The AI revolution is driving nearly insatiable demand for hardware like high-performance GPUs, CPUs, networking equipment, and storage. Demand for this hardware is so strong that AI companies are buying Nvidia's GPUs and memory chips from Micron hand over fist. As a result, the share prices of AI hardware companies have skyrocketed in recent years.
However, hardware is only part of the estimated $5.2 trillion in capital spending McKinsey sees on AI infrastructure through 2030. Companies will also need to invest heavily in developing the physical data center shells, power grids, and electricity-generating capacity to support the AI boom. Here's a look at some of the overlooked pick-and-shovel plays that are crucial to the AI gold rush.
Image source: Nvidia.
The cloud is a physical space
AI technology requires specialized computing infrastructure. The centerpiece is an AI factory, a purpose-built data center that houses critical hardware. In addition to the hardware, this physical building shell requires advanced cooling systems and access to a substantial amount of reliable power.
Several companies specialize in developing data centers (shells and operating facilities). For example, data center REITs such as Equinix (EQIX +0.63%) and Digital Realty (DLR 0.06%) play a crucial role in helping develop physical real estate to support AI. Equinix is building out several xScale data centers (large-scale, AI-ready facilities). In 2024, it formed a $15 billion joint venture to purchase land and build several new state-of-the-art xScale facilities. Meanwhile, Digital Realty launched its first U.S. Hyperscale Data Center Fund in 2025 to support up to $10 billion of data center investments. It also partnered with Blackstone in late 2023 on a $7 billion joint venture to build large-scale data centers.
NASDAQ: EQIX
Key Data Points
Brookfield Infrastructure (BIPC 0.37%)(BIP +0.28%) is another notable data center developer. The company has acquired several global data center platforms over the years to build up its capabilities. It currently operates over 140 data centers worldwide, with a total capacity of 1.6 gigawatts (GW). It sees potential to develop an additional 3.4 GW of capacity across its platform in the future. Additionally, Brookfield is investing in other data infrastructure to support AI, including plans to deploy advanced fuel cell technology from Bloom Energy to power data centers.
Powering the AI revolution
The advanced GPUs, cooling systems, and other hardware required to support AI demand a massive amount of electricity. Many AI data center campuses need over 1 GW of electricity, enough to power about 750,000 homes. According to an estimate by leading AI company Anthropic, the U.S. will need at least 50 GW of power for AI by 2028 alone.
Several energy companies are investing in expanding the energy infrastructure necessary to support AI. They're constructing new energy transmission lines, natural gas pipelines, and clean power generation assets.
One of the leaders in energy infrastructure development is NextEra Energy (NEE 0.05%). The company is exploring the potential to invest over $25 billion in the coming years in developing additional electricity transmission projects to support the grid. Additionally, it has an extensive gas pipeline network that it expects to expand in the future to support growing demand for gas power. On top of that, NextEra is a leader in developing renewable energy. The company also formed strategic partnerships with Google to advance the deployment of nuclear energy and co-develop multiple large-scale data center campuses.
NYSE: NEE
Key Data Points
Williams (WMB +0.10%) is another leader in helping support the AI power boom. The natural gas pipeline company has several projects underway to increase the supply of gas across the country that should enter commercial service through 2030. The company is evaluating another 30 projects totaling over $14 billion of investment that it could build in the 2027 to 2033 time frame. Additionally, Williams has $5.1 billion of projects under construction to develop and operate gas-power generating capacity for data center customers. It has several additional power innovation projects in the backlog.
Building the physical infrastructure to support AI
AI has the potential to be one of the most impactful technologies in history. It can't achieve its immense potential without the physical infrastructure needed to turn chips into AI supercomputers. As a result, companies from Brookfield Infrastructure to NextEra Energy will play a vital role in supporting the AI era, enabling their investors to cash in on the gold rush.
