation centering on chips and software, the defining constraint on the buildout has shifted to energy. Every dollar of AI compute spend now carries a direct and nondiscretionary power requirement that cascades through the full infrastructure stack.
The broader AI thesis remains intact and, if anything, is accelerating. Wedbush frames the roughly $2 trillion AI spending wave as comparable in scale to the interstate highway buildout of the 1950s, with the power layer an underappreciated derivative of that spend. The Ratepayer Protection Pledge, signed at the White House on March 4, 2026 by Amazon, Google, Meta, Microsoft, OpenAI, Oracle, and xAI, shifts buildout costs onto technology companies rather than ratepayers. Wedbush sees that policy commitment as accelerating the cycle rather than slowing it.
The AI Power Report
The list spans 30 names across four segments, each targeting a physical chokepoint that cannot be engineered around regardless of capital deployment.
- Power Generation & Fuel Supply: Constellation Energy (CEG), Vistra (VST), NextEra Energy (NEE), Dominion Energy (D), Southern Company (SO), NRG Energy (NRG), Talen Energy (TLN), Kinder Morgan (KMI), Williams Companies (WMB), and Bloom Energy (BE), covering nuclear baseload, natural gas generation, and fuel transmission.
- Grid Infrastructure & Data Centers: Equinix (EQIX), Digital Realty (DLR), American Tower (AMT), Quanta Services (PWR), and Willdan Group (WLDN), covering the physical buildout and facility scarcity layer.
- Equipment & Power Management: Vertiv (VRT), GE Vernova (GEV), Eaton (ETN), Schneider Electric (SBGSY), Siemens Energy (SMNEY), Johnson Controls (JCI), Hubbell (HUBB), nVent Electric (NVT), Monolithic Power Systems (MPWR), and Baker Hughes (BKR), covering the cooling transition, the transformer bottleneck, and power delivery at the chip level.
- Materials & Enabling Technologies: Cameco (CCJ), Southern Copper (SCCO), Coherent (COHR), BWX Technologies (BWXT), and NuScale Power (SMR), covering critical inputs and the next generation of nuclear.
Constituents are scored across five weighted dimensions, infrastructure exposure (30%), competitive positioning (25%), valuation (20%), growth quality (15%), and financial strength (10%).
Demand Is Outrunning Every Official Forecast
Wedbush built its demand model from the ground up, metro by metro, and it puts U.S. data center electricity consumption at 470 TWh (terawatt-hours) by 2030, roughly 10% above the International Energy Agency base case, as AI workloads demand 10 to 50 times the power density of traditional cloud infrastructure. The team also pushes back on the idea that efficiency gains will moderate demand. From 2020 to 2024, inference costs fell 95 to 98% while global queries grew from 100 million to 5 billion per day, a real-time confirmation of Jevons Paradox in which cheaper compute drives exponentially more usage rather than less.
Supply Cannot Move Fast Enough, and Baseload Power Wins
On the supply side, transmission interconnection queues now average five or more years, and large power transformers face a deficit of 600 to 800 units against lead times of roughly 2.5 years. Because renewable capacity factors of 20 to 40% fall well short of the 90%-plus baseload requirement AI workloads carry, Wedbush views nuclear as the only generation source that clears the bar on physics alone. Hyperscaler behavior reinforces that view, from the 20-year nuclear PPA (power purchase agreement) Microsoft signed with Constellation to Amazon’s Talen agreement running through 2042.
Equipment Captures Three Transitions at Once
The equipment layer carries its own thesis independent of which AI models or chips win. Data centers are moving from air to liquid cooling, from standard to high-amperage power distribution, and from legacy AC to 800 VDC architecture, and each shift drives replacement demand and margin expansion on its own. Wedbush points to direct liquid cooling earning gross margins near 25% versus 15% for air, with penetration projected to reach 65% by 2028, and to AI-optimized distribution equipment commanding 20 to 30% premiums over commodity components. In the team’s view, these product cycles remain buried inside diversified industrial financials and have yet to be reflected in valuations.
Conclusion
Wedbush’s core argument is that the AI Power Report thesis holds even if AI monetization disappoints in the near term. Multiyear hyperscaler commitments, grid modernization, and reshoring-driven power demand create structural lock-in that is not tied to any single model or chip cycle. With 2026 marking the shift from AI as theme to AI as fundamentals, Wedbush sees the companies that own the physical constraints across generation, grid, equipment, and materials as positioned to outperform regardless of which layer of the AI stack ultimately wins.
Important Information
This content draws from the March 2026 AI Power Report by Dan Ives and the Wedbush Research team. Excerpts are used under fair use for educational and informational purposes. All rights reserved to Wedbush Securities Inc.
This material is for informational purposes only and does not constitute investment advice or an offer to buy or sell any security. The views expressed are those of the authors and are subject to change. Investing involves risk, including loss of principal. Past performance is not indicative of future results.
References to specific securities are for illustrative purposes and do not constitute a recommendation.
