Trigger: Lonis Hamaili's Cambrian Thesis (2026-05-04) flagged CLF as steel for data centers + grid build-out — the only one of his 22 names not in any of our 41 watchlists. This deep-dive validates the AI-attributable revenue angle before deciding whether to add to ai-power watchlist.

Editorial Note

The AI angle is more structural than the headline suggests. CLF isn't just "auto-grade steel" — they've already committed $150M to a West Virginia plant retrofit for grain-oriented electrical steel (GOES), the specific input for grid transformers and HVDC. Expected $500M annual EBITDA contribution once online. This is a real picks-and-shovels exposure to the grid build-out, not a marketing pivot. The bear case is the rest of the business: auto-dominated revenue, 5 straight losing quarters, $7.76B net debt, and recent tariff coverage labels CLF as "struggling" vs STLD which is "thriving." Bull case = if GOES ramp + tariff regime + 43% short volume ratio compound into a turnaround. Verdict: add to ai-power watchlist as a power-infrastructure cohort member alongside GEV / ETN / VRT, but don't enter as a position without a specific trigger (GOES plant commissioning timeline, Q2/Q3 EBITDA contribution disclosure, or short-squeeze setup).

The Story Right Now

CLF sits in an awkward middle. The narrative bull case (US steel + tariffs + AI grid + GOES capacity) is real. The financial reality (5 consecutive quarterly losses, negative gross margins until very recently, operating cash flow −$325M in Q1 (correction 2026-08-03, D038 class: the original "FCF -$465M" was OCF −$325M plus the entire −$140M investing line — the investing-CF-as-capex substitution; per EDGAR, true Q1 FCF ≈ −$477M on $152M of real capex, so the direction and rough scale survive, but the printed figure was resolver-made), $7.76B in long-term debt with effectively zero cash) is also real. The 30-day +21% rip is the market sniffing the turnaround — Q1 2026 revenue $4.92B with gross margin -1.7% (vs -8.4% Q1 2025, narrowing fast), revenue acceleration +23%, 100% earnings beat rate.

The GOES story changes the framing. Lonis put CLF in his "priced in but safe" quadrant at 200%+ off-scale coverage — meaning their order book is already huge relative to AI demand. He's right that most of CLF's order book is auto/industrial (not AI-specific), but the $150M West Virginia retrofit is a new capacity addition specifically targeted at the AI-grid story. That's not "priced in" the same way the existing book is — it's incremental capacity at premium product pricing (transformer steel commands a margin premium over commodity flat-rolled).

Two voices in 3 days have flagged the same structural setup from different angles: NuttyCLD's AI Power Crisis series (delivery-side voltage cascade) + Lochmiller/Crusoe (gas-turbine pricing 3x'd) both implicitly need transformer capacity on the back-end. CLF's GOES plant is the upstream supplier for that demand. Same story Lonis is telling about CEG/VST/TLN as IPPs — except CLF is the input chain TO those IPPs.

What I don't love: CLF is more leveraged than the IPP cohort, more cyclical than GEV (auto exposure swings), and the GOES story is a 2026-2027 EBITDA build-out rather than a near-term catalyst. This is "earn the conviction over the next 2-3 earnings calls" — not "buy now."

Quick Snapshot

Action Matrix

Price Data

Company Overview

Cleveland-Cliffs Inc. (NYSE: CLF) — North America's leading flat-rolled steel producer. Headquartered in Cleveland, OH. ~25,000 employees across US + Canada. Vertically integrated from iron-ore mining → DRI → primary steelmaking → finishing/stamping/tooling/tubing.

Four operating segments (consolidated to one reportable: Steelmaking):

• Steelmaking — primary business
• Tubular — pipe/tube products
• Tooling and Stamping — downstream finishing
• European Operations

Primary competitive position: leading supplier of automotive-grade steel in the US. Auto-grade is the highest-quality, most operationally intensive end-market — hence higher through-the-cycle margins.

Key vertical advantage: mining → DRI → steel → finished goods, all in-house. Insulates from input cost shocks BUT exposes them to commodity price volatility on the output side (steel/scrap/iron ore).

The AI Angle (this is the new thing)

The GOES product specifically is what makes CLF interesting for our ai-power thesis. Most flat-rolled steel commodifies; GOES is specialty steel with limited US producers (CLF is one of two — AK Steel which CLF acquired in 2020 was the historical leader). It's the input layer for what GEV / ETN / VRT sell downstream.

Financials (5-quarter trail)

Trajectory: losses narrowing fast — gross margin -8.4% → -1.7% over five quarters. Revenue stabilizing at ~$4.6-4.9B/quarter. The path to break-even is visible but not imminent.

Balance sheet (recent):

• Total assets: $20.1B
• Total liabilities: $14.1B
• Total equity: $6.0B
• Long-term debt: $7.76B
• Cash: $0
• Debt-to-equity: 1.29

Cash flow:

• Operating CF: -$325M
• CapEx: -$140M
• Free cash flow: -$465M, FCF margin -9.4%

The leverage profile is the bear case. With ~$8B in debt and zero cash, every quarter of negative FCF is operationally fragile. They have asset coverage (mining + plant base) but not liquidity coverage.

Short Interest (squeeze setup)

43% short volume is the squeeze tinder. Historical CLF squeezes have been violent (+30%+ in days). Combined with the GOES + tariff narrative bid, the asymmetric upside on a positive catalyst is real.

Bull case

1. GOES capacity comes online on schedule with a $500M annual EBITDA contribution materializing in 2026-2027 disclosures. Re-rates the multiple.
2. Tariff regime favors US steel. April 24 piece pegged CLF as "struggling" but Steel Dynamics (STLD) "thriving" — same regime, different cost structures. CLF's vertical integration could turn into the asset that makes it work if input costs settle.
3. Auto cycle bottoms. Auto-grade demand has been weak through 2025; cyclical recovery + tariff protection on imports = margin expansion.
4. AI grid build-out. ~70% of US grid is end-of-life; transformer demand is structural and multi-year. CLF is one of two US-based GOES producers.
5. Short-squeeze tinder. 43% short volume ratio + narrative bid = asymmetric upside on positive catalysts.

Bear case

1. Balance sheet fragility. $7.76B debt, $0 cash, -$465M FCF. Another commodity downturn or auto contraction and refinancing risk activates.
2. Auto-dominated revenue dilutes the AI signal. GOES is real but a small fraction of total revenue. The "AI infra play" framing is partial — CLF's results are mostly determined by auto cycle + steel commodity prices, not AI demand.
3. Tariff regime is reversible. Single executive-order shift could undo the tailwind. STLD's better cost structure suggests CLF is the marginal player even within the protected regime.
4. GOES timeline risk. $500M EBITDA is forward-looking. Plant commissioning, ramp, customer qualification, end-product validation — multiple ways the timeline slips.
5. Negative gross margins for 5 quarters. Until they print a positive gross margin quarter, "turnaround" is a story, not a fact.

Catalysts

Recent News

• 2026-04-24 — Tariffs Rose: 1 Steelmaker Thrived, 1 Still Struggles (Investing.com) — STLD up; CLF flagged as the struggling side
• 2026-04-24 — Resilient Q1 results and strong steel demand (Simply Wall St) — Q1 showed narrowing losses + automotive demand firm
• 2026-02-13 — Stock a Steal Buy After Falling Off the Cliff? (Motley Fool) — bottom-fish thesis after the Feb-9 sell-off
• 2026-02-09 — Q4 2025 revenue miss + full-year net loss → -10%+ stock move
• 2025 — Announced $150M West Virginia GOES plant retrofit
• 2024 — DOE selects CLF for $575M industrial decarbonization investments

Cross-References

Verdict + recommended action

Conviction: medium. Real GOES angle, weak balance sheet, turnaround narrative tracking but not yet proven.

Recommended actions (NOT trade entries — research-side decisions):

1. Add CLF to ai-power watchlist — alongside the power-infrastructure cohort (GEV / CEG / VST / TLN / ETN / VRT). Treats CLF as the upstream-input layer feeding the transformers + grid build.
2. Hold off on key_ticker promotion to ai-power-bottleneck perspective until GOES plant disclosure firms (commissioning timeline + first EBITDA contribution). Currently track-only.
3. Monitor for short-squeeze setup — 43% short volume ratio + narrative bid means a single positive catalyst (DOE funding milestone, GOES commissioning, tariff ratchet, auto recovery print) could trigger a violent move.
4. Re-evaluate after Q2 2026 earnings (mid-Aug) — first chance to see whether the trajectory of narrowing losses continues through a print and whether GOES timeline gets concrete dates.

Trade trigger if entering (low-probability scenario):

• GOES plant commissioning announcement with a 2026-Q4 or 2027-H1 ramp date → entry into $9.68-$10.19 zone
• Stop $8.15 (below 30D low of $8.37 with cushion)
• Target $13.25 (+30% from entry)
• Position size small (volatility + balance sheet fragility) — bench-signals or YOLO sizing only

Sources

• Article capture: inputs/articles/2026-05-04-loniss-cambrian-thesis.md
• Triggering input event: 2026-05-04-ai-supply-chain-2x2-ai-beta-coverage-of-15gw-substitution-risk
• Web search 2026-05-04: Cleveland-Cliffs (CLF) Q1 valuation analysis, Why CLF Stock Is Surging as AI Demands Steel, Cleveland-Cliffs Q1 2026 Results, DOE $575M industrial decarbonization investments, AI 2026: Data Centers Restart Growth of a Stagnant U.S. Electrical Grid
• Local data: 2026-05-04 (Polygon + leading indicators + 10-K excerpt)
• 10-K: 2026-02-09 filing — risk factors discuss steel/scrap/iron ore commodity price volatility as primary risk