We ran every actively traded common stock on the NYSE and NASDAQ above twenty billion dollars through three filters at once. Not a scoring model, not a ranking — three hard conditions that a company either meets or does not. Six passed.
The three tests
Each one is deliberately unremarkable on its own. Plenty of companies clear any single hurdle. The point is the intersection.
- Return on capital employed above 25% — the business earns a genuinely high return on the money invested in it.
- Debt to equity below 0.25 — those returns are not manufactured with borrowing.
- Revenue growth above 15% — and the business is still getting bigger.
We also excluded ETFs, funds, ADRs, preferred shares, and non-dollar listings, and de-duplicated dual share classes. That cleanup matters more than it sounds — a naive version of this screen returns preferred shares with nonsensical market caps and the same company twice.
The six
- NVIDIA — 82.4% ROCE, 85.3% revenue growth, 52.4% margin. Score 96.
- Alphabet — 29.3% ROCE, 20.8% revenue growth, debt to equity 0.10. Score 79.
- Arista Networks — 27.5% ROCE, 33.0% revenue growth, and effectively zero debt. Score 94.
- Texas Pacific Land — 41.8% ROCE, 62.2% net margin, no debt. Score 80.
- Comfort Systems USA — 39.7% ROCE, 47.5% revenue growth, 9.0% margin. Score 83.
- EMCOR Group — 37.1% ROCE, 21.5% revenue growth, 7.1% margin. Score 65.
Two of these are not technology companies
Comfort Systems and EMCOR install and service mechanical, electrical, and plumbing systems in commercial buildings. They are contractors. They earn single-digit net margins — 9.0% and 7.1% — and they clear a returns hurdle that almost every software company in America fails.
This is the most useful thing on the list. A 7% margin business earning 37% on capital is turning that capital over roughly five times a year. Margin tells you what a company keeps per sale; returns tell you how hard the money works. Confusing the two is how investors end up believing high margins and high quality are the same thing.
It is also worth noticing what they build. Both are exposed to datacenter construction, which is the same spending wave showing up as falling free cash flow at Microsoft and Alphabet. One company's capital expenditure is another's revenue.
What the absences tell you
Consider who is not here. No Apple, no Microsoft, no Meta, no Amazon, no Broadcom, no TSMC. Most fail on growth — 15% is a high bar at multi-trillion-dollar scale. Some fail on leverage. Meta and TSMC come close enough that a slightly looser screen would include them.
And a screen is not a buy list. Nothing above filters on price. Nvidia and Arista both carry demanding multiples; Texas Pacific Land is a land and royalty company whose economics do not resemble the other five at all. Six names passing three tests means exactly that, and no more.
The bottom line
Out of roughly seven hundred US-listed companies above twenty billion dollars, six earn more than 25% on their capital, carry almost no debt, and are still growing revenue at 15%+. That is under one percent of the large-cap market.
The scarcity is the finding. Companies that combine high returns, low debt, and real growth are rare enough that you can list them, and that is precisely why the market rarely offers them cheaply. If you want the reasoning behind the first filter, ROCE explained covers why we weight it above almost everything else.