The AI Energy Economy — Part 2.2: The Last-Meter Bottleneck

The AI Energy Economy — Part 2.2: The Last-Meter Bottleneck

Part 2 of this series focuses on the companies that build the physical infrastructure required to power artificial intelligence — the turbines, transformers, substations, transmission lines, and grid hardware that move electricity from where it is generated to where it is consumed. In Part 2.1, we examined the firms that connect the grid itself, carrying massive amounts of power across long distances and into major load centers.

But even after electricity reaches a data-center campus, the job is not finished.

Power still has to be distributed safely inside buildings, routed to individual racks, protected from electrical faults, and continuously cooled. As AI data centers grow larger and denser, this final step — the space between the grid and the computer chip — is becoming one of the most important constraints in the entire system. This section focuses on that “last-meter” layer of AI electrification — the point inside facilities where power must be safely delivered and heat must be removed for AI systems to operate at all.

1. What “Torque” Means — in Practical Terms

Throughout this series, I use the term “torque” as a metaphor borrowed from mechanics to describe how strongly a company’s earnings and stock price respond when AI drives electricity demand higher. Just as torque in a car engine determines how quickly a vehicle can accelerate, financial torque reflects how quickly a company’s performance can accelerate when market conditions shift in its favor.

Low-Torque vs. High-Torque Businesses

Some companies benefit slowly and steadily from rising AI demand. Regulated utilities, for example, earn approved returns over long periods. Their profits rise gradually as demand grows, but their upside is limited by regulation and long planning cycles. These are low-torque businesses — stable, dependable, but not explosive. Think of them like a heavy truck: reliable and powerful, but slow to accelerate.

Other companies sit closer to the points where AI creates real stress in the electricity system. When infrastructure becomes constrained, or when demand spikes faster than expected, their products become urgently needed. Orders accelerate, pricing improves, and earnings can re-rate quickly. These are higher-torque businesses — more sensitive to AI demand and more responsive when bottlenecks emerge. Think of them like a sports car: capable of rapid acceleration when conditions are right.

The Key Principle: Torque Increases Closer to the Constraint

Bulk generation and long-distance transmission sit upstream in the power system, far from where things tend to fail quickly. Power distribution and cooling systems, however, sit right where AI infrastructure can break if something goes wrong. That proximity to critical failure points is what gives certain companies their torque — when problems arise at these choke points, solutions become urgently valuable, and the companies providing them can see their financial performance accelerate rapidly.

2. Why AI Changes Everything Inside the Data Center

Traditional data centers were designed for modest power loads and air cooling. AI has rendered those assumptions obsolete.

Modern AI servers consume far more electricity and generate vastly more heat than earlier computing equipment. Entire AI campuses can now demand as much power as a small city, while individual racks draw levels of electricity that were once unimaginable in commercial buildings.

At those levels, problems appear quickly:

• Electricity must be routed precisely to avoid overheating or electrical faults
• Power systems must be enclosed and protected to prevent fires and downtime
• Heat must be removed continuously to keep equipment operating

If any of these fail, the entire facility can shut down.

This is where nVent (NVT) and Vertiv (VRT) become essential.

3. nVent (NVT): Making Extreme Power Density Safe

nVent’s role in the AI energy economy is best understood as the physical safety and routing layer of electricity.

The company designs and manufactures enclosures, busway systems, and power-distribution hardware that allow very large amounts of electricity to be delivered safely inside data centers, utility facilities, and industrial sites. These systems are engineered to prevent electrical faults, overheating, and catastrophic failures as power density rises.

As AI pushes electricity use higher and concentrates it into smaller physical spaces, simply delivering more power is not enough. Power has to be contained, spaced, protected, and routed correctly. nVent’s products exist specifically to solve that problem.

After divesting its thermal-management business, nVent is now a focused electrical-infrastructure company centered on Systems Protection and Electrical Connections. Its exposure to infrastructure end markets has increased dramatically, with the infrastructure segment — including data centers and power utilities — now accounting for more than 40% of revenue, up from low-teens percentages at the spin-off.

From an investment standpoint, this positioning gives nVent more torque than broad industrial companies. It does not benefit from every aspect of electrification, but it benefits directly from one of AI’s most pressing challenges: delivering large amounts of power safely in confined spaces. As rack densities rise and liquid cooling becomes more common, demand for nVent’s solutions can grow faster than overall electricity demand.

4. Vertiv (VRT): Keeping AI from Overheating

If nVent is about electrical safety, Vertiv is about thermal survival.

Vertiv specializes in cooling, thermal management, and power-conditioning systems for data centers. Its products remove heat from high-density AI servers, maintain stable operating temperatures, and ensure that computing workloads can run continuously without thermal shutdowns or damage.

AI workloads generate far more heat than traditional computing. In many cases, air cooling alone is no longer sufficient. This has driven rapid adoption of liquid cooling, advanced thermal systems, and tightly integrated power-and-cooling designs — areas where Vertiv has deep expertise.

Vertiv’s torque is high because cooling becomes critical very quickly. Once power density crosses certain thresholds, cooling is no longer an incremental upgrade; it becomes essential. If heat cannot be removed efficiently, servers cannot operate, regardless of how much power is available.

As a result, Vertiv is one of the most direct beneficiaries of AI-driven infrastructure stress. Its growth and valuation are closely tied to how aggressively hyperscalers push compute density and performance.

5. How nVent and Vertiv Fit Together

Although both nVent and Vertiv benefit from AI, they solve different problems.

nVent focuses on how electricity is safely delivered and contained. Vertiv focuses on how the heat created by that electricity is removed. They often operate side by side in the same facilities, but they address different failure modes.

Together, they represent the physical constraint layer of AI infrastructure — the point where power and heat stop being abstract concepts and start determining whether systems can function at all. This is also why their torque is higher than that of broader electrification players such as Eaton or Schneider Electric, which supply essential equipment across many markets and technologies.

6. A Note on nVent and Vertiv as Investments

From an investment perspective, companies like nVent and Vertiv are best understood as higher-torque infrastructure plays, not steady regulated utilities or broad industrial conglomerates. Their earnings and stock prices tend to respond more sharply when AI-driven infrastructure demand accelerates — and, as a secondary effect, when investor sentiment around AI is elevated.

These companies can benefit disproportionately when hyperscalers ramp spending, but they can also be more volatile than diversified electrification leaders such as Eaton or Schneider Electric. Their performance depends less on slow-moving rate cases or multi-year planning cycles, and more on how quickly AI infrastructure pushes against physical limits in power delivery and cooling.

At current levels, both companies look fundamentally strong, with clear exposure to durable AI-driven infrastructure demand. At the same time, valuations reflect a meaningful amount of optimism, which means returns are likely to be more sensitive to changes in AI spending expectations than in lower-torque parts of the energy system.

In that context, both nVent and Vertiv are best viewed as Buys — but not defensive ones. nVent offers focused exposure to power distribution and electrical safety, with growth tied to rising rack density and infrastructure hardening. Vertiv offers more direct exposure to AI-driven cooling demand, with potentially higher upside — and higher volatility — as compute intensity increases.

In practical terms, these companies fit best as targeted allocations within a broader AI energy portfolio, complementing lower-torque utilities, grid builders, and automation companies rather than replacing them.

7. Why This Matters — and What Comes Next

For AI, electricity generation, transmission, and grid expansion are highly visible challenges. But the bottleneck does not end where the grid connects to a facility. As power density increases, the real limits increasingly shift inside buildings — to how electricity is distributed, protected, and cooled. If those systems fail, it doesn’t matter how much money is spent on power plants, long-distance transmission lines, or substations — the facilities that step electricity down from high-voltage lines so it can be safely used.

What this means, in practical terms, is that AI infrastructure only works if electricity can be safely delivered and heat can be continuously removed inside the facility itself. Power that cannot be routed, contained, and cooled cannot be used. That reality is what makes companies like nVent and Vertiv so important today — and why the next layer of the AI energy story focuses on the automation, controls, and software that keep these environments operating safely and reliably at scale, the subject of Part 3.

• The AI Energy Economy Series:

  • The AI Energy Economy — Part 1: The Nuclear & Utility Winners of the AI Power Boom
  • The AI Energy Economy — Part 2: The Pick-and-Shovel Suppliers Powering AI Electrification
  • The AI Energy Economy — Part 2.1: Companies That Connect the Grid
  • The AI Energy Economy — Part 2.2: The Last-Meter Bottleneck
  • The AI Energy Economy — Part 3: Industrial Automation, Cooling & Controls
  • The AI Energy Economy — Part 4: Merchant Power, Nuclear Scarcity, and AI Contracts