Round 1 - IG4 Capital / Shine I FIP

Cutoff 2026-08-15. [V] verified in local primary evidence; [I] inference.

Post-run correction, 16 August 2026. Q2's low cash conversion was depressed by a price/inventory working-capital reset as well as the loss of financing arrangements. If naphtha prices and inventory volumes stabilize, that operating outflow need not recur even before the LC book is fully restored. The structural financing need remains. The refreshed mediator estimate is 15% upfront warrants, stepping to 22.5% on objective performance misses; the original valuation tables below are preserved as the cases tested during Round 1.

1. Situation

[V] Capital: 451,668,652 ON + 345,060,392 PNA + 478,790 PNB = 797,207,834 sh (ITR n.24.1). Authorized 1,152,937,970 → headroom 355,730,136 sh (44.62%; max 30.85% of a fully-authorized post-money); ON headroom only 83,993,079. Anything larger needs an AGE - an ON majority - which only Shine (50.1108% of ON) delivers. Shine 273,628,795 = 34.3234%/50.1108% ON; Petrobras 288,188,691 = 36.1498%/47.0316% ON; together 97.14% of voting; float 29.53% (ADR 13.24%, other 13.14%, NSP 3.14%).

[V] Debt (deck p.030, 30-Apr): US$9,497m, cash interest US$665m/yr (7.0% blended); bonds US$7,018m face; LCs US$1,308m; RCF US$1,004m due 31-Dec-26. Jun-30: gross US$10.3bn, adj. net US$9.5bn, 6.74x; consolidated equity −R$13,087m; going-concern uncertainty; Fitch C / S&P D.

[V] Market marks. BRKM5 R$6.36, BAK US$2.44 (=2 PNA) → FX 5.2131; market cap US$973m. From the release's 2Q26 yields ('28 40.76%→price 62.0; '30 22.29%→58.3; '31 63.0; '33 60.4; '34 61.7; '41 54.4; '50 43.3; hyb 51.0) → blended 58.1c, MV US$4,076m. Implied EV ≈ 0.97 + (5.98 − 0.8) ≈ US$6.15bn = 4.1x 2027E (1,494), 2.05x a 3.0bn case. The fulcrum is in the unsecured debt; equity is a stub option. Argue from that, not around it.

[V] Cash conversion - the central fact. Best quarter since 2021 (PE-naphtha US$773/t vs 387 avg; EBITDA US$1,043m) gave OCF US$383m, recurring cash US$208m, US$160m pre-debt-service, net cash −US$149m: 15% conversion. But the gap is financing, not operations - banks called LCs (US$179m settled Q2, US$136m July, reclassified to loans) against scheduled supplier-LC run-off of US$1,682m (2026) + US$602m (2027) = US$2,284m of creditor-controlled withdrawal. Restore it and the same spread converts to cash. That is my case.

[V] Spread correction. The Q2 deck is stamped "updated as of June 30, 2026"; its 3Q26e US$316/t is June-vintage forecast, not August fact. (A) Persistent: FY26 ≈ US$3.0bn (H1 actual 1,234 + 3Q ≈1,200 + 4Q ≈600), calibrated off the 1Q→2Q response (+US$408/t → +US$852m EBITDA, 65% attributed). (B) Normalization: 2027E 1,494 / 2028E 1,524.

[V] Model tie-out. H2-26 status quo: recurring OCF +978, financial WC −860, strategic capex −148, cash interest −325, amortizations −1,242 = −1,597 → EoP −696. Add back extension 1,242 + LC 860 + PIK 325 = +830 → EoP ≈1,748 vs the deck's restructuring case 1,727. My model ties to p.028.

Stay (~25 Aug) and the Valor leak. [V] 60-day stay 26-Jun, Ch.15 recognition 30-Jun, default from July, obligations reclassified current; [I] renewal likely, not certain. [V] On 31-Jul Braskem neither confirmed nor denied controllers' unofficial RJ consent and confirmed creditor proposals involving capitalization and security over assets. Both readings live: real - RJ genuinely dominates a bad consensual deal and the cliff forces contingency planning; leak - "albeit unofficially and without reflecting their actual preference" is exactly how a controller signals credible willingness to file without owning it. My read 70/30 leak - but I must make it true enough to be credible: a filing-ready package before 25 August.

2. Capacity - honest

IG4's ~US$1bn AUM is not uncalled capital for Braskem. Realistic: in-vehicle uncalled [I] US$100-200m; syndication to Brazilian institutionals and the NSP creditor banks [I] US$150-250m; new fund capital needs 6-12 months I don't have.

Constraint not in my brief [V]: the June 5 filing commits Shine to an OPA under art. 254-A LSA / bylaws art. 10 / CVM Res. 215 for up to all outstanding ON and PN, registration denial being a resolutive condition of the CCVJ (Petrobras waived tag-along 23-Apr). Exposure ≈ 235.4m sh × R$6.36 = US$287m at market, competing directly with new money. Consideration paid was 820,886,385 NSP debentures (3/share), so the reference price may be far lower [I] - but the contingency is real.

Fund US$200m at close; backstop US$75-150m; syndicate the rest. Hard ceiling US$350m, and only against Petrobras at ≥1.25×.

3. Leverage

Real: 50.1108% of ON delivers or blocks the AGE that any capitalization >355.7m shares requires; the 3-Jun-26 shareholders' agreement makes control consensual, so creditors must win both of us; the EJ/RJ filing is a board decision I co-control; art. 171 preemption (incl. §3, convertible debentures) blocks cheap creditor placements outside art. 172; I supply the turnaround team; and time - RJ kills the LC market.

Not real: no creditor votes, no defence against cram-down. [I, legal, unverified locally] Under Law 11.101 as amended by 14.112/2020 creditors may propose an alternative plan (art. 56 §4) converting debt to equity with art. 171 disapplied - a live risk. Also weak: art. 170 §1 - with parent equity −R$12,586m, almost any positive issue price defends against "unjustified dilution."

4. BATNA, quantified

Stake today: 273,628,795 × R$6.36 / 5.2131 = US$334m (Petrobras US$352m; minorities US$287m). Basis unknown and probably low - paid in NSP debentures bought from the NSP banks at an undisclosed discount [V structure, I price]. Do not argue from basis; creditors will destroy me on it.

Equity = max(0, 6.0× EBITDA − pro-forma net debt), US$m:

Brutal conclusion: at US$1.5bn EBITDA I am strictly better off crammed down (275-315) than winning a no-haircut extension (−154 net). Deleveraging beats percentage. My strategy is a levered bet on EBITDA ≥ ~US$2.0bn. So my real red line is not "no equitization" - it is no equitization at a trough price and none I'm not paid an option for; I should demand a contingent equitization, not refuse one. What genuinely justifies fighting is franchise: IG4 losing control in year one of a flagship control deal is a fund-level catastrophe independent of the mark.

5. Opening package - "Bridge, Burden-Share, Warrant Ladder"

Post-deal liquidity (US$m): EoP 2026 2,087 (1,727 −190 cash int +550) / 2027 1,805 / 2028 1,443, plus the US$1.5bn LC → total US$2.9-3.6bn; persistent case adds ~US$1.0-1.5bn.

Pro-forma Dec-28 leverage (gross 9,497 + ~1,020 PIK = 10,517; net 9,074):

Honest: at 1.5bn this structure does not work - which is what the warrant ladder and contingent equitization are for.

Creditor NPV test (representative '30, mkt 58.3): my package = 68.9 @12.3% / 59.6 @15% / 53.1 @17.3%; the June-25 proposal was 21.1 @22.3% / 35.8 @15%. Positive-NPV against the mark below ~15%; June's was destruction at every rate.

6. Dilution matrix

Rights/convertible US$550m @ R$6.00 = 477.9m new sh (+59.9%); post-RO 1,275.1m.

Non-subscribers lose ~44% of their percentage (34.32→19.31) vs ~10% for full subscribers (34.32→30.89). Percentage dilution ≠ value destruction: a subscriber pays R$6.00 for shares worth materially more at 2.0bn EBITDA; a non-subscriber suffers no value transfer only if the price is fair - which is why R$6.00, not a trough price, is non-negotiable.

Mechanics. Art. 171 §1(b): preemption within class first, then cross-class for leftovers; issue in current class proportion (ON 56.66% / PNA 43.28% / PNB 0.06%) so no class is advantaged. Art. 171 §6: rights freely assignable - list on B3 with a ≥30-day window plus a sobras round with proportional allotment. PNA keeps its priority non-cumulative 6% dividend on unit value (R$8,043m/797.2m = R$10.09 → R$0.605/sh) and Class A's equal claim on residual profit and on capitalization of reserves; new PNA must carry identical rights. Art. 172 exclusion of preemption is not used - my fairness anchor and my defence against a creditor placement.

BAK/ADS (13.24% of capital; 30.60% of the PNA class). Rule 801 unavailable - US holders far exceed the 10% test [I] - and F-3/F-1 registration is impractical on this timeline. Realistic outcome: the depositary sells rights on B3 and remits net proceeds; BAK holders are cashed out of their subscription right and diluted to 16.61%. Mitigants I commit to: the ≥30-day rights window so rights price properly; sobras open to the depositary; Shine will not buy depositary-sold rights below the rights' VWAP; identical economics if any US tranche is registered. Not perfect - but creditors cannot use it as a cudgel while proposing a placement with no minority participation at all.

7. Brazil political economy (a binding input)

• Election: 1st round 4 Oct, runoff 25 Oct - ~7 weeks out. A state-controlled company writing a large cheque into a distressed private petrochemical group in the campaign's final weeks is radioactive: TCU/CVM scrutiny, Lei 13.303 and CVM Res. 81 related-party review, an independent laudo, minority challenge. So Petrobras's contribution must be below the "statization" threshold, arm's-length and pro-rata on terms identical to Shine and every minority (my art. 171 structure delivers exactly that), and either signed by ~late September or deferred past 25 October. I will offer Petrobras a deferred-funding option to Nov-2026 with Shine bridging at cost.
• Petrobras's public "no statization / no debt consolidation" stance [I] caps its cheque and pushes burden onto me. Defend my ceiling by ratio (Petrobras ≥1.25× Shine), never absolute number.
• Industrial policy raises my equity option - price it into warrants, don't give it away. [V ITR] PRESIQ (Law 15,294/25) effective 2027-31; definitive 5-yr PE antidumping duties vs US/Canada (GECEX 876/2026); 20% resin import tariff (GECEX 800/2025) expiring 16 Oct 2026; [V deck] REIQ/PRESIQ Insumos US$305m (2026), ~US$225m/yr 2027-31, Investimentos US$114m/47m/45m; [I] the Mar-2026 REIQ uplift to 5.8% (LC 228) sits in the same package. The 16 Oct expiry falls between the two election rounds - renewal is likely on employment politics but not automatic, and is worth roughly the gap between 70% and 75%+ Brazilian utilization. I propose a warrant step-up trigger tied to non-renewal, so creditors are paid for that specific risk instead of pricing it into a demand for control today.
• Employment/regional politics. Camaçari (BA), ABC Paulista (SP) and Triunfo (RS) are unionized, politically salient clusters; Transforma Rio (220 kt/yr from 2029, ~US$900m capex 2026-30) is BNDES-linked and federally favoured. The AHG's demand [V] that it be funded from internal cash rather than BNDES asks Braskem to fight the development bank and three state governments seven weeks before an election. Not winnable; I'll say so.
• Alagoas is a political liability, not a negotiable claim. [V] Provision R$3,247m (1,024 current + 2,223 non-current) ≈ US$623m plus R$1,168m payables; R$1.2bn State Agreement (10-Nov-25); H1-26 outflow R$479m; Braskem lists Maceió compliance among its non-negotiable values. Any package appearing to subordinate reparations to bondholder collateral draws the MP, ANM and the Alagoas courts - the fastest route to judicial intervention. The carve-out de-risks creditors' own plan.
• Brazilian courts protect large industrial employers in RJ, worsening the creditors' BATNA: long stays, plan deference, multi-year tolerance, reluctance to displace management at a systemic employer with a sovereign-adjacent shareholder. Add the certainty that US$1.3-1.5bn of LC capacity disappears entirely for an RJ debtor, plus Idesa's parallel Ch.11, and RJ is a 2-4 year value-destroying process the 58.1c mark already partly embeds. Consensus is worth more to creditors than to me - the asymmetry I own.

8. Rejection terms

1. Shine <35% of ON, or loss of the shareholders' agreement / Shine's CEO-CFO-transformation nominations, absent ≥US$3.0bn principal cancellation.
2. Creditor equity >35% without ≥US$2.5bn principal cancellation.
3. Issuance to creditors below R$6.36 without a simultaneous, identically-priced art. 171 offering open to all classes including the depositary.
4. Automatic conversion on technical default, or a DIP converting on covenant breach; contingent equitization must be leverage-tested at a fixed date with a price floor.
5. Blanket first liens on core Brazilian crackers or Braskem America opcos; any pledge of Shine's own shares.
6. IG4 cash >US$350m, or any amount without Petrobras at ≥1.25×. 7. Cash interest >US$450m/yr before Jan-2029. 8. Consolidation of Idesa's US$2.1bn, or new cash to Idesa beyond US$34m. 9. Sweep below US$1.0bn liquidity, or any sweep/lien reaching the US$250m Alagoas reserve.
7. Any deal with <US$1.3bn committed LC capacity - without it no maturity or coupon relief converts spread into cash, and I file instead.
8. Any sponsor/monitoring/backstop fee to IG4 or Petrobras (self-imposed, to kill the leakage argument).

9. Strongest argument against me - and my answer

Them: "IG4 paid no cash - it swapped discounted Novonor bank claims for 34% of a company with −R$13.1bn equity, a going-concern qualification, C/D ratings and bonds at 58c. Its US$200-275m is a rounding error against ~US$4.3bn of creditor value already lost, and it wants 82.5% of the upside. And Q2 refutes its thesis: the best spread quarter in five years produced US$160m pre-debt-service and net cash still fell US$149m. Cram it down."

Answer. (1) Concede the fact, relocate the cause: the US$660m EBITDA-to-OCF gap was you withdrawing trade finance - US$179m of LCs called in Q2, US$136m in July, against US$2,284m of scheduled run-off. Restore US$1.3-1.5bn and, at US$1,200m quarterly EBITDA, pre-debt-service cash reaches ~US$750m/quarter (~US$3.0bn/yr). The LC facility is worth more than the entire interest concession (US$665m/yr) - and it vanishes completely in RJ. (2) My basis is the wrong question: Petrobras cannot inject unilaterally seven weeks before a national election; Novonor is in RJ; you do not want to own a Brazilian cracker complex carrying Maceió and a Mexican Ch.11. I alone deliver the AGE, the board, the team and Petrobras consensus in weeks. (3) It isn't 82.5% free: US$550m of permanent, non-repayable, loss-absorbing junior capital with preemption for everyone; zero sponsor fees; a binding leakage prohibition; a coupon increase; cross-guarantees; a collateral trigger; warrants to 22.5%; and pre-agreed US$1.5bn contingent equitization if leverage exceeds 5.0x at Dec-28 - paying you precisely where my thesis fails. At 1.5bn you get your deleveraging anyway, later but without RJ's destruction; at 2.5bn you get par plus step-up, ~70% above 58c. (4) I'll name my own weakness first: my numbers say that at US$1.5bn I am better off crammed down (275-315) than winning this extension (−154 net). My resistance is a bet on ≥US$2.0bn, not ideology. Price the option and we deal today. Take it for free and I spend the seven weeks to 4 October making an RJ real, before courts that protect large employers, against a stay already granted once - and your LC market will not survive the experiment.