Λ

Λ3THER RESEARCH BRIEF

The New Oil: AI's Unquenchable Thirst for Power

May 27, 2026

The digital AI revolution is creating a physical-world bottleneck of staggering proportions, shifting the entire investment landscape from software and chips to the raw materials of computation: electricity and the infrastructure that delivers it.

EXECUTIVE THESIS // TACTICAL VIEW

The exponential growth of AI is no longer a story about algorithms; it's a story about energy. The market is awakening to the reality that data center power demand is set to double or triple by 2030, a surge the aging U.S. grid cannot handle. This creates a structural supercycle for the "picks and shovels" of the AI buildout: baseload power generators, grid equipment manufacturers, and on-site power specialists. The key trigger is the multi-year lead times for critical hardware like transformers, forcing hyperscalers into long-term, high-priced energy contracts that fundamentally revalue the utility and industrial sectors.

1. The Macro View: A Physical-World Bottleneck

👋 1 big thing: The AI industry's demand for electricity is growing at a pace that makes Moore's Law look quaint. This isn't an incremental increase; it's a step-function change that is breaking forecasting models and straining physical infrastructure to its limits.

Why it matters: Global data center electricity consumption is projected to more than double from 460 TWh in 2022 to around 1,050 TWh by 2026. In the U.S. alone, data centers could consume up to 12% of the nation's total electricity by 2028, up from just 4.4% in 2023. This explosive growth is creating a desperate scramble for power that renewables, with their intermittent nature, cannot satisfy alone.

Zoom in: AI workloads require 24/7, high-density, reliable "baseload" power. This has triggered a renaissance for nuclear and natural gas, the only sources that can provide this level of stability at scale. Hyperscalers like Meta and Microsoft are now signing massive, multi-decade Power Purchase Agreements (PPAs) directly with nuclear operators, effectively bypassing the grid to secure their energy supply.

This is the market realizing that the most valuable asset in the AI race isn't just the GPU, but the guaranteed megawatt that powers it.

2. The Anatomy of the AI Power Stack

The bottom line: An AI data center is a marvel of engineering, but it's fundamentally a power conversion and heat management problem. The supply chain to solve this problem is now the primary constraint on AI's growth.

Building a gigawatt-scale campus requires a vertical stack of specialized hardware, and every layer is experiencing unprecedented demand and multi-year backlogs.

This isn't a temporary surge. Eaton estimates there is a backlog of 165 to 228 gigawatts of data center capacity planned in the U.S. through 2030, representing what could be over a decade of construction at current installation rates.

3. Equity Analysis: The AI Bottleneck Portfolio

Go deeper: The most durable investment theme of the AI revolution may not be the model-builders, but the companies solving the physical constraints of power and infrastructure. Here's a look at the key players.

🏦 The Power Generators: Vistra (VST) & Constellation (CEG)

These companies own the largest nuclear fleets in the U.S., making them the primary beneficiaries of the demand for 24/7 carbon-free energy. Their long-term PPAs with tech giants de-risk their business models and provide decades of visible, high-margin cash flow. Constellation is the largest nuclear producer, while Vistra is aggressively leveraging its nuclear and gas portfolio to sign deals with Meta and Amazon.

🛠️ The Grid & Equipment Builders: Quanta (PWR), GE Vernova (GEV), Eaton (ETN)

This trio forms the backbone of the grid buildout. Quanta Services is the premier contractor, reporting a record backlog of $48.5 billion in early 2026. GE Vernova and Eaton are essential equipment suppliers. GEV is a key manufacturer of the large power transformers facing severe shortages, while Eaton dominates the on-site power distribution and cooling markets. Eaton's data center orders are exploding, and GEV's electrification segment orders nearly doubled year-over-year, driven by data center demand.

⚡ On-Site Power Solutions: Bloom Energy (BE)

With grid connection queues stretching for years, data centers are increasingly turning to on-site power generation to get online faster. Bloom Energy's solid-oxide fuel cells, which run on natural gas, offer a rapidly deployable solution. The company has seen a surge in demand, securing major contracts with companies like Oracle and utility AEP to bypass grid delays. Bloom is on track to double its production capacity by the end of 2026 to meet this demand.

☁️ The GPU Cloud Operator: CoreWeave (CRWV)

CoreWeave, which went public in March 2025, represents the demand side of this equation. As a specialized AI cloud provider, its entire business model relies on securing massive clusters of GPUs and, critically, the power to run them. The company has pioneered using Nvidia GPUs as collateral to raise tens of billions in debt to finance its expansion. While not a power producer, CoreWeave's massive financing and backlog—$99.4 billion at the end of Q1 2026—serves as a powerful leading indicator of the immense and growing demand for the power and infrastructure provided by the other companies in this ecosystem.

A3THER Research is an independent financial publisher. All reports are for educational purposes and involve risk. Please review our full Regulatory Disclosures & Risk Warnings.

REGULATORY DISCLOSURE

A3THER Research LLC (“A3THER”) is a financial publisher and research provider. A3THER is not registered as an investment adviser, commodity trading advisor, broker-dealer, or financial planner with the U.S. Securities and Exchange Commission (SEC), the Financial Industry Regulatory Authority (FINRA), or any state or foreign regulatory authority. No content published by A3THER constitutes a recommendation that any particular security, portfolio of securities, transaction, or investment strategy is suitable for any specific person. The information is not tailored to individual circumstances and does not take into account the unique financial situation, investment objectives, or risk tolerance of any recipient. Nothing contained herein establishes a fiduciary, advisory, or client relationship between the reader and A3THER.

INVESTMENT RISK WARNING

All financial markets involve substantial risk, including the complete loss of principal. Highly volatile asset classes—including equities, sovereign bonds, commodities, foreign exchange, and digital assets—can experience sharp fluctuations. The research, opinions, analyses, projections, and reports published by A3THER are for general educational, historical, and informational purposes only. Readers must conduct their own independent due diligence and consult with a licensed financial adviser, tax professional, and legal counsel before making any investment decisions. Under no circumstances shall A3THER, its members, officers, or contributors be liable for any direct, indirect, incidental, special, or consequential trading losses or damages incurred as a result of relying on any data, models, charts, analysis, or strategic reviews published herein.

PROPRIETARY DATA & PROJECTIONS

Historical calculations, dual-axis charts, correlation matrices, and yield projections are synthesized using public APIs, scraping engines, and proprietary quantitative models. All charts and historical figures are for illustrative and conceptual purposes only. Hypothetical or backtested performance has inherent limitations, is not reflective of actual trading, and does not guarantee future results. Information is provided on an "as is" basis, without express or implied warranties of any kind regarding accuracy, completeness, timeliness, or reliability.

PRIVACY POLICY

At A3THER Research, we prioritize the privacy and security of our institutional clients and subscribers. We collect minimal personal data, restricted to registration emails and session preferences required to operate our analytical platform. Your email address is exclusively utilized for dispatching research updates and will never be shared, sold, or distributed to third parties. We leverage industry-standard cryptographic protocols to protect all stored information. If you wish to purge your subscription data or request details on collected metrics, please contact operations@a3ther.research.

TERMS OF SERVICE

By accessing the A3THER Research Portal or subscribing to our daily briefs, you agree to comply with and be bound by these Terms of Service. All content, analytical tools, models, and projections are the intellectual property of A3THER Research LLC and are licensed for personal, non-redistributable use. Redistribution, replication, or commercial exploitation of A3THER reports is strictly prohibited without prior written authorization. The services are provided "as is" and "as available" without warranties of any kind. We reserve the right to suspend subscription access or terminate user privileges at our discretion for violations of intellectual property guidelines.

FORM ADV PART II (SUMMARY)

By accessing the A3THER Research Portal, you acknowledge that A3THER Research LLC is not a registered investment adviser (RIA) and does not file a Form ADV with the SEC or state regulators. This portal provides research and publishing services exclusively. For entities requiring regulatory disclosure packets or operational due diligence (ODD) questionnaires, A3THER maintains an institutional fact sheet outlining its operations, data sources, and analytical methodologies. To request the institutional factsheet, please submit an inquiry through your family office or institutional representative to legal@a3ther.research.