Grid Picks and Shovels - Six Names Across Build, Components, and Metals

TL;DR: AI is rewiring the US grid, and the grid in turn drives a picks-and-shovels cycle through contractors, components, and raw materials. Deep-backlog contractors, oligopolistic power-component makers, and domestic aluminum — three buckets that map directly onto the $1.4T rebuild.

The "Picks of the Picks" Frame

In the gold rush, the people who made the most money weren't the miners — they were the ones selling picks and shovels. The frame compounds with every cycle:

• AI is the picks-and-shovels for data centers (Nvidia)
• The grid is the picks-and-shovels for AI (power infrastructure)
• Grid contractors, components, and raw materials are the picks-and-shovels for the grid

The third layer is where I focus most. It's the furthest from the headline, so the market reprices it last, which means the window stays open longer.

1. Grid Contractors - Backlog Decides

Quanta Services (PWR)

North America's largest grid infrastructure contractor. Transmission lines, substations, all the physical build. $44B backlog, roughly three years of revenue locked in. Recent deal with NiSource for 3 GW of new grid capacity. Stock up +43% over six months. Already moved, but if backlogs keep building the trend has more runway.

MasTec (MTZ)

Builds both power lines and natural gas pipelines. Since the executive order keeps gas plants alive, MTZ benefits from grid expansion and gas pipeline demand simultaneously.

Argan (AGX)

Power plant EPC (engineering, procurement, construction) specialist. With the government keeping retiring coal/gas plants online and adding new capacity, AGX is among the most directly positioned to land contracts.

2. Critical Components - Moat Decides

The parts that go into the grid and into data centers come from a small group of specialists. High barriers to entry — that's the moat.

Vicor (VICR)

Power modules. Every chip from Nvidia, AMD, and Google needs power conversion, and Vicor is one of the best in the world at it.

Powell Industries (POWL)

Switchgear and electrical distribution. Think of them as the circuit breakers for data centers — controlling where electricity flows and protecting against overloads. Stock +150% over six months.

Vertiv (VRT)

Thermal management and cooling. AI chips run hot, and traditional air cooling doesn't cut it. Liquid cooling and immersion cooling are the answer, and VRT leads.

Comfort Systems (FIX)

Mechanical and electrical contracting for data centers. Unglamorous, but every new data center needs it. Bread-and-butter execution layer.

3. Raw Materials - There's a Second Entry Point

Building a grid ultimately means aluminum (transmission conductors) and copper (wiring, transformers).

Century Aluminum (CENX) is up 300% over the last year. But Alcoa, in the same aluminum sector, has gone essentially nowhere YTD. Same macro thesis, one stock rips, the other sits — in my view, that's exactly what a "second entry point" looks like.

Layer on Trump's tariffs on foreign metals. If America rebuilds the grid with American labor and American materials, domestic aluminum producers become not just an infrastructure play but a tariff play as well.

Side-by-Side

Ticker	Bucket	6-month return	Core moat
PWR	Contractor	+43%	$44B backlog
MTZ	Contractor	Variable	Power + gas dual
AGX	Contractor	Strong	Power plant EPC
VICR	Component	Strong	Power module tech
POWL	Component	+150%	Switchgear oligopoly
VRT	Component	Strong	AI cooling leader

How to Choose

I'm not saying buy all six. The point is consistency in your selection criteria:

• Want near-term revenue visibility? Big-backlog contractors (PWR, MTZ)
• Want technical moats? Oligopolistic component makers (VICR, POWL, VRT)
• Want a later entry? Less-extended names in the same sector (Alcoa-style aluminum laggards)

One rule across all cases — wait for next quarter's backlog prints. If backlogs keep growing, the trend is alive. If backlog growth rolls over, that's the signal to trim.

Closing Thought

I'd rather arrive at the party a little late than too early. The total return is smaller, but risk-adjusted return is cleaner. Infrastructure cycles typically run 5-10 years. With $1.4T deploying over that horizon, being six months late doesn't end the trade.