AI Infrastructure Investment: The Hidden Boom in Construction Materials

• The artificial intelligence ("AI") boom isn't just about semiconductors or cloud software – it's also about concrete, steel, copper, and cables.

• Hyperscaler capital expenditures (Amazon, Microsoft, Google, Meta Platforms) are driving a physical build-out of U.S. AI hubs.

• Materials firms may be the quiet winners.

The Physical Side of the AI Boom

AI is growing faster than almost any other industry. In 2023, the global AI market was valued at $189 billion. By 2033, it's expected to reach $4.8 trillion. And with this growth comes the need for data centers... lots of them.

AI is driving innovation across nearly every sector. But behind the scenes, AI data centers provide the computational horsepower to train and run AI models. As demand rises, so does the need for more data centers and power stations built specifically for AI.

The physical side of AI infrastructure – construction and materials – is often overlooked. Yet it's a critical part of the AI story. And for investors, it's a space ripe with opportunity. Not just in chips and semiconductors, but in the materials that make AI possible.

Five Key Takeaways – AI Infrastructure Investment Outlook

• • 1. UBS projects that global AI infrastructure spending will reach $375 billion in 2025 and $500 billion by 2026.
  • 2. AI data centers will drive copper use from 500,000 metric tons per year to 3 million by 2050.
  • 3. The American Cement Association ("ACA") predicts the U.S. will need around 1 million metric tons of cement to construct AI data centers by 2028. And the demand for steel will be just as great… one hyperscale AI data center requires up to 20,000 tons.
  • 4. ESG-focused (Environmental, Social, and Governance) and green building materials, as well as renewable energy sources, are growing trends in AI data center construction.
  • 5. Companies supplying materials and infrastructure are positioned for long-term growth... these are the "quiet winners" of the AI boom.

What AI Infrastructure Is and Why It Matters Now

AI infrastructure combines specialized hardware and software to support AI and machine learning ("ML") workloads... ranging from data processing and storage to model training and deployment. This infrastructure delivers the computing power, data handling, and management tools necessary to develop and scale AI.

The major components include:

• • • Specialized processors such as GPUs (graphics processing units) and TPUs (tensor processing units), a type of application-specific integrated circuit designed for ML applications
  • • High-speed networks for fast data flow
  • • Enormous storage systems for massive datasets
  • • Data center switches to "direct traffic"
  • • Data-processing libraries that process and prepare volumes of data for AI applications
  • • Power generation and electrical infrastructure to make it all run

Investors are in love with semiconductor, energy, and data center infrastructure stocks... and rightly so. But the materials used to build data centers are just as important. Think of them as the "scaffolding" holding up the AI infrastructure (more on that below).

The AI Data Center Boom: Scale and Spending Trends

By the end of 2025, there will be more than 6,100 public AI data centers worldwide. That includes over 5,500 colocation sites (where companies rent space, power, and equipment) and more than 560 hyperscale sites (huge facilities designed for massive computing needs). By 2030, that total number could reach 8,400.

UBS projects global AI infrastructure spending will hit $375 billion in 2025 and $500 billion by 2026. And McKinsey predicts that by 2030, data centers will require $6.7 trillion in worldwide investment to meet global demand... primarily driven by AI.

Consider this... only a few years ago, far more investment was made in office space construction than data centers in the U.S. Today, investment in data centers is nearly equal to or more than office development. Some estimates project up to $50 billion in data center spending by the end of 2025.

As Bob Dylan once said, "the times they are a-changin'."

So, where are these facilities being built in America?

The usual hubs, like Northern Virginia, an established global data center capital. And Silicon Valley, which has the existing infrastructure as well as a local community of tech giants.

Texas – particularly the Dallas-Fort Worth Metroplex, Austin, and San Antonio areas – offers wide swaths of land and the power infrastructure AI data centers require. Atlanta also has a significant market for data centers.

Other portions of the U.S. are also emerging as potential AI data center hubs... Salt Lake City, Phoenix, Washington, Oregon, Las Vegas, Reno, Chicago, and areas in the center of the country, such as Iowa. These areas offer the right mix of investment, infrastructure, and energy capacity to support large-scale AI growth.

Steel and Concrete: Foundations of AI Infrastructure

The literal foundation of AI infrastructure is the same as other industries... raw building materials. And the demand for those materials is rising.

• • • Concrete: The ACA predicts that the U.S. will need approximately 1 million metric tons of cement to construct massive AI data centers by 2028. Its fire resistance, thermal stability, physical security, and long-term structural integrity make concrete a high-demand commodity for AI data center construction.
  • • Steel: One hyperscale AI data center requires up to 20,000 tons of steel to support the enormous structural load a hyperscale AI data center requires. With more than 500 of these hyperscale centers existing worldwide – and with several more to come – the need for steel is obvious. Even standard AI data centers are built with 1,500 to 2,000 tons of steel.
  • • Aluminum: The use of aluminum throughout AI data centers is widespread. Its light weight and high thermal conductivity make it an ideal material for a variety of uses.

However, bear in mind that all three – steel, concrete, and aluminum – are already massive industries. The global steel industry was estimated to be $1.47 trillion in 2024. The concrete market is estimated to be just more than $1 trillion by the end of 2025. As for aluminum – it's around $178 billion.

All three materials already have established demand drivers from construction, traditional infrastructure, consumer items, and more. While data centers are a relatively new source of demand, it's expected to be minimal in the grand scheme of each industry. If you're investing in materials companies, we encourage you to research individual companies for exposure to the data center market.

Copper, Cables, and Critical Inputs for AI Data Centers

What's inside the structure – and what makes an AI data center the literal powerhouse it is – comprises a long list of materials worthy of investment.

• • • Copper: The U.S. boasts approximately 48 million metric tons of identified copper resources… plenty to support AI infrastructure for decades. Projections show that copper consumption from AI data centers will grow from around 500,000 metric tons to 3 million metric tons per year by 2050. Its superior conductivity, power distribution, durable connections, and ability to transmit enormous amounts of data make copper a critical data center component and a key sector for AI infrastructure investment.
  • • Fiber optics, cables, and transformers: The foundational infrastructure for data transfer and power delivery is fiber optics, cables, and transformers. Combined, they deliver high-speed connectivity, immense power, and the cooling required to support high-demand workloads. Without these components, there would be no AI data centers.

Green Materials and ESG Innovation in AI Construction

Though the federal government is peeling back green initiatives while pushing for more fossil fuel usage, builders and AI data center operators continue to strategically search for energy-efficient materials and climate- (and cost-) conscious solutions.

While most AI data centers are built with traditional materials such as concrete, steel, and aluminum, ESG-focused (environmental, social, and governance) and green materials (as well as renewable energy sources) are emerging, growing trends in the construction of AI data centers.

Some green/ESG materials and innovations to note:

• • • Recycled aluminum and glass: The use of recycled materials significantly cuts the environmental footprint versus using new materials.
  • • Cross-laminated timber: This type of engineered wood product is a lower-carbon alternative to traditional construction materials.
  • • Low-carbon concrete: Traditional cement, while durable and reliable, is a significant source of CO2 emissions. Low-carbon concrete is an example of new concrete mixtures that reduce or eliminate the need for standard cement.
  • • Battery energy storage systems ("BESS's"): A battery energy storage system is essentially a giant rechargeable battery that stores electrical energy for later use. The BESS holds electricity when it is readily available (such as when there is high renewable energy generation or low energy demand). When electricity is needed, the BESS then releases the stored energy to maintain power grid stability and provide backup electricity during power outages. As a bonus, the BESS cuts fossil fuel reliance.
  • • Renewable energy: Wind and solar are clean sources of power that effectively power AI data centers while eliminating fossil fuel usage altogether. However, these energy sources are, by their nature, intermittent... making a BESS (mentioned above) important for any renewable energy power source.
  • • Hydrogen fuel cells: While not yet widely adopted, hydrogen fuel cells are used in some data centers to provide a zero-carbon backup energy source.

Other exciting green innovations are in progress... such as AI-optimized concrete, CO2 mineralization within concrete mix, and hydrogen-based green steel. And these are primed to further accelerate the AI data center boom.

The Key Stocks to Watch in the AI Materials Boom

With all the components involved in building and operating an AI data center, there's no shortage of AI materials investment opportunities in a variety of sectors. Here are some of the top stocks to watch.

• • • Steel
  • • • Nucor (NUE): North America's largest steel producer and recycler. Nucor uses electric arc furnace ("EAF") technology to make lower-carbon steel, a big plus as tech companies push for sustainability. It also supplies structural steel and other parts for data center construction.
  • • • Cleveland-Cliffs (CLF): Offers a wide range of steel products, especially for structural use in data centers.
  • • • Steel Dynamics (STLD): Another domestic EAF steel producer, serving the construction sector with materials tailored for data centers.

Risk to Consider: The steel industry is tied to economic cycles, specifically as it relates to construction and vehicle demand. In the event of a global economic slowdown, weakening demand from traditional sources could hurt steel companies even if demand for data center building continues to increase.

• • • Concrete/Cement
  • • • Vulcan Materials (VMC): The largest U.S. producer of construction aggregates and a major cement supplier, Vulcan is active in key regions for data center growth.
  • • • Amrize (AMRZ): A spin-off of global building materials leader Holcim, Amrize focuses on the North American market and offers cement, aggregates, ready-mix concrete, and more.
  • • • Martin Marietta Materials (MLM): A leading U.S. supplier of aggregates, cement, and concrete products.

Honorable Mention: CarbonCure Technologies, an innovative private company that injects captured carbon dioxide into concrete during mixing, turning it into a mineral that strengthens the mixture and eliminates the carbon.

Risks to Consider: Like steel, concrete demand is tied to economic cycles, specifically new construction and infrastructure starts. In the event of a recession, demand is likely to weaken, hurting concrete demand. And as mentioned above, while data center construction is a new demand driver, it is still relatively small.

• • • Copper
  • • • Freeport-McMoRan (FCX): A global leader in copper mining with diverse worldwide operations.
  • • • Southern Copper (SCCO): Controls major copper reserves in Peru and Mexico, with low-cost operations.
  • • • BHP Group (BHP): A copper giant forecasting huge copper demand growth by 2050, BHP is actively acquiring new copper assets.

Risks to Consider: A theme is emerging. Like the previous industries, copper demand is tied to economic cycles. In fact, copper demand is often viewed as a barometer of economic health. Hence the name, Dr. Copper.

• • • Cables and Networking Solutions
  • • • Corning (GLW): Makes fiber-optic cables and data center connectivity solutions. Working with Broadcom on co-packaged optics to boost AI processing.
  • • • Broadcom (AVGO): A global leader in semiconductors and infrastructure software for AI data centers.
  • • • Credo Technology (CRDO): Focuses on high-speed connectivity, including active electrical cables for AI centers. Its stock has surged in 2024 and 2025.

Risks to Consider: In addition to company-specific risks, all three firms are currently seeing heightened demand as a result of AI capital expenditures ("capex"). Should AI capex slow from the major hyperscalers, it would likely put pressure on all three stocks.

What to Watch Next: Permits, Power, and Policy

As AI data center demand grows, investors should keep an eye on these key developments:

• • • Supply chain bottlenecks: Shortages in advanced semiconductor chips, construction materials, electrical gear (transformers and backup generators), and even water and energy are causing delays and driving up costs.
  • • Permitting fights at local levels: The president's executive order speeds up federal permits, but local resistance is strong. Public opposition delayed or blocked an estimated $64 billion in U.S. data center projects between May 2024 and March 2025.
  • • ESG mandates shaping material choices: Environmental rules are shaping material choices. Builders are being pushed to use sustainable materials, reduce carbon emissions, and adopt circular economy practices.

The 'Invisible Scaffolding' of the AI Boom: Why Materials Matter More Than You Think

The AI boom isn't just about chips and software… it's about what holds it all together. The steel, concrete, copper, and cables that power AI data centers are the hidden drivers of long-term growth. These materials form the "invisible scaffolding" of AI infrastructure.

These stocks may not make headlines, but they're building the future of AI – literally.

David Engle

P.S. The big names like Nvidia, Meta Platforms, and Microsoft get all the headlines. But Stansberry Research isn't only interested in covering the largest AI names, we’re looking for ways to democratize access to artificial intelligence for everyday investors. If that sounds of interest, you can learn more about Whitney Tilson’s N.E.W. System right here.

As Whitney puts it, it’s like having "the same kind of powerful advantage Wall Street firms use to make billions of dollars every year in all market conditions… But with a big twist that has never been possible for regular people until today." Again, get all the details right here.

Get exclusive early access now. Watch the full briefing here.

Frequently Asked Questions ('FAQs')

1. What materials are most in demand for the AI data center boom?

Concrete, steel, aluminum, copper, fiber optics, cables, and transformers are among the most in-demand materials for AI data center construction.

2. Are steel and concrete companies investable AI infrastructure stocks?

Yes. There is expected to be even greater demand for materials like steel and concrete as AI data centers continue to be built around the world. However, as mentioned, demand for steel and concrete is likely to mainly come from traditional sources. We recommend investors do their own due diligence to learn which companies have exposure to new data center construction.

3. How is copper tied to AI infrastructure growth?

Copper is critical to the growth of AI infrastructure because its superior electrical and thermal conductivity is vital for the energy-intensive data centers that power AI. Copper is integrated into nearly every component of AI hardware and facilities, from high-speed data transmission to cooling.

4. What are the ESG/green material trends in AI construction?

ESG and green material trends in AI construction include actually using AI to optimize designs for low-carbon materials, improving supply chain sustainability by identifying ethical material sourcing and reducing waste, and enhancing building performance through AI-driven energy management. AI also facilitates the repurposing of materials like recycled aggregates.

5. What risks could slow AI infrastructure investment?

A variety of factors could potentially slow AI infrastructure investment. These include supply chain bottlenecks, major strains on energy grids, and political and community pushback. Also be aware that new data center construction is being driven by billions of dollars in capital expenditures from hyperscalers. Microsoft alone committed $44.4 billion in capex in 2024. And while hyperscalers continue to commit tens of billions of dollars more in 2025, any slowdown would negatively impact AI infrastructure providers.

6. How can investors find overlooked AI boom investment opportunities in materials?

Stansberry Research has you covered. We've provided you with recommendations regarding stocks we're confident will sustain steady, long-term growth and deliver significant profit. You can also check out Stansberry Research's AI Infrastructure Investment Playbook, where we have identified three under-the-radar materials firms poised for dramatic upside as the AI hub build-out accelerates. These are the names our team believes are set to benefit most from the boom… and few investors are watching them.