The explosive growth of artificial intelligence is creating an unprecedented energy crisis, and the companies that solve it stand to reshape global infrastructure for decades. Two very different energy players are now at the center of this transformation: Constellation Energy, which operates the largest fleet of carbon-free nuclear reactors in the United States, and GE Vernova, which manufactures the turbines and grid technology needed to distribute that power globally. Both companies are racing to meet the surging electricity demands from AI data centers, but they're taking fundamentally different approaches to the same problem.

Why Is Data Center Power Suddenly the Biggest Infrastructure Challenge?

The numbers tell the story. Constellation Energy reported nearly $25.5 billion in revenue for fiscal year 2025, an 8.3% increase over the previous year, driven largely by long-term power agreements with major technology companies. Microsoft alone signed a 20-year deal to restart the Crane Clean Energy Center, while Meta Platforms committed to purchasing power from the Clinton plant. These aren't small contracts; they represent a fundamental shift in how the world's largest tech companies are planning their infrastructure.

The challenge extends beyond simply generating more electricity. Data centers, especially those running advanced artificial intelligence workloads, consume enormous amounts of water for cooling. Microsoft has tackled this problem head-on, reducing its data center water intensity by nearly 90% since the early 2000s, dropping from 2.3 liters per kilowatt-hour to just 0.27 liters per kilowatt-hour by 2025. The company has achieved a 25% reduction in water-use intensity across its owned data center fleet and is targeting a 40% improvement by 2030.

How Are Companies Reducing Data Center Water Consumption?

• Direct Air Cooling with Evaporative Assist: Since 2008, Microsoft has used outside air for cooling and adds water only when temperatures exceed 85 degrees Fahrenheit, meaning in cooler regions like Northern Europe, no water is needed year-round.
• Zero-Water Chip-Level Cooling: In 2024, Microsoft introduced a new data center design for AI workloads that uses closed-loop, direct-to-chip cooling, which recirculates water inside sealed systems and targets heat at the chip level rather than cooling entire rooms.
• Alternative Water Sourcing: Microsoft is increasing its use of recycled, reused, and non-potable water, with facilities in Quincy, Washington using 74% recycled water and Singapore facilities using 99%, while also installing rainwater harvesting systems across multiple countries.
• Operational Optimization: The company refines temperature and humidity settings to reduce overcooling and audits actual water use against design expectations using real-time weather data and operational analytics.

These improvements are producing measurable results. In Phoenix, Arizona, where water stress is particularly acute, Microsoft reported a 23% year-over-year improvement in water use effectiveness in fiscal year 2025. The company is now deploying similar improvements across its data centers globally, recognizing that water availability has become as critical to site selection as power and fiber access.

What's the Difference Between Constellation Energy and GE Vernova?

Constellation Energy operates as the power producer. The company serves roughly 2.5 million customer accounts, including 75% of the Fortune 100 companies, and operates the largest carbon-free energy fleet in the United States. Its business model is straightforward: generate electricity from nuclear reactors and sell it under long-term contracts. The company reported net income of approximately $2.3 billion in fiscal year 2025, though its net margin fell to 9.1% from 15.9% in the previous year, reflecting the capital-intensive nature of nuclear operations.

GE Vernova takes a different approach. Rather than generating power, it manufactures the equipment that generates and distributes it. The company has an installed base of over 7,000 gas turbines and 59,000 wind turbines, and it sells to utilities, governments, and large industrial users like Amazon. For fiscal year 2025, GE Vernova generated close to $38.1 billion in revenue with net income of approximately $4.9 billion, and notably, the company carries zero total debt relative to shareholder equity.

The distinction matters for investors. Constellation Energy offers steady cash flow and regular dividends, making it appealing to conservative investors who want exposure to the growing demand for carbon-free electricity. GE Vernova, by contrast, is positioned to capitalize on the global infrastructure upgrade cycle, as utilities worldwide modernize aging electrical grids to handle increased energy demands from technology companies. However, GE Vernova's stock carries a higher valuation that already reflects this growth optimism.

What Are the Key Risks Each Company Faces?

Constellation Energy confronts significant integration challenges following its acquisition of Calpine, which added 23 gigawatts of capacity to its fleet. The company must also navigate volatile wholesale market prices that can cause earnings to fluctuate significantly, and its nuclear operations require ongoing license extensions from the Nuclear Regulatory Commission and continued federal support to remain financially viable.

GE Vernova deals with different pressures. The company operates on long-cycle fixed-price contracts where inflationary pressures can lead to cost overruns and lower profitability. It's also exposed to supply chain disruptions for critical materials, often involving geopolitical tensions, and faces intense competition from global giants like Siemens Energy and Mitsubishi Heavy Industries while navigating complex international trade regulations.

Beyond the companies themselves, the broader infrastructure challenge is becoming a governance issue. Microsoft has invested more than $500 million in over 75 water and wastewater infrastructure projects since 2020, and near its data center in Leesburg, Virginia, the company is funding more than $25 million in water and sewer improvements. This reflects a wider trend where data center growth now depends on more than land, power, and fiber access; water availability, utility capacity, and community trust are becoming core business risks.

The convergence of these trends suggests that the next phase of AI infrastructure will be defined not by computing power alone, but by the ability to secure reliable, sustainable energy and water resources. Companies that can solve this puzzle will shape the digital economy for decades to come.