From Data Centers to Copper: AI Infrastructure Race Goes Deeper 

The latest update to Mogin Law’s AI Deal Table shows that artificial intelligence competition is continuing to move past models and cloud capacity and even deeper into the power, wiring, materials, and industrial systems needed to operate AI at scale, in other words, AI infrastructure. 

The history of technology is often told through inventions, but those innovations would still be on dusty shelves without new infrastructure designed to support them. The automobile required roads, fueling networks, and manufacturing capacity. The internet depended on fiber-optic cables, data centers, and telecommunications networks. Electrification demanded power plants, transmission lines, transformers, copper wiring, and decades of investment before electricity became widely available. 

Artificial intelligence has reached that kind of moment. 

The latest update to the Mogin Law AI Deal Table makes it clear: the artificial intelligence revolution has entered its infrastructure and electrification phase. The defining AI competition questions are no longer limited to models, applications, or software platforms. They increasingly involve power generation, data centers, compute capacity, optics, networking systems, copper-intensive electrical infrastructure, and access to the physical systems needed to run AI at scale. 

From AI Infrastructure Awareness to Infrastructure Control 

The first wave of the AI boom centered on foundation models and applications. Companies raced to develop more capable large language models. Investors focused on model developers, AI assistants, and software platforms. That phase has not ended, but the competition has moved deeper. 

The central question is no longer merely who has the best model, or even who has the most cloud capacity. It is who can secure the power, land, chips, data-center capacity, networking equipment, cooling systems, transmission resources, and materials needed to keep advanced AI systems operating at scale. 

Our report on the SpaceX/Reflection deal described how the AI compute debate was already moving deeper into infrastructure. The latest AI Deal Table update suggests the next step in that progression: the race now extends beyond compute and data centers into power, copper, cooling, networking, and the industrial inputs needed to operate AI at scale. 

The Race to Secure Power 

Another revealing example is Microsoft’s long-term arrangement with Chevron to support a major West Texas data-center project. Chevron announced a 20-year power agreement for a co-located facility expected to provide approximately 2.67 gigawatts of dedicated capacity to a Microsoft-operated data center. 

AI uses a lot of electricity, so major AI participants want to lock in access to dedicated energy resources. Companies that can obtain large, reliable, long-term electricity supplies may be better positioned to build and expand the data centers on which advanced AI depends. 

Data centers illustrate this deeper competition. They concentrate chips, servers, cooling systems, networking equipment, power contracts, land, permitting, and capital. Building them is slow, expensive, and constrained by local energy availability, grid capacity, construction timelines, and access to specialized equipment. Firms that can finance, build, reserve, or partner for capacity may gain advantages over rivals that must wait for space, power, chips, or interconnection. 

The Race Extends to Copper, Cooling, and the Industrial Supply Chain 

This is why the recent focus on copper is noteworthy. The story is not really about copper as a commodity. It is about wiring, transmission, transformers, electrical infrastructure, cooling systems, and the industrial capacity required to support a rapidly expanding compute economy. 

Recent analysis by S&P Global describes copper as essential to electrification, digitalization, AI, data centers, electric vehicles, and defense, noting that the metal has been central to electrification since Edison’s early Lower Manhattan power network. That connection helps explain why AI’s infrastructure race is now reaching into materials and industrial systems that once seemed far removed from software competition. 

If the most important AI firms can secure preferred access to power, data-center capacity, chips, networking systems, or critical materials, rivals may face higher costs, delays, or practical limits on their ability to scale. 

Amazon illustrates the competitive significance of infrastructure control. The company built logistics and cloud infrastructure to serve its own business, but those systems later became indispensable to many other firms. With Amazon Web Services, Amazon already controls a major layer of the cloud and compute infrastructure on which AI development depends. The AI infrastructure race raises a similar question: when firms build systems for themselves, do those systems become gateways others must use to compete? 

The Competitive Stakes 

• AI developers may be disadvantaged if they cannot obtain compute, cloud services, power, or networking resources on competitive terms. 

• Data-center operators may become gatekeepers if capacity remains constrained and demand continues to rise. 

• Utilities and energy producers may gain new leverage as AI companies compete for large, reliable, long-term power supplies. 

• Industrial suppliers of transmission equipment, cooling technology, optical systems, and materials may become more strategically important to the AI economy. 

• Regulators and antitrust enforcers may need to look beyond AI products and examine whether control of infrastructure inputs creates bottlenecks, exclusionary risks, or durable advantages. 

Who could be affected by AI infrastructure chokepoints?

AI developers and model companies. As competition shifts from software toward power generation, data centers, networking, cooling systems, and electrical infrastructure, AI developers may find that access to critical inputs becomes as important as algorithmic innovation. Companies that cannot secure sufficient infrastructure could face competitive disadvantages regardless of the quality of their models

Businesses adopting AI. Companies that rely on AI tools for operations, customer service, software development, research, or analytics ultimately depend on the infrastructure supporting those systems. Competition for power, compute resources, and data-center capacity may influence pricing, availability, reliability, and the pace of AI innovation.

Energy, utility, and infrastructure providers. Electric utilities, power generators, transmission operators, data-center developers, networking suppliers, and manufacturers of copper-intensive electrical equipment are increasingly becoming strategic participants in the AI economy rather than background service providers.

Investors and infrastructure developers. Capital is increasingly flowing into power generation, data centers, networking systems, cooling technologies, and related infrastructure needed to operate AI at scale. Investors evaluating AI opportunities may need to look beyond software companies and consider the businesses that control essential inputs.

Companies that depend on fair access to infrastructure. Businesses seeking access to data-center capacity, compute resources, networking systems, or energy infrastructure may face challenges if key inputs become concentrated among a small number of firms. Questions about access, interoperability, exclusivity, and infrastructure bottlenecks are likely to become increasingly important competition issues.

Some of the most important AI-related antitrust stories of the next decade may not concern the newest model release, but who controls the supporting systems and who struggles in the face of critical chokepoints.

Edited by Tom Hagy, Editor-in-Chief, Mogin Law Blog.