Powell Industries ($POWL)) makes medium voltage bespoke switchgear and packaged power solutions and is HQ'd in Houston, has ~3k employees, and is ~80 years old.
The Thesis
I believe $POWL — and I've taken a position as such at $20 and then again at $350 — is set to become a $100B–$200B market cap company in 10–20 years in such an odd way.
I might buy again at $750, fingers crossed.
Has ballooned to 45% of my portfolio, that too way above where I thought it would be.
It's currently a $6.6B market cap company.
Competitive Moat: A Western Monopoly
I do believe $POWL can hold its customers hostage.
100% domestic manufacturing out of Houston — breakers, enclosures, bus bars, integration, all under one roof. No other company on earth does this for bespoke MV systems. To pivot into it is annoying for even the majors because it represents a small % of their total market (Eaton, Hitachi, Siemens, etc.).
Why the Competition Is Weak
The only potential competition I've heard of is weak.
Lucy Electric is a UK company that makes standardized distribution switches for utilities, not custom Power Control Rooms for LNG terminals and data centers. Zero U.S. manufacturing, ineligible for Buy American.
TGOOD is the closest global analog (7,000+ prefab substations/year), but faces 20–47%+ U.S. tariffs, holds no ANSI/UL/IEEE certifications, and no U.S. operator will put Chinese switchgear in critical infrastructure.
Neither makes their own breakers, neither has U.S. installed base. Powell remains a category of one domestically. I believe they are a western monopoly on bespoke medium voltage PPS (packaged power solutions) and switchgear.
Will wait and see how this plays out.
I exited Cloudflare at $190 and $220 to take this position.
We may very well be in an era of 30+ PE Ratio industrial companies. Time will tell.
When You Don't Need Powell
You don't need Powell when you're wiring up a house, a restaurant, an office building, or even a small warehouse. You call a regular electrician, they go to a supply house (there are many), buy standard parts off the shelf, and wire it up.
Done in a few weeks.
This is like buying furniture from any store — pick it from the catalog (Eaton, Hitachi Energy, Siemens, etc.), assemble it, move on.
When You DO Need Powell
You're building something where if the power goes out, bad things happen.
Examples:
A data center where Amazon or Google stores your photos, Netflix shows, and banking data. If the power flickers for even a fraction of a second, millions of people lose access and the company loses millions of dollars per hour.
An LNG terminal where natural gas is being cooled to -260°F and loaded onto ships. If the power system fails, you've got a potential explosion risk and a $10 billion facility sitting idle.
An offshore oil platform in the middle of the Gulf of Mexico during hurricane season — the power system has to work perfectly, in a steel box, on a platform that might be rocking in 30-foot waves.
A hospital or transit system where people's lives literally depend on the power staying on.
For all of these — buying from catalog doesn't work practically. The power system has to be custom-designed or bespoke for your specific safety requirements, project, and environment. Hundreds of engineering decisions all work together.
Powell infra has high complexity and supply chain orchestration requirements.
The Customer Journey: When Powell Becomes Relevant
Powell infra has high complexity and supply chain orchestration requirements.
This is an example in the customer journey when the moment Powell Industries becomes relevant.
The engineering team might say something along the lines of: "We need 15,000 volts distributed across 30 different circuits, with arc-resistant housing rated for hurricane winds, explosion-proof certification, and the whole thing needs to show up factory-tested and ready to plug in — because we can't afford to spend 18 months debugging it on-site."
There are really only two options at that point:
Option A: Piece It Together Yourself
Hire a team of engineers to design the system, then buy the circuit breakers from ABB in Germany, enclosures from a fabricator in Mexico, controls from Siemens in another country, bus bars from someone else, ship it all to your job site, and pray it all works together. If something goes wrong, everybody points fingers at everybody else. This takes 3–4 years.
Option B: Call Powell
They design it, build the breakers, build the enclosures, wire it all together, test the entire thing as one unit in Houston, put it on a truck (or a barge), and ship it ready to plug in. One phone number to call if anything breaks. One warranty. 12–18 months.
Powell becomes relevant the moment the electricity problem is too complex, too dangerous, or too mission-critical for off-the-shelf parts.
And right now, almost every big thing being built in America — data centers, LNG plants, grid upgrades, renewable energy farms — has exactly that kind of electricity problem.
Demand Drivers
Data centers: Somewhere between 50–100 GW of new capacity is expected to come online by 2030, potentially doubling the sector's size — each MW requiring custom MV power distribution (JLL).
LNG plants: U.S. LNG export capacity is on track to roughly double from ~11–15 Bcf/d to potentially 25–29 Bcf/d by 2029, with every new train needing bespoke MV power systems (U.S. EIA).
Grid upgrades: Estimates suggest $1–$1.4 trillion in U.S. power sector investment from 2025 to 2030, roughly double what was spent in the prior decade (Deloitte).
Aging infrastructure: Roughly 60–70% of power transformers and circuit breakers are past their intended useful life, and the ASCE gave U.S. energy infrastructure a D+ grade (Veckta).
Regulatory Tailwinds
Three major U.S. laws — the IIJA ($1.2 trillion), IRA ($370 billion), and CHIPS Act ($280 billion) — all require or financially incentivize the use of domestically manufactured equipment in infrastructure, energy, and semiconductor projects, with domestic content thresholds ratcheting up to 55% by 2027 and penalties for non-compliance getting progressively more severe.
Powell manufactures everything — breakers, enclosures, bus bars, complete PCRs (Power Control Rooms) — in Houston, making them 100% compliant by default, while competitors like ABB, Siemens, and Schneider manufacture primary components overseas and face growing audit risk around the "manufactured in America" vs. "assembled in America" distinction.
The Predictive Maintenance and Supply Chain Angle - How to view the Remsdaq Acquisition
When I was at Georgia Tech about two decades ago, I had a professor who used to step outside military bases and do vibration testing on equipment, and with a 96% confidence interval he could tell you the likelihood of a part having an outage in 1, 3, 5, and 10 years and how much to stock up. Very smart professor — need to go find his info. I mention this because I think Powell's Remsdaq acquisition enables something similar at scale across their entire installed base.
Remsdaq essentially creates years of baseline temperature trends, partial discharge data, and vibration signatures that enable predictive maintenance. Examples of what I mean, given to me by some engineers I've met in the industry:
A partial discharge trend from 10 pC to 50 pC over six months means insulation is actively degrading — you have maybe 6–12 months before catastrophic failure (pC = picocoulomb).
A bus bar drifting from 63°C to 78°C with no load change means the bolted joint is loosening and contact resistance is climbing.
A breaker that opened in 48 ms when new but now takes 58 ms is wearing toward the point where it won't clear a fault fast enough.
A vibration spike from 1.8 g to 2.4 g over two months means something mechanical shifted — find it during a planned inspection, not a real emergency.
Every one of these data points is a future Powell work order — breaker replacement, retorque service call, a retrofit upgrade , etc.... — that Powell can see before the customer calls.
The Strategic Value
If Powell has visibility into the degradation curves across its entire installed base, they can forecast demand for replacement breakers, spare parts, and field service calls with much better accuracy than waiting for reactive orders. That's not just a customer lock-in play — it's a supply chain planning advantage that should make Powell progressively more profitable over time as the monitoring footprint grows.
My best guess is that the market is valuing Remsdaq as a small tuck-in acquisition, when the real value is the operational intelligence layer it creates across Powell's entire business. I could be totally wrong — but the logic makes sense to me and I haven't heard a better explanation.
