This is private exploration and general reflection, not financial, investment, tax, or legal advice.
The question I'm actually trying to answer is not just "own more physical-world businesses." It's closer to this: if someone already thinks, works, and probably invests in a software-heavy world, what does a serious hedge look like if the goal is to change the dependency stack instead of buying another digital story with different branding?
I start with the obvious idea: buy or start an HVAC company, maybe plumbing, maybe electrical. The problem is that direct ownership solves one issue by loading you up with several others. You reduce software exposure, but you add operator risk, local-market risk, leverage risk, technician retention risk, and a huge amount of execution dependence on a business you may not know how to run. That is not diversification. That is a transfer of concentration from one domain to another.
So the thesis shifts. Instead of asking which trade business to own, I start asking which parts of the essential-services ecosystem get paid across repair cycles, compliance cycles, route density, and building maintenance whether or not AI makes one more category of white-collar work cheaper.
Part of what keeps this lane interesting is that the physical labor underneath it is still real and still needed. The Bureau of Labor Statistics still projects 8% employment growth for HVAC mechanics from 2024 to 2034, 9% for electricians, and 4% for plumbers, pipefitters, and steamfitters. At the same time, the Department of Energy says data centers could rise from 4% of total U.S. load in 2023 to as much as 9% of annual electricity generation by 2030. That does not mean every related stock is automatically attractive. It does mean the "physical layer still matters" part of the thesis is not imaginary.
This is a sleeve, not a total portfolio
I'm not trying to design a 100% allocation for somebody's life savings. I'm trying to sketch a thematic sleeve that can sit next to broader equity exposure and do a different job. The job is not maximum upside. The job is to hold businesses that still have to install, maintain, inspect, route, repair, or finance the built environment if software gets cheaper and more abundant.
That matters because it changes the standard. I'm not looking for the single best stock in isolation. I'm looking for a basket where each name plays a role: distribution, field execution, regulated service, route density, or credit. The weights are not meant to look like an optimizer output. They're a rough expression of function, conviction, and how much duplication I'm willing to tolerate inside the sleeve.
The 10-name version I would start with
If I force myself to cut the idea down to 10 names, this is the version I start from:
| Ticker | Weight | Role in the sleeve | Why it made the cut |
|---|---|---|---|
| WSO | 12% | HVAC distribution | Lets me own replacement and repair demand without having to underwrite one local contractor. |
| FERG | 10% | Plumbing and HVAC distribution | Adds another picks-and-shovels layer across plumbing, HVAC, and adjacent maintenance categories. |
| FIX | 11% | Mechanical systems execution | Gives direct exposure to commercial HVAC and mechanical work where complexity still matters. |
| EME | 11% | Electrical and mechanical contracting | Adds electrical intensity, industrial exposure, and more direct participation in capex-heavy physical systems. |
| APG | 10% | Fire and life safety | One of the cleanest compliance-and-inspection businesses in the group. |
| OTIS | 10% | Elevator service and modernization | Very similar to what I like in HVAC, but with even more recurring maintenance and code sensitivity. |
| WM | 9% | Waste route density | Essential local-service economics with pricing power and very hard-to-ignore demand. |
| ROL | 8% | Pest-control recurring service | Route density, repeat service, and low direct AI displacement risk. |
| CTAS | 7% | Uniforms, facility services, and compliance | A little broader than the pure trades thesis, but still tied to recurring physical operations. |
| BXSL | 12% | Private-credit proxy | Keeps one slot for the lending layer so the sleeve is not only public-equity operating exposure. |
That exact list is the point where the idea stops being a vibe and starts becoming a portfolio. Not a finished portfolio. Not a backtested one. But at least a real construction with actual tradeoffs.
Why the first 42% goes into the service infrastructure itself
The center of gravity in this sleeve is WSO, FERG, FIX, EME, APG, and OTIS. That is deliberate. I want the portfolio to lean hardest into businesses that either supply or directly maintain the systems buildings cannot ignore.
WSO and FERG are the least romantic names in the bunch, which is part of why I like them. They let you own a broad stream of repair and replacement demand without pretending you know which local contractor is best run in Phoenix, Dallas, or Tampa. They are the picks-and-shovels version of the thesis.
FIX and EME are a different kind of exposure. They are closer to the actual work getting done. That makes them more cyclical and more execution-sensitive, but it also gives the sleeve a cleaner connection to HVAC, electrical, and mechanical complexity in the real world. If AI drives more power demand, more cooling demand, and more physical infrastructure buildout, that pressure shows up here more directly than it does in the route businesses.
APG and OTIS are probably the most structurally attractive names in the whole sketch. Fire and life safety, inspections, elevator maintenance, modernization, and related service work all sit near the sweet spot I keep looking for: recurring, compliance-linked, field-heavy, and difficult to wish away. They do not depend on the market deciding software deserves a premium multiple forever. They depend on buildings continuing to function.
Why I still want route-density businesses in the mix
The second layer was about behavior, not just theme. WM, ROL, and CTAS are there because they should not fail for the same reasons as the mechanical and electrical names.
Waste collection, pest control, uniforms, first-aid replenishment, and facility-service routes are not identical businesses, but they share a pattern I want in the sleeve: recurring service, operational density, and a lot of local-world friction that software alone does not erase. These are not glamorous businesses. That is a feature, not a bug.
I also like that this group dampens the portfolio's tendency to become a disguised "construction boom" thesis. FIX and EME can benefit from physical buildout. WM, ROL, and CTAS are more about ongoing operating reality. If the economic cycle gets ugly, I would rather have some exposure to services people keep paying for because the trash still has to move, pests still show up, and workplaces still need supplies and compliance support.
Why BXSL stays even after I cut the rest of the credit sleeve
BXSL is the odd name in the final 10, and I think it deserves an explicit defense. In the broader version of the idea, I'm also looking at ARCC, OBDC, and MAIN as ways to get yield and senior-secured lending exposure without becoming an operator. When I force the sleeve down to 10 names, I do not want four versions of the same credit idea. I want one representative slot for it.
BXSL stays because I still like the role. A public-equity sleeve can give you business ownership exposure, but lending exposure changes the return shape. It adds income, it adds a different downside profile, and it gives at least some participation in middle-market operating businesses without asking you to guess which contractor platform deserves 8% of the whole sleeve.
The tradeoff is that BXSL is also the least pure expression of the thesis. It is not a direct "essential maintenance" business. It is a credit vehicle that may have some overlap with that world while also taking broader middle-market risk. I keep it anyway because I do not want the whole portfolio to depend on multiple expansion and operating execution alone.
What got cut, and why
The broader sketch has more names in it. Cutting down to 10 forces the more honest conversation.
| Name | Why it was interesting | Why it did not survive the final cut |
|---|---|---|
| RSG and WCN | Good waste businesses with similar route-density logic to WM. | I did not need three waste names in a 10-stock sleeve, and WM felt like the cleanest single anchor. |
| SCI | Probably the most literal non-discretionary demand business in the wider set. | It added stability, but it weakened the "maintenance and infrastructure" center of the thesis. |
| PAVE | Useful way to express grid and infrastructure tailwinds. | Too broad for a final 10. It diluted the sleeve away from service-heavy businesses into a looser industrial-capex bet. |
| ARCC, OBDC, and MAIN | All were reasonable income-oriented credit proxies. | Keeping all of them would have turned one tactical credit sleeve into a quarter of the portfolio by itself. |
This is also where the portfolio gets more opinionated. A lot of drafts sound smart because they never force a cut. Once you force a cut, you have to admit what you are actually prioritizing. I am prioritizing service infrastructure, recurring physical maintenance, and one controlled dose of credit. I'm not trying to own every respectable "boring business" at once.
The real tradeoffs are not subtle
The first tradeoff is valuation. A lot of these businesses are attractive for exactly the reasons everybody else likes them: recurring demand, pricing power, and durability. That means the market often knows what it owns. A thesis can be right and still produce mediocre returns if the starting multiple already prices in too much perfection.
The second tradeoff is that "essential" does not mean "immune." FIX and EME still have project exposure. Distributors still feel inventory cycles. Waste and route-density businesses can still get hit by cost inflation, labor shortages, or overenthusiastic acquisition behavior. BXSL can still get hurt if credit quality weakens. There is no magic here. There is only a different failure stack.
The third tradeoff is purity versus resilience. A purer sleeve would probably drop BXSL and maybe even CTAS, then reallocate harder into the core service names. A more defensive sleeve might add back another waste name, bring in SCI, or reduce the contractors. There is no single correct answer. The whole point is being explicit about what job each name is doing.
What I want to backtest before taking the weights seriously
I don't pretend these allocations are optimized. They're not. They're a first serious map. If I come back to this later, I want to test at least five things:
First, I would compare the sleeve's correlation, drawdown behavior, and upside capture against SPY, QQQ, and a software-heavy benchmark instead of assuming the diversification is real just because the stories sound different.
Second, I would test equal weight versus these role-based weights. The current mix reflects judgment, not math. Sometimes that is fine. Sometimes it hides overconfidence.
Third, I would isolate the credit decision. Does BXSL genuinely improve the sleeve's behavior, or does it just add yield-shaped risk that looks comforting until a credit cycle turns?
Fourth, I would separate service-heavy revenue exposure from project-heavy revenue exposure inside the contractors. The portfolio probably behaves very differently depending on how much of the economic engine is maintenance versus new build.
Fifth, I would spend more time on valuation entry ranges. This whole idea works better if the discipline is "own the maintenance layer of civilization when the price is sensible," not "pay anything for quality because AI sounds scary."
What still holds my attention about the thesis
The reason this still holds my attention is that it is not just an investing thought. It is a broader systems thought. When risk is concentrated, the right answer is often not to get fancier inside the same dependency chain. It is to change the chain.
That is what this sleeve is trying to do. Not predict the end of software. Not declare an "AI apocalypse" with cinematic certainty. Just ask a harder portfolio question: if software becomes cheaper, noisier, and less scarce, what parts of the economy still have to show up in the real world and do the work?
My current answer is still the same one I keep circling in the source conversation: I would rather start with HVAC distribution, electrical and mechanical execution, fire and life safety, elevator maintenance, waste routes, pest routes, facility services, and one controlled credit sleeve than with another layer of software-adjacent optimism. Even if I later change the exact names, I think that dependency shift is the right starting point.