AI ignites electricity demand! Morgan Stanley: Fuel cells, Bitcoin mining farms, and other niche sectors are underestimated by the market

With the rapid advancement of artificial intelligence technology, global demand for electricity is entering a new round of growth.

On Wednesday local time, Morgan Stanley's analyst team led by Stephen C Byrd released the 2025 Global Generative AI Power Outlook, citing the International Energy Agency (IEA) predicting that by 2026, electricity consumption in data centers, AI, and the cryptocurrency industry could double, with electricity consumption in data centers potentially exceeding 1000 TWh, equivalent to Japan's annual electricity consumption.

In this major trend, certain sub-sectors are being underestimated by the market. According to Morgan Stanley's report, areas such as fuel cells, Bitcoin mining sites, electricity transmission, and grid enhancement technologies are among them.

Morgan Stanley believes that the repurposing of fuel cells and Bitcoin mining sites will provide new energy solutions for data centers, and that electricity transmission and grid enhancement technologies are crucial for the efficient and safe electricity use of data centers, with a focus on power generation facilities like nuclear power plants near data centers.

Additionally, legislative proposals in Texas may promote the construction of natural gas power plants, and the new Trump administration will take measures to support the growth of AI infrastructure.

Fuel Cells, Bitcoin Mining Sites, Electricity Transmission... Being Underestimated

The report points out that some areas play a key role in the growth of AI electricity demand, but the market has not fully recognized their value and potential, including:

1. The Development Prospects of Fuel Cells

Morgan Stanley is particularly optimistic about the application prospects of fuel cell technology, stating that this technology provides an efficient energy solution for data centers due to its significant time and operational advantages.

The report notes that compared to traditional power supply, fuel cells can be deployed more quickly, have higher energy efficiency, and lower emissions, making them an ideal choice to support AI infrastructure.

Morgan Stanley is optimistic about China's fuel cell industry, expecting it to show rapid development, cost reduction, continuous improvement in infrastructure construction, and accelerated promotion of hydrogen fuel cell vehicles in the future.

2. The Repurposing of Bitcoin Mining Sites

Morgan Stanley points out that many Bitcoin miners' valuations are low compared to their electricity generation capacity, indicating that the market has not fully recognized the potential of these sites to be converted into data centers.**

Bitcoin mining sites typically have substantial electricity infrastructure that can be repurposed to support the operations of data centers, especially in energy-intensive AI and big data processing fields.

As AI technology develops, the demand for electricity is sharply increasing, and the energy infrastructure originally used for Bitcoin mining can be transformed into valuable assets supporting AI data centers. This conversion can not only enhance the economic benefits of these sites but also reduce the time and cost required to build new data centers 3. Power Transmission Value Chain

The report mentions that power transmission is a key link connecting power plants and data centers. Especially in the context of the rapid development of AI technology, the demand for electricity from data centers is continuously increasing, making power transmission even more important.

Morgan Stanley believes that with the development of AI and digital infrastructure, the technology and services of power transmission companies will face greater demand, thus playing a more significant role in the power transmission value chain.

4. Grid Enhancement Technologies (GETs)

Grid enhancement technologies refer to a series of hardware and software tools that can reconfigure the transmission grid and adjust its parameters, increasing grid capacity and enhancing the reliability and safety of power transmission, further releasing more grid access for data centers and new factories.

Morgan Stanley believes that the Trump administration prioritized funding for foundational AI research and computer infrastructure, focusing on eliminating regulatory restrictions on the scheduling of AI technologies, which may promote the application and development of grid enhancement technologies.

Catalysts Driving AI Power Demand Growth

Additionally, Morgan Stanley proposed several potential catalysts in its report that may drive the growth of electricity demand in the AI sector.

1. Co-location of Nuclear Power Plants and Large Data Centers

The report emphasizes that with the development of AI technology, the electricity demand of data centers has surged, particularly in the United States, where nuclear power generation has significant potential to meet these demands. Establishing nuclear power plants in close proximity to data centers will reduce losses during the power transmission process and allow for uninterrupted power supply without external connections.

Therefore, Morgan Stanley suggests that more attention should be paid to companies providing on-site "main power supply" for new data centers, as there has been a recent surge in interest from data center developers in installing main power generation on-site.

2. Off-grid Data Center Strategy

The off-grid data center strategy involves technologies such as fuel cells, turbines, and energy storage. Morgan Stanley believes that these technologies can help data centers operate without relying on traditional grids, enhancing the stability and reliability of power supply.

For example, ECL announced the world's first modular, sustainable, off-grid data center, which uses green hydrogen as its primary power source, achieving 99.9999% uptime. The design of this off-grid data center includes using water generated from hydrogen power for cooling, eliminating reliance on external resources, and allowing for rapid deployment, with a planning to construction cycle of only 6-9 months, far less than the 18-24 months typical for traditional data centers. Additionally, this type of data center has very high energy efficiency, with a Power Usage Effectiveness (PUE) of 1.05, and can reach up to 50KW per rack.

3. Texas Legislative Proposals

The report points out that legislative proposals in Texas may promote the construction of new natural gas power plants, which are crucial for supporting the energy demands of new data centers. The logic behind these proposals is that with the rapid growth of energy consumption in data centers, especially in the context of the rapid development of AI technology, the demand for stable and sufficient power supply is becoming increasingly urgent. As an energy powerhouse, Texas's legislative proposals indicate a focus on investing in energy infrastructure, aiming to meet the growing energy demands of data centers by increasing power generation capacity, ensuring that these critical facilities can operate stably, and supporting the development of the digital economy.

4. Support from the Trump Administration

Morgan Stanley stated in its report that the Trump administration is expected to take measures to support the growth of AI infrastructure, particularly through the Federal Energy Regulatory Commission (FERC).

The Trump administration's policies tend to relax regulations on AI issues while strengthening export controls, focusing on competition with specific countries, and reducing antitrust enforcement related to the European Union. Particularly in the field of artificial intelligence, the Trump administration has prioritized funding for foundational AI research and computing infrastructure, machine learning, and automated systems, aiming to eliminate regulatory restrictions on the scheduling of AI technologies.

Additionally, the Trump administration may also revoke regulations imposed by previous administrations that restrict fossil fuel production and consumption, which would further promote the growth of AI-related data capacity and increase electricity demand, necessitating significant investments in new power generation and transmission infrastructure.

5. Announcement of New Large Data Centers

Morgan Stanley believes that the construction of large data centers will be driven by collaboration between super giants and fully integrated regulated utilities. This collaboration can ensure that the energy demands of data centers are met, as these partnerships typically involve power development companies that can provide the necessary power supply to support the operation of data centers.

Morgan Stanley also predicts that by 2025, tech giants will invest a total of approximately $300 billion in hyperscale capital expenditures, primarily to support the development of generative AI and massive language models