Class 1 railroads are the six largest freight railroads in the US and Canada, defined by annual revenue above roughly $900 million (a threshold set by the Surface Transportation Board and adjusted for inflation). They own the track, terminals, and locomotives that move roughly 40% of US long-distance freight by ton-miles — bulk commodities (coal, grain, chemicals), intermodal containers, and finished goods. The industry generated roughly $80 billion in combined revenue in 2024 across all six carriers est.. Four are publicly traded under the tickers covered here; the other two are BNSF (wholly owned by Berkshire Hathaway, not publicly traded) and CN (Canadian National, ticker CNI, US-listed but not included in this group). A pending UNP–NSC merger, if approved, would reduce the six to five and create the first US coast-to-coast railroad.
These four companies own physical infrastructure — steel rail, land rights-of-way, bridges, and yards — that cannot be replicated. No new Class 1 railroad has been built in the US in over a century. Money comes in as freight revenue per carload; money goes out as crew labor, fuel, equipment maintenance, and track upkeep, captured in a single metric called the operating ratio (operating expenses as a percentage of revenue — lower is better). The industry-wide OR runs 58–66%, meaning 34–42 cents of every revenue dollar becomes operating income.
The product is freight transportation by rail, measured in two units. A carload is one railcar of freight (coal, grain, chemicals, lumber, autos). An intermodal unit is a container or trailer riding on a flatcar — these carry consumer goods, retail inventory, and imports from ports. Revenue ton-miles (RTM) captures both weight and distance: one ton of freight moved one mile = one revenue ton-mile.
Railroads charge shippers a rate per carload or per container. Rates are set through two channels: contract rates (multi-year agreements with large shippers, often indexed to inflation) and tariff/spot rates (published price lists, adjustable). On top of the base rate, railroads add a fuel surcharge that floats with diesel prices, partially hedging their largest variable cost. Total freight revenue = (rate per unit + fuel surcharge) × units shipped.
The cost structure is dominated by four items: (1) compensation and benefits (~25–30% of revenue — train crews, maintenance workers, dispatchers), (2) fuel (~10–15%, varies with diesel price), (3) depreciation (~10–12% — the track, bridges, and locomotives wear out), and (4) purchased services and materials (~10–15% — outsourced track work, parts). Capital expenditure runs 14–18% of revenue annually, spent on track maintenance, new locomotives, and terminal upgrades. est.
The operating ratio (OR) = total operating expenses ÷ total operating revenue. An OR of 60% means 40 cents of every dollar is operating income. The six Class 1s have driven ORs from the mid-70s in 2010 to the high-50s/low-60s today through "precision scheduled railroading" (PSR) — operating practices that reduce dwell time, increase train length, and eliminate unnecessary switching.
Cash returns to shareholders come via dividends (all four pay, yielding 0.7–2.1%) and share buybacks (UNP is the most aggressive, targeting $4.0–4.5B in 2025). Railroads carry substantial debt — the assets are long-lived and generate predictable cash flow, so the balance sheets are leveraged 2–3× debt-to-EBITDA.
Source: UNP, CSX, NSC, CP 2024 10-K filings and earnings releases; AAR Freight Rail Facts & Figures.
Rail freight demand tracks industrial production, GDP growth, agricultural output, and international trade (imports through ports ride intermodal to inland destinations). US Class 1 carloads have been roughly flat to declining for a decade, offset by intermodal growth. In 2024, total US carloads were 11.34 million, down 2.9% year-over-year, driven entirely by coal's secular decline. Excluding coal, carloads rose 1.4%. Intermodal units hit 13.84 million, up 9.3% — the third-highest level ever recorded — driven by international container imports through West Coast ports.
Source: AAR via Calculated Risk, Jan 2025.
Contracted: A significant portion of railroad revenue is under multi-year contracts with large industrial shippers (utilities buying coal transport, chemical companies, grain elevators). These contracts typically have volume commitments and rate escalators tied to the Railroad Cost Adjustment Factor (RCAF) published by the STB. The exact share of contracted vs. spot revenue is not disclosed uniformly, but railroads consistently describe their book as "predominantly contracted" in filings. contracted
Forecast: CPKC guides for "mid-single digit volume growth, as measured in Revenue Ton-Miles" for 2025. UNP's management noted potential headwinds from "a mixed economic backdrop, coal demand, and challenging year-over-year international intermodal comparisons." Coal — which was ~16–20% of carloads a decade ago est. — continues its structural decline as power plants retire; the AAR data shows carload declines are entirely attributable to coal. Intermodal is the growth engine, driven by truck-to-rail conversion (rail is 3–4× more fuel-efficient than trucking per ton-mile) and port volumes.
AI benefits railroads through operational efficiency rather than demand creation. AI-optimized train scheduling (reducing dwell time and improving velocity), predictive maintenance on track and rolling stock, autonomous train operations (reducing crew costs), and dynamic pricing. Each 1% improvement in operating ratio at a Class 1 railroad translates to hundreds of millions in profit est.. Rail demand itself is not directly driven by AI build-out the way copper or power demand is — data centers need electricity and metals, not carloads of bulk freight. The connection is indirect: if AI accelerates GDP growth, freight volumes follow; if AI disrupts trucking (autonomous trucks), intermodal could face competition from cheaper long-haul trucking.
Source: 500-stocks sector scan, "Class 1 Railroads" section; CPKC Q4 2024 earnings release; UNP Q4 2024 earnings release.
The US freight rail network is ~140,000 route-miles (AAR). No new Class 1 railroad has been built; right-of-way acquisition through developed land is effectively impossible. Existing railroads can increase throughput by running longer trains (some now exceed 3 miles), double-stacking intermodal containers, adding sidings for passing, and reducing terminal dwell time. But adding entirely new routes or corridors is not realistic — the last major new rail corridor built was CPKC's connection of CP and KCS networks through the 2023 merger.
Railroads do not disclose a single "capacity utilization" number. The binding constraints are network fluidity (how fast trains move through the system — measured by average train speed and terminal dwell time) and crew availability. After aggressive headcount cuts during PSR implementation (2017–2022), several carriers experienced service crises from too-few crews. Railroads have since rebuilt crew pipelines. In Q1 2025, Oliver Wyman reported that UNP maintained the fastest velocity, while CSX experienced "significant congestion" from weather and the Howard Street Tunnel reconstruction in Baltimore.
Source: AAR Freight Rail Facts & Figures; Oliver Wyman Q1 2025 Class I Performance report; UNP Q4 2024 earnings release; equipment and track costs from general industry knowledge.
Unlike copper or semiconductors, railroads do not face a dramatic demand-supply imbalance in either direction:
| Factor | Demand side | Supply side |
|---|---|---|
| Current level | 11.3M carloads + 13.8M intermodal units (2024) | ~140,000 route-miles, not fully utilized |
| Direction | Flat to low-single-digit growth (coal declining, intermodal growing) | Fixed network; capacity added at margins through operational efficiency |
| Speed of change | Slow — tracks GDP/industrial production | Very slow — network expansion nearly impossible |
| Pricing implication | Railroads consistently price above inflation (RCAF escalators) | Oligopoly of 6 carriers with no new entrants possible |
| Key risk to demand | Autonomous trucks making long-haul trucking cheaper; coal retirement | Service failures (derailments, congestion) shift freight to trucks |
Pricing direction: Inflation-adjusted revenue per ton-mile has declined 44% since deregulation in 1981 (AAR) — but this reflects efficiency gains passed to shippers, not weak pricing. In nominal terms, revenue per carload has risen steadily, and railroads routinely price contracts at or above inflation. The oligopoly structure (6 carriers, with many routes served by only 1 or 2) gives them leverage. The STB regulates rates for "captive shippers" (those with access to only one railroad) but does not set prices broadly.
Slow-growth, high-margin, irreplaceable-asset business where supply is permanently constrained and demand grows with the economy. No acute shortage or glut — the cash flows are durable and predictable because the asset base cannot be replicated.
Source: AAR (rate data, network facts); sector scan; Oliver Wyman Q1 2025 report.
Four publicly traded Class 1 railroads, plus context on the two that are not investable individually. All financial data is FY2024 from 10-K filings and earnings releases unless noted.
| Ticker | Network | Revenue | OR | Net income | FCF est. | Capex | Total debt |
|---|---|---|---|---|---|---|---|
| UNP | Western 2/3 of US (23 states) | $24.3B | 59.9% | $6.7B | ~$5.9B | $3.5B | $31.2B |
| CSX | Eastern US (23 states, DC) | $14.5B | ~63.9% | $3.5B | ~$2.7B | $2.5B | $18.5B |
| NSC | Eastern US (22 states, DC) | $12.1B | 66.4% | $2.6B | ~$0.6B | $3.5B | $17.2B |
| CP | Canada coast-to-coast + US (KCS, Gulf Coast, Mexico) | $14.5B | 64.4% | $3.7B | ~$2.4B | $2.9B | $22.6B |
| Not individually investable: BNSF (Berkshire Hathaway subsidiary, ~$23B revenue est.) and CN/CNI (Canadian National, US-listed but not in this group) | |||||||
Source: UNP Q4 2024 earnings release; CSX 2024 10-K (MarketBeat); NSC Q4 2024 earnings release (Railpace); CPKC Q4 2024 earnings release (Stock Titan); Oliver Wyman Q1 2025 report.
What an owner pays today for $1 of these companies' earnings or cash flow. All prices as of market close June 3, 2026.
| Ticker | Price | Mkt cap | P/E | EV/EBITDA | Div yield | Debt/EBITDA |
|---|---|---|---|---|---|---|
| UNP | $262 | $156B | ~21.6× | ~13.3× | 2.1% | 2.5× |
| CSX | $46 | $86B | ~25× est. | ~15× est. | ~1.0% | ~2.7× est. |
| NSC | $305 | $69B | ~26× est. | ~14× est. | ~1.8% | ~2.8× est. |
| CP | $89 | $79B | ~27× | ~18× est. | 0.7% | ~3.0× est. |
Money in: An owner buying UNP at $262 is paying roughly $156B for a business that generated $6.7B in net income and ~$5.9B in free cash flow in 2024 — about 23× FCF est.. UNP returns cash through dividends ($5.28/share, 2.1% yield) and buybacks ($1.5B in 2024, guided to $4.0–4.5B in 2025). At the guided buyback pace, UNP would retire ~1.7% of its shares per year at current prices. est.
Money out: These are leveraged equity stubs sitting atop large debt loads. UNP carries $31B in debt against $12.5B in EBITDA (2.5× leverage). The debt is long-dated and investment-grade (all four are rated BBB+ to A− by S&P). Interest expense at UNP was $1.3B in 2024 — about 14% of operating cash flow est.. A severe recession that cut volumes 15–20% would compress FCF substantially while debt service remains fixed.
Deal terms (announced July 28, 2025): Each NSC share receives 1.0 UNP share + $88.82 cash. At UNP's current $262 price, that implies ~$351 per NSC share. NSC closed at $305 on June 3, 2026, a ~15% discount to the implied deal value — reflecting regulatory risk (STB review) and time value (expected close by early 2027). Shareholders approved the deal with ~99% of votes. The combined entity would have ~$36B in revenue and cover the entire US from coast to coast. The $2.5B termination fee is payable if the deal is blocked. contracted
NSC at $305 is primarily a merger-arbitrage position: the question is whether the STB approves the deal. If the deal closes, NSC shareholders receive ~$351 in value (at current UNP price). If it breaks, NSC would likely trade back toward its pre-deal unaffected range of ~$230–250. est.
Source: StockAnalysis.com (prices, market caps, Jun 3 2026); UNP/CSX/NSC/CP 2024 10-K and earnings releases (financial data); UNP S-4 filing (merger terms); MarketBeat (UNP detailed financials).
Source: sector scan; UNP S-4 filing; Oliver Wyman Q1 2025.
What was used:
Hard vs approximate: