Plot twist in the AI race—the next big flex is actually physical infra?

I. Power and Energy Networks: The “Blood” of AI

1. Constellation Energy(CEGON)| Market Cap: Approximately $99 billion

• Main Business: The largest nuclear power operator and carbon-free power generator in the U.S.
• Core Advantage: Nuclear power is the only energy source capable of providing 24/7 zero-carbon “base-load power.” As tech giants scramble to secure long-term power purchase agreements (PPAs), CEG is the most direct beneficiary thanks to its vast nuclear power portfolio. Q1 2026 revenue reached $11.1 billion (significantly exceeding expectations); if PPAs continue to materialize, the valuation still has upside potential.

2. GE Vernova (GEVON) | Market Cap: Approximately $300 billion

• Main Business: A global power equipment giant; core products include gas turbines, grid equipment, and energy storage systems.
• Core Advantages: When renewable energy cannot meet the base-load demand of AI, natural gas power generation becomes a critical transitional solution. GEV holds a dominant position in the global gas turbine market. By the end of Q1 this year, its gas power generation orders surged to 100 GW, and the order backlog is expected to reach at least 110 GW by year-end, indicating extremely strong earnings certainty.

3. Eaton (ETNON) | Market Capitalization: Approximately $164 billion

• Main Business: A global leader in smart power management, covering the entire value chain “from the grid to the chip.”
• Core Advantages: Provides integrated power distribution, circuit protection, and liquid cooling solutions for data centers. Driven by strong demand from AI data centers, both revenue and earnings exceeded expectations in Q1 2026. The company raised its full-year organic growth guidance to 10% and expects EBITDA growth to climb to 18%–24% in the second half of the year.

4. Vistra (VSTON) | Market Cap: Approximately $56 billion

• Main Business: The largest unregulated electricity producer and retail energy supplier in the U.S.
• Core Advantages: Possesses 44 GW of installed generating capacity (including natural gas and nuclear power), serving nearly one-third of Texas’s electricity consumers. Recently signed a major nuclear power supply agreement with Meta and acquired Cogentrix to add 5.5 GW of natural gas capacity; the company is a key beneficiary of the AI-driven surge in electricity demand.

5. Oklo (OKLOON) | Market Capitalization: Approximately $10 billion

• Main Business: A pioneering developer of small modular reactors (SMRs).
• Core Advantages: Focuses on the “Aurora Powerhouse” fission power plant, which employs a “nuclear power as a service” model based on long-term power purchase agreements. Although still in the early stages, the company aligns with the long-term narrative of AI data centers’ need for clean, reliable baseload power and enjoys a very high sector-specific growth premium.

II. Data Center Physical Infrastructure and Thermal Management: The “Skeleton and Cooling” of AI

6. Vertiv Holdings (VRTON) | Market Capitalization: Approximately $138 billion

• Main Business: The undisputed global leader in critical data center infrastructure and liquid cooling/thermal management.
• Core Advantages: As AI chip power consumption pushes beyond its limits, liquid cooling has shifted from an “option” to a “necessity.” Vertiv has a backlog of approximately $15 billion in orders and is collaborating with NVIDIA on joint development. It has recently completed two consecutive strategic acquisitions in thermal management, further solidifying its dominant position in the high-performance computing cooling sector.

7. GE Vernova (GEVON) | Market Capitalization: Approximately $110 billion

• Main Business: The world’s largest data center REIT (Real Estate Investment Trust), providing colocation and interconnection services.
• Core Advantages: Operates 260 data centers across 71 markets worldwide. As the “landlord” of AI computing infrastructure, Equinix directly benefits from the demand for AI cluster expansion. Higher power density translates to higher per-rack rental rates, creating a “double boost” from both rising rents and expanded capacity.

8. CoreWeave (CRWVON) | Market Cap: Approximately $60 billion

• Main Business: A cloud infrastructure platform built specifically for AI workloads, offering GPU computing power leasing, AI-native cloud services, and data center colocation.
• Core Advantages: Deeply integrated with NVIDIA; its client base includes leading AI players such as OpenAI, Anthropic, Meta, Google, and Microsoft. As of Q1 2026, its revenue backlog stood at $99.4 billion, with full-year revenue guidance of $12.0–13.0 billion. As a “specialized” player focused solely on AI computing infrastructure, CoreWeave directly benefits from the industry shift from AI training to inference.

III. Key Raw Materials: The “Cornerstone” of AI

9. Freeport-McMoRan (FCXON) | Market Cap: Approximately $99 billion

• Main Business: One of the world’s largest publicly traded copper producers.
• Core Advantages: Copper is the “lifeblood” of the electrification era. From power grid upgrades and transformer manufacturing to internal wiring in data centers, every stage of AI computing power expansion relies on massive amounts of copper. Against the backdrop of difficult approvals for new mines, FCX, with its high-quality copper mining assets, stands to benefit long-term from the supercycle in copper demand driven by AI.

10. MP Materials (MPON) | Market Cap: Approximately $10 billion

• Main Business: The largest producer of rare earth elements in the Western Hemisphere, operating the only active large-scale rare earth mine in the United States.
• Core Advantages: Rare earth elements are core materials for AI robot servo motors, high-performance chips, and advanced packaging. In Q1 2026, the company set a record for neodymium-praseodymium oxide production. As the only U.S.-based company with a large-scale rare earth supply chain, MP aligns perfectly with the “supply chain security” strategies of Europe and the U.S., commanding a significant geostrategic premium.

Conclusion