CLF: The Recovery Is Real. Can It Pay Down the Debt?

The 60-second read

Five things to know

1. The recovery is profitable under GAAP. Operating income reached $311M.
2. Cash has not accelerated with EBITDA. Q3 FCF was $72M versus $73M in Q2.
3. Liquidity is not cash. The $3.2B liquidity headline includes financing capacity; cash was $77M.
4. Next year is conditional. The >$4B outlook depends on contracts and index-price forecasts.
5. HOLD at $13.80. Base value is $15; an illustrative 25% margin-of-safety entry is $11.25 if fundamentals hold.

Three scenarios, one probability-weighted number

Analyst scenarios, rounded per-share values; constant present net debt.

Signposts: what would change our mind

Green and red thresholds are our monitoring rules, except management’s stated leverage goal. Future release dates are unannounced; review dates are analyst checkpoints.

The tape

A strong print has already moved the price.

Cleveland-Cliffs reported before the market opened on October 19, 2026. The issuer timestamp is 6:00 a.m. Eastern, or 5:00 a.m. Central. The earnings call was scheduled separately for 8:30 a.m. Eastern. Our valuation snapshot uses the FMP quote of $13.80 at 10:04:59 a.m. Central. It is an intraday mark, not the closing price and not an execution recommendation. The daily-price endpoint was still changing during collection; the completed-session price chart therefore stops before this trading day.

The important change from our Q2 episode is the question being asked. That episode examined a return to positive cash flow and a jump in EBITDA. This quarter puts a stronger pricing recovery against a large financing burden and a much larger conditional outlook for next year. A positive operating turn can be real while the equity still offers an inadequate margin of safety after a sharp rally. We keep those two judgments separate.

The price-reaction series deliberately does not claim that earnings caused every move. Broader steel prices, interest rates, trade policy and market risk appetite move CLF too. Historical releases can occur before or after the close, and a simple release-date comparison does not adjust for those clocks. The current observation is especially provisional because the market session is incomplete. A later price can change the prospective return without changing any reported quarter figure.

The print

Pricing and volume improved; adjusted EPS beat the provider estimate.

Revenue was $5,612 million, against $4,734 million a year earlier and $5,226 million in Q2. The computed growth rates are 18.5% year over year and 7.4% sequentially. Operating income turned positive at $311 million. Consolidated net income was $165 million, but the amount attributable to Cliffs shareholders was $152 million after $13 million of noncontrolling interests. Those are different measures; neither should be silently substituted for the other.

FMP carried an adjusted EPS estimate of $0.214 against the reported $0.28: a computed 30.8% beat. Revenue of $5.612 billion was 1.5% below the provider consensus of $5.695 billion. That mixed surprise argues against reducing this release to a single beat-or-miss label. The EPS comparison is adjusted-to-adjusted; the GAAP $0.26 is not the numerator in that surprise calculation. Estimates are a provider snapshot and may have changed after publication.

Management had guided to $575 million of Q3 adjusted EBITDA and delivered $612 million, a $37 million or 6.4% improvement. The next guide is approximately $715 million for Q4, including the company’s stated effects of seasonality, maintenance, input costs and incentive compensation. Adding that guide to reported first-nine-month adjusted EBITDA implies about $1.708 billion for 2026. That is arithmetic based on management’s outlook, not a guaranteed full-year result.

The operating engine

One dominant steel business; use the actual segment bridge.

Steelmaking generated $5,447 million of revenue and $685 million of cash margin. Depreciation, depletion and amortization of $231 million leaves $454 million of segment gross margin. Cash margin is useful for understanding the manufacturing spread, but it excludes depreciation and does not pay corporate costs, financing costs or capital spending. Consolidated adjusted EBITDA is therefore not interchangeable with segment cash margin.

Realized selling price rose 14.1% year over year while shipments rose 3.0%. That mix helps explain why revenue and cash margin improved much faster than physical tons. It does not establish how much came from contract repricing versus product mix. The release gives market and product categories, not a full price-volume-mix reconciliation. We do not label the entire average-price increase a pure market-price benefit.

The current Steelmaking sales mix was 34% distributors and converters, 28% direct automotive, 28% infrastructure and manufacturing, and 10% steel producers. Direct automotive exposure makes contract resets important, while distributor behavior can magnify inventory swings. These categories describe market exposure, not the identities or concentration of individual customers. They do not prove that any single customer is diversified away.

Stelco improved enough for management to describe its strongest quarter since acquisition, while galvanizing lines are being idled because of Canadian oversupply. Management expects hot-rolled production to preserve total tonnage. That is a portfolio response with execution and employment consequences, not evidence that all Canadian markets have recovered. We credit the reported improvement and retain a separate risk for the remaining product-market imbalance.

Where profit goes

The operating recovery survives GAAP, but debt absorbs a large share.

The headline is not solely an adjusted-accounting recovery. GAAP operating income is positive, and consolidated net income is positive after interest and tax. Nevertheless, the $612 million adjusted EBITDA headline travels a long way before reaching common shareholders. Depreciation of $259 million and interest expense of $148 million are substantial. Treating EBITDA as cash available to the equity would ignore both the assets needed to make steel and the financing of those assets.

Below operating income, pension and other benefit credits added $64 million, derivative losses subtracted $15 million, and other nonoperating income added $1 million. Together with $148 million of interest expense, these produce pretax income of $213 million. The tax provision was $46 million, followed by a $2 million discontinued-operations loss. Pension credits improve accounting earnings but do not represent a customer payment collected this quarter.

For the first nine months, consolidated operating income was only $49 million and attributable net loss remained $230 million. That cumulative view matters when testing the sustainability of a $4 billion-plus outlook for next year. A recovery can accelerate, but the burden of proof increases with the distance between the latest run rate and the proposed destination. We require operating and cash evidence rather than capitalizing the destination immediately.

Earnings quality

A real recovery with modest cash conversion after investment.

The adjusted net-income gap is comparatively small: $161 million adjusted attributable income versus $152 million GAAP. The reconciliation removes losses and also removes gains, including the disposal gain. Showing only positive add-backs would exaggerate the adjustment. Our chart uses the contribution needed to move from GAAP attributable income to adjusted attributable income, so the disposal and tax items appear with negative signs.

Quarter operating cash flow of $224 million exceeded consolidated net income, but capital spending absorbed $152 million. The remaining $72 million was essentially flat against Q2’s $73 million even as adjusted EBITDA more than doubled. Receivables used $116 million and inventory used $170 million of cash, partly offset by $135 million from payables and accrued expenses. Higher sales can consume working capital before the cash arrives; the next quarter must demonstrate collection and inventory discipline.

First-nine-month operating cash was $129 million against capital spending of $461 million, leaving computed free cash flow of negative $332 million. Property sales can reduce debt, but sale proceeds are investing cash and are not operating free cash flow under our definition. This prevents a financing improvement from being mistaken for recurring cash generation. It also explains why we do not treat the liquidity headline as surplus cash.

The 2025 annual filing corrected immaterial employment-cost accrual errors in prior periods. We use the latest comparative figures where supplied and explicitly retain original-release presentation elsewhere. Corrected prior-quarter adjusted EBITDA is negative $179 million for Q1 2025 and positive $94 million for Q2 2025. The correction deserves attention, but the current filing evidence does not support labeling the issuer’s present statements nonreliable. No standalone quarterly SBC amount is manufactured from the provider’s zero field.

Balance sheet and capital

Available borrowing capacity is not cash in the bank.

The balance sheet contained $77 million of cash and $7,645 million of long-term debt. Our simple net-debt calculation is therefore $7,568 million. Management’s $3.2 billion total-liquidity measure includes available financing capacity; it cannot replace the $77 million cash balance in an equity-value calculation. The business has access to liquidity, but drawing that capacity creates or maintains a liability.

Net debt fell by a computed $65 million from Q2, a useful but modest move relative to the balance. The quarter’s operating-income-to-interest ratio was only 2.1 times. That is a narrow accounting coverage measure, not a covenant calculation and not a forecast of default. It shows why financing costs still deserve substantial space in an otherwise encouraging operating discussion.

Legacy pension, environmental and retirement obligations also matter. We do not simply add every gross liability to net debt while ignoring associated assets or cash-flow effects. Our normalized cash-flow assumptions and cautious multiples recognize recurring burdens, and the packet shows the reported balances so readers can stress them separately. The equity is sensitive to assumptions because a large fixed financing claim sits ahead of it.

Capital allocation has changed across the cycle. The company spent $733 million on repurchases in 2024, repurchased no common shares in 2025, and raised $951 million from common issuance in 2025. The annual report retained $1.4 billion of authorization, but authorization is neither an obligation nor evidence of current purchases. In this balance-sheet position, debt reduction and cash retention deserve priority over treating unused authorization as a near-term catalyst.

Valuation

Three routes cluster near $15; the wide scenario range stays visible.

Our HOLD is a valuation judgment, not a denial of the recovery. We use a $15 base value, conviction three out of five, and High uncertainty. The three routes below land between roughly $13.84 and $15.62. We round the central judgment to $15 rather than pretend model precision is price precision. The scenario range is wider, $6 to $27, because cyclical operating earnings and leverage create asymmetric equity outcomes.

At the fixed quote, enterprise value is $15.441 billion. At our selected 5.5-times normalized EBITDA multiple, that price implies approximately $2.808 billion of sustainable annual EBITDA. Q3 annualized is $2.448 billion, whereas management’s conditional 2027 outlook exceeds $4 billion. The market-implied figure sits between them. Annualizing one quarter is a reference point, not a forecast, and the company’s outlook is not assured merely because the implied figure is lower.

The cash-flow route is intentionally demanding. It requires operating recovery to turn into cash after maintenance investment and working capital. Its terminal value carries substantial weight, so the result is sensitive to discount rates and sustainable cash generation. We do not present it as an independent proof of the EBITDA model: all three routes share an economic recovery premise. Agreement among related assumptions reduces arithmetic inconsistency, not business uncertainty.

The EBITDA starting point follows the company’s adjusted measure, which excludes EBITDA attributable to noncontrolling interests. Our normalized earnings and cash-flow routes are framed on that common-economic basis too; we do not subtract a second minority claim without a consistent consolidated denominator. Actual future share issuance, acquisitions, debt reduction or asset-sale proceeds would require refreshing the model. Holding net debt constant makes the present comparison transparent.

A $15 base value offers only a computed 8.7% upside to $13.80, before fees, taxes or timing. We require a 25% discount to base value for a new BUY under High uncertainty, which gives an illustrative $11.25 entry threshold, provided the operating thesis remains intact. A falling price caused by deteriorating fundamentals does not automatically qualify. Better verified cash conversion or durable contract economics could instead justify raising the estimate.

Wall Street context

Fresh source times matter more than a target headline.

The provider target snapshot had a $12.07 consensus, $11.50 median, $9 low and $17.48 high. Our $15 base value is above the central Street marks, yet still does not create enough prospective return at the fixed quote for a BUY. These are observations with different publication dates and assumptions. A target distribution is neither a probability distribution nor a substitute for the operating model.

The gap between approximately $2.022 billion of provider 2027 EBITDA and management’s greater-than-$4 billion outlook is economically important, but its interpretation is limited. The provider field may use a different EBITDA definition and may not yet incorporate the release. We show the gap as a question to investigate, not as evidence that analysts are demonstrably wrong or that a wave of upgrades must follow.

The useful revision test is whether fresh, dated estimates converge toward the company’s contracted economics after analysts digest the release. We would want a consistent definition, enough contributing analysts and explicit assumptions about steel prices, volumes and costs. A higher target without those inputs says little about downside protection. Likewise, an older bearish mark should not be treated as a fresh rebuttal of new operating evidence.

Management and governance

Credit the delivered quarter; separate commitments from conditions.

Management delivered above its prior quarterly EBITDA guide, and the current release identifies pricing, volumes and unit costs as drivers. That earns credit for execution. The larger 2027 statement is explicitly conditional on signed contracts and prevailing price forecasts. Signed commercial terms can improve visibility, but index exposure, customer volumes and cost assumptions remain important. We do not convert conditional language into a guaranteed earnings floor.

The covered security is NYSE-listed Cleveland-Cliffs common stock, with no ADR conversion in the per-share model. The annual report identifies Deloitte & Touche as auditor. This is not a foreign ADR holding-company case requiring an ADS ratio or an assumed contractual-control discount. The annual report’s ordinary consolidation policy discusses variable-interest entities; that wording alone does not establish either zero VIE exposure or a separate structural impairment.

The annual report showed no preferred shares outstanding and described the common issuance used to repay ABL borrowings. We do not infer insider ownership or voting concentration from the free-float percentage. A precise beneficial-ownership judgment requires the relevant proxy and current holdings. The largest decision-relevant exposures demonstrated here are manufacturing cyclicality, direct automotive demand, financing costs and capital allocation, rather than an unsupported governance score.

1. How much of the 2027 EBITDA outlook comes from fixed contract changes versus index-price assumptions?
2. What quarterly cash conversion is required to reach the mid-2027 leverage goal, after maintenance investment?
3. How much expected debt reduction depends on asset-sale proceeds rather than operating free cash flow?
4. What is the full cash cost and restart condition for the Stelco galvanizing idling?
5. How will management prioritize debt repayment, buybacks and discretionary expansion if steel prices weaken?

The annual employee series was approximately 28,000 in 2023, 30,000 in 2024 and 25,000 in 2025. Acquisitions, disposals and idling affect that series, so it should not be used as a clean productivity ratio without a comparable operating perimeter. We retain the figures as context and avoid presenting a reduction in headcount as an automatic increase in economic efficiency.

Ranked risks

The balance sheet makes a cyclical miss expensive for the equity.

The bull case deserves a fair hearing. Contract resets could make the latest quarter a stepping stone rather than the peak. Higher realized prices, steadier volumes and better costs could allow EBITDA to approach the company’s outlook while working-capital needs normalize. Cash would then retire debt and increase the value left for common shareholders. That combination explains why a large upside scenario remains plausible despite our cautious central call.

The bear case is equally concrete. Index forecasts can weaken, customer orders can fall and inventory can absorb cash. Fixed operating and financing costs leave less room for error. If normalized EBITDA is only $2.2 billion and investors pay five times that amount, subtracting present net debt leaves roughly $6 per share. The scenario does not require the business to disappear; it requires a less favorable operating cycle and a constrained equity claim.

We land between those cases because the reported recovery is convincing but the financing and cash evidence has not caught up. Our scenarios use a constant present net-debt balance and rounded share values. They are decision tools, not price predictions or a complete distribution of outcomes. Extreme operational disruption, new issuance or a major transaction could produce values outside the displayed range.

What changes the call

Dates below distinguish announced events from analyst checkpoints.

The next useful catalyst is evidence, not simply another target-price headline. We will compare Q4 cash generation with EBITDA, distinguish property-sale proceeds from recurring cash, and reconcile any debt reduction to the balance sheet. When the full quarterly filing appears, it should provide details beyond the earnings exhibit. Until then, unavailable fields remain unavailable rather than being backfilled with a provider zero.

The HOLD can change in two ways. A lower price can create a sufficient discount if the underlying business remains on track, or stronger verified cash and contract economics can raise our estimate of sustainable earnings. Conversely, a miss on cash conversion, a worsening debt bridge or weakening realized prices can lower the value even if reported revenue keeps growing. The signposts at the top make those conditions reviewable next quarter.

Data, sources and method

USD millions unless explicitly stated; computed items are labeled.

Source hierarchy: the current issuer earnings release and its 8-K exhibit govern current financial figures; the annual report and latest quarterly filing govern structural and accounting context. We used the SEC submissions index to verify the exact accession, and the issuer-hosted copies of the corresponding filing and exhibit for their contents. Direct SEC archive retrieval was unavailable during collection; no successful direct-archive body read is implied. The issuer mirrors and collected files are hash-bound in the research evidence.

• Cleveland-Cliffs Q3 2026 release, October 19
• Exact current 8-K earnings exhibit
• Q2 2026 Form 10-Q
• 2025 Form 10-K and comparative financial statements
• Issuer filing archive for historical quarterly exhibits and annual reports
• FMP quote, historical prices, estimates, targets and ratings snapshots
• Prior CLF Q2 2026 episode: HOLD; cash-flow and EBITDA recovery

Historical comparability is imperfect. The annual filing corrected immaterial prior employment-cost accruals, and Stelco changed the operating perimeter after its acquisition. We identify specific corrected comparisons instead of silently overwriting every original historical line. The current-release columns are the authoritative source for the present year-over-year and sequential comparisons. Historical charts should be read as a documented series, not a fully restated same-perimeter econometric dataset.

The prior public Q2 episode establishes that the earlier call was HOLD and centered on improving EBITDA and positive cash flow. We do not invent a prior numerical target or signpost score from its title. A full like-for-like target track record requires the exact prior model; the current packet’s dated numeric signposts provide that foundation going forward. This protects the distinction between a verified prior stance and an unverified retrospective performance claim.

Definitions: free cash flow here equals operating cash flow less purchases of property, plant and equipment. Simple net debt equals reported long-term debt less cash and cash equivalents. Enterprise value equals the fixed quote times the reference share count plus simple net debt. EBITDA is not cash flow. Ending-balance receivable and inventory days are directional proxies. Scenario probabilities, normalized earnings, valuation multiples, discount rates and signpost thresholds are our analytical judgments.

The dataset retains the full quarterly source tables, including operating costs, below-the-line items, reconciliations and balance sheets, rather than just the highlighted rows above. Some details are not separately available in an earnings release, including a new quarter-end cover share count and standalone quarterly SBC. The valuation uses the 570.54 million reference count tied to the prior filing/provider observation. It does not infer shares by dividing rounded income by rounded EPS.

No constant-currency conversion is applied: the issuer presents these consolidated statements in U.S. dollars. Canadian operations introduce economic currency exposure, which is already reflected in the consolidated reported figures and disclosed currency adjustments. The packet makes no claim to isolate every foreign-exchange effect. Charts, scenarios and financial arithmetic were generated from the collected dataset; all forecast inputs remain explicitly labeled assumptions.