Steel Dynamics

Refresh delta (since 2026-06-18)

What changed

1. Q2 2026 printed far hotter than the prior dossier's model. Diluted EPS $3.69 on net sales $6,091.6M (+33% YoY) and operating income $700.5M (+83%) — a record 3.7M-ton steel quarter. H1 2026 diluted EPS is $6.47. The prior dossier's full-year base case was ~$10.00; the company banked 65% of it in six months.

2. Founder succession — and not to the presumed heir. On 2026-08-03 Mark Millett gave notice he retires as CEO effective 2027-01-01, becoming Executive Chairman. CFO Theresa E. Wagler becomes President & CEO; EVP & Treasurer Richard A. Poinsatte becomes CFO. Effective 2026-09-01, Barry Schneider moves from President & COO to EVP & Chief Technology Officer, with three divisional COOs installed beneath (Alvarez — aluminum + recycling; Anderson — construction products; Graham — flat-rolled). The prior dossier called Schneider "the clear COO heir, which de-risks it." That is now falsified.

3. The tariff wall is leaking, on management's own account. COO Barry Schneider: "We are seeing certain countries shipping through the 232s… very high accelerated rates of all steel exports from primarily three countries in Asia. These are disruptive" — with imports up month-over-month and "July… up noticeably versus June". Corroborated externally: hot-rolled sheet imports jumped 128% MoM in May, and Nucor paused its HRC spot hikes on import risk.

4. Aluminum inflected — but missed its own guidepost. Shipments 53,000 mt vs 22,500 mt in Q1 (+135% sequential), mill output 84,000 mt (~50% of capability). Segment loss −$49.9M, of which $33M is startup cost (a 48% sequential improvement) and $16M a non-cash impairment on relocating the second satellite slab center from Arizona to Columbus, MS. Management had guided 60–70k mt: "the 53 could have quite easily have been 60… just a matter of the team working out all the bugs" (Millett). Third cold mill online Aug 2026; earnings-positive committed for H2 2026; exit-2026 run-rate ≥90% of capacity.

5. Cash conversion repaired. The prior dossier's flagged earnings-to-cash gap narrowed sharply: OCF $427.9M on capexCapital expenditureMoney spent on long-lived things — buildings, machines, servers — rather than on running costs. of only $123.8M → FCFFree cash flowCash left after paying to run and maintain the business. Unlike profit, it is hard to flatter with accounting choices. $304.1M in Q2, vs FCFFree cash flowCash left after paying to run and maintain the business. Unlike profit, it is hard to flatter with accounting choices. of ~$10M in Q1. CapexCapital expenditureMoney spent on long-lived things — buildings, machines, servers — rather than on running costs. is rolling off hard (H1 $261.8M vs $593.8M LY; H2 guided $300–350M). Interest coverage improved to 15.63x from 13.7x; debt/cap 0.31.

6. The stock went the other way. ~$267.79 (2026-08-10) vs $282.76 at the prior dossier — −5.3%, off an all-time closing high of $282.12 (2026-06-12). Earnings power up, price down.

What held

The structural thesis stands, and Q2 strengthened three of its four legs:

• The circular/vertical-integration moat is doing exactly what it is supposed to. STLD mills ran 90% utilization vs an 81% industry rate, with intra-segment shipments up 49% YoY (550k vs 368k tons) — the fabrication arm absorbing internal tons is the mechanism, and it is visible in the numbers.
• Demand is broadening, not narrowing. Fabrication backlog ~+45% YoY (was +38% at Q1), now extending into Q1 2027; Dodge Momentum Index +30% YoY at its highest recent level; contractor backlogs >9 months.
• Balance sheet remains a fortress. Total debt $4.18B unchanged, liquidity $2.0B, coverage 15.6x vs a 2.50x covenant floor.
• The scrap-tightening bear point weakened. Management now reports "ample supply" with scrap expected to "go sideways," and notes new entrants (Nippon, Hyundai-Posco) are adding DRI capacity, which balances iron units rather than bidding up scrap. This is a genuine, sourced softening of the prior dossier's chokepoint #1 — not a removal of it.

What did not hold is the prior dossier's valuation conclusion, and it failed in the investor-friendly direction. See Lens 11.

Refreshed lenses

Company Overview — refreshed (minor)

Three disclosure changes since June:

• SDI Biocarbon Solutions is now 100% owned (was 75%), effective 2026-06-19. The decarbonization input is fully consolidated.
• The second 150,000-mt satellite slab center is now "planned for construction in Columbus, Mississippi" — relocated from Arizona, at a $16M write-off and a $10–20M capex increase at Columbus that Schneider attributes to "inflation of construction contractors more than anything else." UBC (used-beverage-can) sourcing from the West Coast drops from 100% to ~60–70% of that unit's feed.
• Aluminum is no longer a rounding error: 8% of consolidated net sales in Q2 2026 vs 1% a year ago. Steel Operations fell to 66% of net sales from 72%.

Total aluminum program cost is now publicly framed as ~$2.5B for the mill plus ~$400M for the two satellite slab centers.

Competitive Advantages — refreshed (material change)

Two of the four moat pillars moved in opposite directions this quarter.

Stronger — the utilization moat is now quantified. 90% vs 81% industry, and management names the mechanism outright: "Our steel fabrication platform provides meaningful support for our steel mills, particularly critical in softer demand environments, allowing us to operate at higher through-cycle utilization rates than our peers" (Schneider). Value-added spreads to hot band recovered +$70/ton off the Q4-2025 lows, and STLD is the largest flat-rolled coater in North America.

Weaker — the tariff is not the wall the June dossier treated it as. The prior dossier framed Section 232 as an on/off risk: 50% holds or it doesn't. Q2 reveals a third state that is more likely than either — the tariff nominally holds while product routes around it. Three Asian exporters are absorbing/circumventing the 50% duty on price, imports are rising monthly, and STLD's response is lobbying ("we are hoping the administration… try to rein it back in"), not pricing. A moat you have to petition for is thinner than a moat you own. Offsetting, partially: new Section 301 investigations are underway which STLD wants "additive to the existing 232 steel tariffs," the Section 301 on Brazilian pig iron resolved favorably for minimills, and STLD is engaged in the USMCA review defending melted-and-poured provisions.

Moat verdict, updated: the operating moat is real and measurable; the policy moat is porous. That is the same conclusion as June but with the emphasis moved — the June dossier worried about tariff reversal, and the actual observed failure mode is tariff erosion.

Segments — refreshed

Q2 2026 vs Q2 2025, all:

The mix shift the June dossier flagged is now happening, on schedule but not on size. Aluminum went 1% → 8% of sales in a year. The segment loss, however, did not narrow YoY (−$40.6M → −$49.9M); it narrowed sequentially ex-impairment (−$64M Q1 → −$33M Q2). Both framings are true; the sequential one is the honest read of the ramp, the YoY one is the honest read of the drag still in the FY number. H1-2026 aluminum OI is −$114.5M, already two-thirds of FY2025's full-year −$173M.

Fabrication is the one genuinely deteriorating line, and it is deteriorating for a benign reason: it is being squeezed by its own steel segment's success. Management expects realized fabrication pricing to stay flat and only improve "in the next six to nine months," because today's higher-priced backlog lands in Q4 2026 and 2027 (Wagler).

Earnings Result (Q2 2026, reported 2026-07-20; call 2026-07-21) — re-run

Income statement, Q2 2026 vs Q2 2025:

• Net sales $6,091.6M, +33% (vs $4,565.1M)
• Gross profit $959.0M — 15.7% margin vs $618.5M / 13.5% (+220 bps)
• Operating income $700.5M, +83% — 11.5% margin
• Interest expense, net $39.1M, +125%
• Pre-tax $683.5M; tax 22.3%
• Net income to STLD $534.1M; diluted EPS $3.69 (vs $2.01, +84%)
• Adjusted EBITDA $921M

vs guidance and vs consensus — the two answers differ, and that matters. STLD guided $3.51–$3.55 on 2026-06-17. Actual $3.69 beat its own guide by ~$0.16. Street consensus was $3.71, so it was scored a $0.02 miss. Revenue beat hard: $6.09B vs ~$5.68B expected (+$410M). Note the company's guide already absorbed the $16M impairment. A revenue-beat / guide-beat / consensus-hairline-miss print.

What drove it: steel metal spread again — ASP +$162/ton against scrap +$5/ton, spread +22%, steel OI +89%. Sequentially, steel ASP rose another $105/ton and steel OI rose 30%.

Balance-sheet flags (6/30/26):

• Cash $567.7M; total liquidity $2.006B; total debt $4.182B, unchanged from YE25.
• Interest coverage 15.63x (vs 2.50x covenant floor); debt/cap 0.31 (vs 0.60 max). Both improved from Q1's 13.7x / 0.32.
• Working capital +$631.8M in H1 — AR +$760.3M, inventory +$217.1M, offset by AP +$252.2M. This is the up-cycle absorption flagged in June, and it is still running.
• Equity $9.22B; total assets $17.11B; shares outstanding 143.61M (down from 144.94M at YE25).
• Buyback: $488.7M remaining on the Feb-2025 $1.5B authorization; Q2 repurchases 859,441 shares at a weighted ~$233.04 [computed from the Item 2(c) monthly table: 407,544@$209.47 + 212,392@$237.06 + 239,505@$269.60 = $200.29M ÷ 859,441].

Market reaction: muted-positive, +1.3% to +2.0% on 2026-07-21. As in Q1, the print is not what moves this stock. Since the June 18 boundary the shares are −5.3%, having peaked at $282.12 on 2026-06-12 — i.e., the drawdown the June dossier said it wanted to buy has partly arrived, and it arrived through a strong quarter.

Earnings Calls (sentiment shift) — re-run, now transcript-grounded

This is the first refresh where the transcript is on disk (transcripts/2026-q2.md), so the June caveat ("treat as directional") is retired for Q2.

Tone vs the prior 3–4 calls: the trajectory from FY24–25's defensive register ("through-cycle discipline," "value-added diversification," Sinton ramp difficulty) through Q1 2026's pivot to demand strength continues, but Q2 adds a register that was absent in Q1: operational candour about a ramp that slipped. Millett on the 53k-vs-60-70k miss: "it's all part of the learning curve… We actually left material on the floor that we couldn't ship for a variety of different reasons… No Achilles heel or whatever, but just a couple of little packaging issues." Wagler on working capital: "working capital really did increase in the second quarter more than we anticipated." Neither is spun. That is a quality tell in the same direction as the June read.

Phrases that grew: trade enforcement rather than trade tailwind — 232 now appears alongside 301, USMCA melted-and-poured, Buy American in the Ships for America Act and the Federal Highway Bill. The frame moved from "the tariff is helping us" to "we are working to keep the tariff from being routed around." Phrases that faded: Sinton (now folded into a completed-capital narrative). New and load-bearing: ">$1.4 billion of through-cycle annual EBITDA capability" from the >$5B invested in Sinton + coated lines + aluminum, and a specific aluminum-platform normalized target of $650–700M EBITDA plus $40–50M for metals recycling.

Commitment escalation on aluminum. Q1 said shipments and earnings would "increase sharply in Q2." Q2 goes further and firmer: asked directly whether Q3 pivots to profitability, Wagler answered "we've been kind of approaching EBITDA-positive to EBITDA-positive, but we would expect to be earnings positive in the second half of the year," and on ≥90% exit utilization: "Absolutely. We feel even more confident about that." Management has now put a falsifiable, dated commitment on the record. That is the single cleanest thing to grade this thesis against.

Sentiment shift verdict: constructive-bullish, unchanged in direction, improved in specificity, with one new note of genuine caution — the import discussion is the most defensive passage on the call, and it came unprompted from the COO.

Note for the record: the call opened with a workplace fatality ("Please keep Elijah's family, friends and teammates in your prayers" — Millett). Not a financial item; it belongs in any honest read of a company whose stated first cultural pillar is safety.

Catalysts and >5% moves — re-run

Moves and catalysts since the 2026-06-18 boundary:

1. 2026-06-12: all-time closing high $282.12, immediately pre-boundary — the terminus of the June Section 232 reaffirmation run the prior dossier documented.
2. 2026-06-17: Q2 guidance $3.51–$3.55 issued with the $16M aluminum write-down disclosed inside it — read as a raise vs the sequential $2.78.
3. 2026-07-20/21: Q2 print, +1.3–2.0%. Small.
4. 2026-08-03/04: CEO succession 8-K. No large price move located in the searches; analyst response was mildly constructive (Barclays $270→$272 Overweight; a $300 target extant).
5. Through July–August: the steel-price and import narrative turned. HRC ~$1,155/ton (CRU futures) — still elevated — but Nucor paused its spot hikes citing import risk, and Morgan Stanley published HRC $1,012 (2027) and $900 (2028), i.e. a called medium-term peak.

No single >5% single-day move was identified in the period. The −5.3% since June 18 is a grind, not an event — consistent with the pattern established in June: the macro sets the level, the print barely registers. The one change to the catalyst map is that trade enforcement has split into two separately-tradeable catalysts: 232 durability (unchanged, runs through Dec 2027) and 232 leakage (new, live now, resolvable by quotas or 301 stacking).

Management — rewritten (the largest single delta)

The succession is announced, and it is not the one the June dossier projected.

Read it honestly, both ways.

The reassuring read: Wagler is not a parachuted outsider. She has been at STLD 28 years, CFO for 19 of them, with an unusually broad remit (treasury, legal, cyber, decarbonization, M&A) and she delivered the operating narrative on the Q2 call, not just the numbers. Millett stays as Executive Chairman rather than leaving. The single-COO role is replaced by three divisional COOs, which is a deliberate widening of the operating bench, and Alvarez — who ran North Star BlueScope, the only North American EAF flat-roll mill in BlueScope's portfolio — takes the aluminum + recycling seat at exactly the moment aluminum is the swing factor. Continuity of strategy looks well-protected.

The read that deserves weight: the company has handed the top job to a finance executive rather than a metallurgist, at a company whose entire founding legend is that Millett personally designed, built and started up the mills, and whose next three years hinge on finishing an aluminum ramp — an operating problem, not a capital-allocation problem. Simultaneously the most experienced operator in the building, the man the June dossier and the market both read as heir, is moved to CTO, a role with no P&L. It may be exactly what it looks like (a technology-and-process seat for the person best qualified to fill it, freeing three divisional COOs to run the plants). It may also be a consolation. The filings do not say, and I will not pretend to know. What is knowable: the succession-risk item the June dossier scored as "de-risked" is now an open question, and the fair grade is neutral-to-mildly-negative pending the first year.

Capital allocation — updated. H1 2026: $315.4M of buybacks and $149.0M of dividends paid, with the dividend raised 6% to $0.53/qtr ($2.12 annualized, ~0.79% yield at $267.79). Wagler on the forward pace: "we want to take the opportunity as we have access to cash flow to lean in on the share repurchases. We think that the current pricing is still incredibly attractive… it's going to trend more toward like what you saw in the second quarter" — i.e. ~$200M/qtr, against $488.7M authorized. Capex discipline is now visible: H1 $261.8M vs $593.8M LY, H2 guided $300–350M → implied FY2026 ~$0.56–0.61B vs $0.948B FY2025.

A live M&A situation the June dossier missed entirely. Since late 2024 STLD has, with Australia's SGH, made four approaches for BlueScope Steel, under a structure where SGH acquires BlueScope and on-sells the North American operations to STLD. The bid was raised in Feb 2026 to a "best and final" A$32.35/share (~A$15B / ~US$11B equity value) and rejected again in March 2026; as of Aug 2026 it is reported as not withdrawn.

Red flags: still none material. Related-party sales collapsed to $22.0M in Q2 from $224.3M a year ago (0.4% of sales, from 4.9%) — the NPS consolidation doing its work, and the right direction. No director or officer adopted, modified or terminated a Rule 10b5-1Rule 10b5-1 planA schedule an insider sets in advance saying when their shares will be sold, so the sales cannot be timed on private news. A sale made OUTSIDE such a plan is a live decision, which is why it carries more information. arrangement in Q2 [Item 5]. Insider ownership % remains n/a (the 2026 DEF 14A, filed 2026-03-27, is still not ingested).

Forensic Red Flags — re-run

The June conclusion (clean books; the watch-item is earnings-to-cash) partially resolves in the company's favour, and two small new items appear.

• Earnings-to-cash gap — improving. Q2 OCF $427.9M against net income $530.8M is a 0.81x conversion, versus Q1's 0.37x. With capex at $123.8M, FCF was $304.1M. Wagler expects working capital to be "a funding to certainly neutral, but I would suggest a funding source in the second half." H1 conversion is still only 0.62x (OCF $576.3M / NI $930.9M), so the gap is narrowing, not closed. Grade: watch, downgraded from concern.
• The $16M impairment is clean. Non-cash, disclosed in advance inside the June guidance, booked in SG&A, tied to a named decision (Arizona → Columbus MS relocation), and the incremental capex ($10–20M) and operating consequence (UBC feed 100% → 60–70% from the West Coast) were both volunteered on the call. This is the opposite of an impairment used to bury something.
• New, minor: a management-vs-filing numeric inconsistency on buybacks. Wagler said on the call "in the first half of 2026 we… repurchased $350 million of our common stock" and separately "we paused in the first quarter." The filings say $315.4M of H1 treasury purchases, of which $200.3M was Q2 — meaning Q1 was $115.1M, not a pause. The filed number governs. This is a ~$35M imprecision in verbal remarks, not an accounting issue — but it is the kind of thing that should be scored, not waved through, and it is worth watching whether the new CFO's numbers tie more tightly.
• Interest expense, as predicted, is now a real recurring line. $39.1M in Q2 (+125%), driven mostly by an $18.8M (68%) drop in capitalized interest as the aluminum mill left construction. H1 run-rate implies ~$150M/yr, above the June dossier's $130–140M estimate. Cash interest paid H1 was $93.1M.
• Revenue recognition, leases, goodwill, SBC, internal controls: unchanged. Goodwill flat at $477.5M, no impairment. Disclosure controls effective; no change in ICFR during the quarter. Item 1A: "No material changes… to the risk factors" vs the FY2025 10-KForm 10-KA company’s audited annual report to the US regulator. The most complete thing it publishes..

Regulatory findings (re-verified this run):

• SEC Litigation Releases: 0. SEC AAERs: 0. EDGAR EFTS search covering 2021-08-10 → 2026-08-10, re-fetched 2026-08-10.
• 10-Q Part II Item 1: routine litigation, environmental and contract disputes, "none of which are currently expected to have a material impact"; EPA/state matters (RCRA, CERCLA, Clean Water Act, Clean Air Act) with fines and penalties not exceeding $1 million in aggregate as of 2026-06-30 — identical language and threshold to 3/31/26 and 12/31/25.
• Conclusion: no material regulatory or legal findings, unchanged from June. (Standing caution carried forward: EPA steel settlements that surface in searches — e.g. the $1.5M Braddock penalty — are U.S. Steel, not Steel Dynamics.)

Net forensic read: clean, and cleaner than in June on the one item that was flagged.

Forward Projection (FY2026 / FY2027 / FY2028 diluted EPS) — re-run; this is where the thesis moved

Model-integrity disclosure first, per the honesty rule. our model was re-run this session and wrote model.xlsx. I am not citing a single number from it, because the read-back reports:

• "No computed values. The workbook has 51 formulas and no cached results" — the DCF has never been evaluated;
• only 2 quarters of financials on the shelf, so "LTM figures are partial" and "revenue growth seeded at 8%" (a default, not a finding);
• share price blank, so value-per-share and upside are undefined.

The opening balance sheet is now sourced from our figures (an improvement — it was hand-typed before), and the balance check cannot be verified until the workbook is opened and saved in Excel. Everything below is with the arithmetic shown, anchored to filed actuals — not model output.

Anchors (all research-layer, filed): H1 2026 diluted EPS $6.47; Q2 $3.69; Q1 $2.78; FY2025 $7.99; H1 2025 $3.44 (∴ H2 2025 = $4.55); shares outstanding 143.61M and falling ~0.9%/half on ~$200M/qtr of buybacks; tax 22.2–22.3%; interest ~$150M/yr.

Bridge inputs:

• Lagging contracts: "approximately 80% or more of our flat rolled steel business is linked to lagging price contracts… generally lagging two months," so Q2's price rises land in Q3 [transcript]. This makes Q3 ≥ Q2 close to mechanical absent a demand break.
• Aluminum: H1 OI −$114.5M actual; management commits to earnings-positive in H2 2026 and ≥90% exit utilization; normalized platform target $650–700M EBITDA. Assume FY2026 aluminum OI ≈ −$110M (H2 ≈ +$5M), FY2027 +$150M base, FY2028 +$330M base — all well below the normalized target, because the normalized target is a through-cycle end-state, not a 2027 number.
• Steel spread: HRC ~$1,155 now; Morgan Stanley models $1,012 (2027) and $900 (2028). Base case takes roughly half that fade to the bottom line, since scrap moves with HRC and ~70% value-added mix cushions.
• Buyback: ~1.8%/yr share shrink → 143.6M → ~141M (FY27) → ~138M (FY28).

FY2026 is largely already determined. H1 is banked at $6.47; only H2 is in question.

This is the headline of the refresh. The June base case was FY2026 ~$10.00, implying ~28x at $283 and underwriting the "priced for permanence" objection. The filed H1 alone is $6.47. A base of ~$13.90 at $267.79 is ~19.3x current-year, and ~20.6x the FY2027 base of $13.00.

Cross-checks, computed, with arithmetic shown:

• Trailing P/E: LTM EPS = FY2025 $7.99 − H1'25 $3.44 + H1'26 $6.47 = $11.02; $267.79 ÷ $11.02 = ~24.3x. Conflict surfaced, not resolved: the June dossier cited "~24x trailing " while also attributing it to FY25 EPS of $7.99 — but $283 ÷ $7.99 = 35.4x, so those two statements could not both be true. My 24.3x is computed from filed EPS and is internally consistent; I am flagging the prior figure as unreliable rather than inheriting it.
• EV/EBITDA: market capMarket capitalisationThe share price multiplied by the number of shares. What the market says the equity is worth. = 143.61M × $267.79 = $38.45B; + net debt $3.61B = EV ~$42.07B. On Q2 adjusted EBITDA of $921M annualized ($3.68B) → ~11.4x. Second conflict surfaced: the June dossier's "~8.0x EV/EBITDA" came from a Sep-2025 Benzinga citation for 2026E — a year stale and struck at a far lower share price. I will not restate 8.0x as current. On a plausible FY2027 EBITDA of ~$4.0–4.3B (aluminum ramped), the forward multiple is ~9.8–10.5x. Peer-relative EV/EBITDA is n/a at current dates; the June tiering (STLD/NUE premium ~8x vs CLF/CMC ~6.5x) is directionally carried but its absolute level is stale.

The honest summary of the valuation delta: on earnings the stock got materially cheaper (28x → ~19x base). On EBITDA it did not obviously get cheaper, because EBITDA-based comps at ~11x trailing / ~10x forward sit above the stale 8x anchor. Those two facts are reconcilable — EPS is being flattered by the buyback and by aluminum's D&A sitting above the EBITDA line while its losses sit below it — and anyone taking the bull side on the P/E should know the EV/EBITDA does not corroborate it as cleanly. That tension is the single most useful thing this refresh produced.

Brier forecast: NOT logged (unattended run — no our model create). The loggable binary, if promoted: "STLD aluminum segment operating income ≥ $0 for H2 2026 (Q3+Q4 combined), resolves 2027-02-28, p≈0.60" — chosen over an EPS threshold because it is management's own dated, falsifiable commitment and it is the actual swing variable.

Bull vs Bear — updated

Bull case (strengthened). The June bull rested on a tariff-fattened spread plus an aluminum option. Q2 converts both from thesis to evidence: spread +22% delivered a record 3.7M-ton quarter at 90% utilization against an 81% industry rate; aluminum shipments more than doubled sequentially with automotive qualifications (5182/5754 achieved, 6000-series in trials) inside twelve months of first production — a genuinely fast qualification cycle; capex has rolled off ($124M in Q2) so FCF turned on ($304M in one quarter) exactly as the June dossier predicted it eventually would. Fabrication backlog +45% into Q1 2027 with the Dodge Momentum Index +30% YoY gives an unusually long line of sight on non-resi. And the whole thing now trades at ~19x a base FY26 that is already 47% banked. The earnings-surprise vector is unchanged but better-supported: aluminum inflecting faster than modeled, into a US aluminum-sheet market with a stated structural deficit >1.4Mt.

Bear case (re-weighted, one risk demoted and one promoted).

1. Promoted — tariff erosion beats tariff reversal. The June bear worried about a legal or political rollback of Section 232. The observed risk is subtler and already happening: three Asian exporters "shipping through the 232s" at accelerating rates, imports up MoM, July up noticeably on June, and Nucor pausing spot hikes on import risk. A 50% tariff that is nominally intact but economically porous produces exactly the spread compression the reversal scenario predicted, without any headline to trade on.
2. Held — the 2027–28 capacity wave and the called price peak. Morgan Stanley's HRC path ($1,155 → $1,012 → $900) is the sell-side's version of the June bear point, now with numbers. Tariff-fattened margins summon supply; STLD's own 650kt of aluminum is part of the wave.
3. Demoted — scrap tightening. Management reports ample supply, sideways pricing, and new entrants adding DRI rather than competing for the same scrap unit. This was the June dossier's chokepoint #1; on current evidence it is not binding.
4. New — key-person and structure risk. A founder-metallurgist hands a mid-ramp operating company to a career CFO while moving the senior operator to a non-P&L seat. The strategy will not change; the question is execution cadence on a mill that just missed its own quarterly guidepost.
5. New — the BlueScope overhang. Four approaches, ~$11B, still not withdrawn. A new CEO who has been the deal architect as CFO, sitting on $488.7M of buyback authorization and rising FCF, could plausibly pivot capital from returns to a large cross-border acquisition. That is not a prediction; it is an unpriced optionality that cuts both ways.

Pre-mortem (18 months out — late 2027 — thesis broke). Section 232 was never repealed; it simply stopped working. Asian exporters kept absorbing the duty, import share climbed back toward 20%, and HRC drifted to ~$950 while domestic mills — STLD included — chased volume to defend utilization. The aluminum mill hit 90% capacity on time but ran at a mix and yield that produced only breakeven earnings, because can-sheet qualification took longer than automotive and the relocated Mississippi slab center did not commission until late 2027. Fabrication's higher-priced backlog landed into a non-resi market that finally rolled with rates. The new CEO, needing a growth story to replace a fading spread, went back to BlueScope at a price the market disliked, and the buyback stopped. EPS printed ~$9 for 2027 and heading lower, and a ~19x multiple re-rated to 11x — a stock in the $110s. Note what is absent from this pre-mortem: any failure of the operating model. The moat performs in every branch. That is both the reassurance and the whole problem.

Are multiples too high? June said "reasonable only if you believe the tariff-fattened spread is the new normal." That framing survives but the burden lightened materially: at ~19x base FY26 and ~20.6x base FY27, you no longer need the spread to be permanent — you need it to fade slowly. The EV/EBITDA cross-check (~11x trailing) is the discipline on that optimism.

Contrarian view (updated). June's contrarian take was "buy the tariff-scare drawdown, not the euphoria high." Half of that has now happened — the stock is 5% off its high on no thesis damage — and the more useful non-consensus point has shifted. The market is watching the wrong tariff variable. Consensus (and the June dossier) treats Section 232 as binary and durable-through-2027, and therefore treats the spread as safe until a legal or political event. The Q2 call says the erosion is already underway, monthly, through commercial channels no proclamation controls. If that is right, the spread degrades continuously with no catalyst to sell on — and the correct instrument for owning STLD is the aluminum ramp and the buyback, whose value does not depend on the tariff at all, priced on a steel business you assume normalizes. That is a materially different reason to own it than "quality compounder in a tariff super-cycle."

Devil's Advocate (short-seller) — updated

Q2 handed the short two new pieces of ammunition and took one away.

• The best short argument is now management's own transcript. You do not need a bearish analyst; you need page 236: "We are seeing certain countries shipping through the 232s… very high accelerated rates… These are disruptive," and the remedy proposed is that "we are hoping the administration… try to rein it back in." A thesis whose central profit driver depends on hoping for a policy adjustment is not a moat, it is a lobbying position. Nucor pausing spot hikes is the second mill confirming it.
• The valuation re-rating is doing more work than the fundamentals. Yes, EPS estimates should rise. But EV/EBITDA at ~11.4x trailing on Q2-annualized adjusted EBITDA is above where this cohort has historically traded, and the P/E looks cheap partly because ~$700M/yr of buyback shrinks the denominator while aluminum's D&A depresses the numerator. Strip the buyback and normalize the spread and the "19x" becomes considerably less attractive.
• The aluminum bet keeps sliding — quietly. Q2 shipments missed the company's own 60–70k guidepost. The second slab center was relocated mid-build at a $16M write-off and a $10–20M capex increase, its commissioning pushed to Q4 2026 with 100% utilization only in H1 2027, and its scrap feed logistics degraded (West Coast UBCs from 100% to 60–70%). Every one of these is small. All of them point the same way. The June dossier said the model's vulnerability is operational ramp; four quarters in, that is still true.
• Concentration is unchanged and still under-discussed. ~96% US, exports ~4% of steel sales, and the 10-K's one named demand chokepoint remains the aluminum customer base. our figures is still empty on the shelf — a year of dossiers and no single-name concentration data has ever been sourced. That is a gap the bull case has never had to answer.
• The succession is a free option for the short. Not because Wagler is unqualified — she plainly is not unqualified — but because every transition creates a window where an incoming CEO rebases expectations, takes a kitchen-sink charge, or buys something. She takes the seat on 2027-01-01, roughly when the HRC fade is modeled to bite, with a rejected $11B acquisition already in her drawer.
• What the short gives back: the scrap-squeeze argument. It was the cleanest structural bear point in June — EAF conversion bids up STLD's #1 input — and management's DRI observation, if accurate, defuses it for several years.
• If growth disappoints 20–30%: base FY27 $13.00 → ~$9.00–10.00; at a cyclical-appropriate 11–13x that is a $100–130 stock, ~55% below $267.79. The asymmetry is less unfavourable than June's read (which implied ~65% downside) because the entry multiple compressed — but it is still asymmetric to the downside.
• Single permanent-impairment scenario & plausibility: continuous 232 erosion plus the 2027–28 capacity wave, arriving as aluminum reaches full volume into a normalizing sheet market. Plausibility: moderate, and higher than June's assessment — because the erosion mechanism no longer requires a policy event to trigger it. It is a live, monthly, observable process, and STLD flagged it themselves.

Lenses carried by reference

• Supply Chain: unchanged in structure; see the previous dossier. Two footnotes from this quarter, neither structural: the ferrous internal-supply ratio moved 66% → 65%, and the scrap chokepoint is materially less binding on current evidence (Lens 3/12 above).
• Comps: not re-run; carried with a staleness warning. The June peer table rested on a Sep-2025 EV/EBITDA tiering and had n/a in most cells. That data is now ~11 months old and struck at very different prices. The tiering conclusion (STLD/NUE quality premium above CLF/CMC) is directionally carried; every absolute peer multiple in the June table should be treated as n/a as of this date. Refreshing this lens properly needs a peer-data pull that this run did not have.
• Management Questions: the 15 questions in the previous dossier stand, with #4 (aluminum pause root cause) and #11 (succession) now partly answered by events. Three questions this quarter adds, in information-value order:

1. Quantify the import leakage — how many tons, from which three countries, at what landed price after the 50% duty, and what is your estimate of its effect on Q3/Q4 realized spread? What specifically are you asking the administration for (quotas, 301 stacking, both)?
2. Ms. Wagler — the aluminum ramp is an operating problem and you are a finance CEO taking over mid-ramp, while Mr. Schneider moves to a role without a P&L. Walk us through how operating decisions get made after 2027-01-01, and what Mr. Schneider's CTO mandate actually controls.
3. The SGH/BlueScope proposal has been made four times and is reportedly still live. Is it? What is the maximum share of capital you would commit to inorganic North American growth versus the buyback, and does the answer change when you become CEO?