CSX: A Great Railroad, Priced Like a Lottery Ticket
The business and its moat
Some moats require faith. This one requires only that freight in the eastern United States continues to roll on rails — which, however AI reshapes logistics over the next decade, remains a near-certainty. CSX operates a vast route network across 23 states, one half of an eastern duopoly. Building a competing rail network is, for practical purposes, impossible: physically and regulatorily, the door is closed.
The pricing record backs this up. CSX has pushed rate increases above inflation for twenty years. The 32.1% operating margin posted in 2025 reflects weaker volumes — coal declines and degraded service — not a collapse in pricing power. Deterioration of that kind is a weather problem, not a climate one.
Earning power and capital allocation
The five-year average ROE is 28.3%, but it arrives with roughly $19 billion of debt on the books (D/E around 1.4). Pretty numbers deserve an audit of their ingredients; this one is partly borrowed.
Annual owner earnings — net income plus depreciation minus all capex — land close to the company’s free cash flow of $1.7–1.8 billion, which is a fair approximation given that most railroad capex is maintenance rather than growth. On normalization, the estimate rises to $2.5–3.0 billion.
Capital allocation is steady on the home turf: buybacks and rising dividends, compounded patiently. It is wobblier at the edges. The Quality Carriers acquisition failed decisively, taking a $164 million goodwill impairment in its third period. Neighboring businesses are evidently not where the skill lies.
Valuation and margin of safety
The last close was $51.54, a P/E of 30.5x — a price that leaves little to the imagination and much to hope. Wall Street is captivated: “Volume Growth Is Finally Reaching Earnings” is representative of analyst thinking, with all three pieces reviewed dwelling on near-term volume recovery and the prospect of riding merger speculation.
Compare that enthusiasm to arithmetic. The estimated owner-earnings yield is 1.8–3.3%, against the 10-year US Treasury at 4.67%. A theoretical buy price with a 30% margin of safety would be a market cap of $36 billion or below — roughly $18–19 per share. Even a generous upper-bound model of $4.5 billion in owner earnings falls far short of justifying the current price.
The quality is real. The price is not.
Word for today
Buy a superb rail network at a price with no value in it, and you are merely a passenger. I will skip the ticket and keep reading the timetable.
Verdict: pass. Wall Street is buying volume recovery and merger speculation; I compare durable owner earnings to a riskless bond and find the yield badly wanting. The business is genuine — the multiple, stacked with takeover premium, is what isn’t.
Correction (2026-09-03): market cap and owner-earnings yield restated at the price on the verdict date ($51.30 on 2026-08-27; the 2026-07-29 screening snapshot adjusted for price): market cap ≈ $95.0 billion (the article said $36.0 billion), owner-earnings yield ≈ 1.75% (the article said 3.3%) vs the 10-year Treasury at 4.67% — verdict unchanged.
This analysis is AI-generated, educational, and not investment advice. Figures may contain errors or be delayed. Disclaimer