ROUND 1 - AD HOC GROUP / BONDHOLDER STEERING COMMITTEE

Braskem S.A. - Financial Creditors' Independent Analysis and Opening Position Date: 15 August 2026. Information cut-off: 15 August 2026. Prepared without reference to any other party's position paper.

Post-run correction, 16 August 2026. The original creditor framing overstated the persistence of Q2's US$547M working-capital outflow. Higher prices and inventory volumes are mainly a level reset if naphtha and volumes stabilize. The durable creditor concern is the loss of supplier finance, LC runoff and maturity pressure. The H2 operational release in the June plan is US$518M, not US$548M. The refreshed mediator estimate uses 15% upfront warrants, stepping to 22.5%, plus a monthly working-capital bridge and cash sweeps.

Evidence convention: [V] = verified against a primary document with citation. [I] = inference/model built by us on stated assumptions. [L] = legal proposition under Brazilian Lei 11.101/2005 (as amended by Lei 14.112/2020), Lei 6.404/76, or US Bankruptcy Code, not verifiable inside the local document set.

0. EXECUTIVE POSTURE

Two facts govern this negotiation and they point in opposite directions. We state both, because we intend to be believed on the numbers even where they cut against us.

1. The equity is out of the money on the company's own plan, and the company knows it. Braskem's own June deck projects mid-cycle EBITDA of US$1,494M (2027E) and US$1,524M (2028E) [V, june25ocr/023]. Its own restructuring case - with a full five-year extension, 100% PIK and no shareholder money - still leaves gross financial debt rising from US$9,696M to US$10,752M and leverage at 8.3x–8.5x gross debt+WK/EBITDA in 2027–28 [V, june25ocr/028]. That is not a restructuring. That is a deferral financed by us.

2. Our BATNA is genuinely poor and our bonds already trade at ~58 cents. A Brazilian RJ on our own arithmetic returns roughly 35 cents of present value (Section 4). We are not going to pretend otherwise inside our own analysis. Our leverage comes from asymmetry of pain, not from the attractiveness of our alternative: an RJ costs us ~23 points against the current mark; it costs Petrobras and IG4 100% of a ~US$1.0B equity stake, IG4's entire control thesis, Petrobras's Camaçari/Rio offtake relationship, and 30 months of political exposure over Alagoas.

The negotiation is therefore not "how much do creditors give up." It is: who funds the US$3,686M of debt service falling due between July 2026 and December 2027 [V, june25ocr/005], plus the US$1,326M January 2028 bond maturity [V, june25ocr/005], against US$795M of unrestricted cash at June 2026 [V, june25ocr/006]. Someone must write a cheque. The shareholders have not offered one. Until they do, every term the Company has proposed is a request that impaired creditors finance an equity option.

1. INDEPENDENT SITUATION ASSESSMENT

1.1 The verified financial position

Braskem is in payment default. In July 2026, following expiry of the cure period on unpaid financial obligations of R$507M (US$98M), "the Company was in default under certain financial instruments" [V, q2itr]. All non-current borrowings reclassify to current from July 2026 because the Company "no longer had the unconditional right to defer settlement... for a period exceeding 12 months" [V, q2itr]. Separately, letters of credit of R$929M (US$179M) and a further R$703M (US$136M) in July were settled by the issuing banks and the reimbursement obligations reclassified from Trade Payables to Borrowings [V, q2itr notes 14 & 15]. That is US$315M of trade credit that has already converted into financial debt ranking alongside us.

1.2 The Q2 EBITDA is real; the Q2 cash flow is not

We are told to weigh the spread windfall. We have. Here is the arithmetic, and it is the most important table in this document.

In the best operating quarter Braskem has had in three years, with a war-driven PE–naphtha spread of US$773/t versus a 2022–25 average of US$387/t [V, q2pres], actual cash still fell US$149M. That bridge includes a working-capital reset and should not be treated as a steady-state conversion rate.

Conversion: 1,043 → 210 = 20.1%. We reconcile the US$833M gap: working capital (547), maintenance capex (133), leasing (54), Alagoas (49), other (52), less a net tax inflow (18), plus Q2 cash interest of 121 sits below the "recurring cash generation" line [I, reconciled against june25ocr/024 monthly detail: capex 41+43+49, leasing 19+19+16, Alagoas 16+16+17, interest 67+38+16]. The reconciliation ties to within ~US$16M.

1.3 Separate the one-time operating reset from structural financing pressure

This is where we are supposed to be sceptical, so we are. 1H26 balance-sheet movements [V, q2itr]:

This is a textbook trade-credit run. Payables are collapsing while the company is being forced to prepay suppliers (advances tripled). Reverse factoring is gone - "the Company was no longer able to access certain payment agreements with financial institutions, including supplier finance arrangements (reverse factoring)" [V, q2itr]. And the LC book that finances raw-material purchases is amortising: US$1,308M runs off entirely by March 2027 (Jul-26: 306; Aug: 100; Sep: 166; Oct: 57; Nov: 67; Dec: 75; Jan-27: 265; Feb: 86; Mar-27: 152) [V, june25ocr/031]. LC-backed raw-material payables have already fallen from R$7.8B to R$5.9B [V, q2itr note 14].

Two conclusions, both of which we will use.

• Against the equity: Q2 did not convert fully into free cash, but the US$547M outflow should not be repeated mechanically. Higher product and feedstock prices and higher inventory volumes are mainly one-time balance increases if naphtha and volumes stabilize. The defensible creditor argument is that management may not receive its full projected US$518M H2 operational-WC release and that the financing side remains impaired, not that another US$547M operating outflow must recur.
• For our price: a material part of the liquidity recovery remains deal-contingent. Payables rebuilding to Dec-25 levels is +US$459M; supplier advances unwinding is +US$209M; inventory days normalising from ~71 to ~60 is +US$300M [I, days from june25ocr/026]. Inventory and receivables can normalize without a creditor deal, but rebuilding supplier terms and replacing the LC book require financing support. Creditors can demand a monthly working-capital bridge, borrowing-base controls and excess-cash sweeps in return.

1.4 Spreads: both company vintages are advocacy, and they contradict each other

We were instructed to correct for the June vintage of the Q2 presentation. We do so - and we note that the correction runs in both directions, which the Company has not disclosed.

Between the June deck and the Q2 presentation the Company halved its own 3Q26 spread forecast, from 627 to 316 - to a level below the 2022–25 average and below 1Q26 actual. The 3Q26e of 316 is a June-vintage consultant base case, not observed August fact. Public Platts reporting for the week ended 22 July showed Americas PE prices stable to rising, and the principal's mid-August observation is that spreads held. We therefore treat 316 as an advocacy number and will not negotiate against it.

Equally, we refuse to annualise 2Q26. Actual Braskem EBITDA, 2021–2025: 5,024 / 1,796 / 612 / 845 / 508 [V, june25ocr/023]. Five-year mean US$1,757M; four-year mean ex-2021 US$940M. Fifteen years of this business says the mid-cycle is nearer US$1.5B than US$3B, and the Company's own plan agrees.

Our two operating cases [I]:

• (A) Persistent spreads. Regressing Q2 vs Q1 (ΔEBITDA US$852M on ΔPE spread US$408/t → ~US$2.09M of quarterly EBITDA per US$1/t): 3Q26 at ~700 → ~US$890M; 4Q26 at ~600 → ~US$680M. FY26 ≈ US$2.8B (range 2.7–3.3B). Crude but disciplined.
• (B) Normalisation (consultants). 3Q at 316 → ~US$90M; 4Q at 412 → ~US$290M. FY26 ≈ US$1.6B.
• Management's own June plan: FY26E = US$2,166M, 2027E 1,494, 2028E 1,524 [V, june25ocr/023].

Note carefully: 1H26 actual EBITDA was US$1,234M (1Q 191 + 2Q 1,043) against the June deck's 1H forecast of US$1,229M [V, june25ocr/022]. Management's plan is tracking to within 0.4%. We will hold them to it. And their plan, at US$2,166M of 2026 EBITDA, still shows unrestricted cash at negative US$821M by Dec-26 and negative US$1,981M by Dec-27 [V, june25ocr/025]. A good year does not fix this. That is the whole point.

1.5 What actually happens when the stay expires ~25 August

The São Paulo 2nd Bankruptcy and Judicial Reorganization Court granted, on 25/26 June 2026, a stay of enforcement actions and attachments by creditors invited to the Wind Mediation Chamber mediation, for 60 days [V, q2release §4.9.2; q2itr note 1]. On 26 June the same entities filed a Chapter 15 petition; on 30 June the US court granted a preliminary stay "for the same period as the Protective Injunction in Brazil, pending its final decision on recognition" [V, q2itr note 1]. Sixty days from 26 June is ~25 August 2026 - ten days from today.

On expiry, absent new relief [L/I]:

1. Brazilian enforcement resumes. Execução, penhora and arresto by invited creditors become available. Braskem is already in payment default [V]. Non-invited creditors were never stayed at all.
2. The Chapter 15 preliminary stay lapses unless final recognition is granted. We should oppose recognition. A tutela de urgência cautelar ancillary to a private mediation is arguably not a "foreign proceeding" under 11 U.S.C. §101(23) - it is not a collective judicial or administrative proceeding under a law relating to insolvency in which the debtor's assets and affairs are subject to control or supervision by a foreign court for the purpose of reorganisation or liquidation. If recognition is denied, §1520's automatic stay never attaches and Braskem America Inc.'s US assets and Braskem America Finance Company (issuer of the '41s) are exposed [L]. This is a real and underpriced piece of our leverage and we will brief it.
3. Acceleration. Payment default beyond grace is an Event of Default; holders of ≥25% of a series may accelerate under standard NY-law indentures [L]. Acceleration cross-accelerates the RCF, the LC facilities, the ECAs and the Brazilian debentures. The ITR already concedes the reclassification to current [V].
4. Suit on the New York guarantee. Braskem S.A. has "fully, unconditionally and irrevocably guaranteed the bonds" [V, q2itr note 15(c)]. We sue the guarantor in New York, obtain a money judgment, and use it (a) to attach US assets, (b) to fix a liquidated claim for any Brazilian proceeding, (c) to make the Company's US operations unbankable.
5. The Company's counter-move is already telegraphed. Deck p.011 shows the plan: file an EJ with at least 1/3 support to obtain a 90-day standstill [V, june25ocr/011]. Expect an EJ filing in the window 20–31 August. We should assume it and pre-position.
6. The 60 days already burned are not free. Under Art. 20-B §1 LRF (as amended in 2020), the pre-insolvency mediation stay is capped at 60 days and is deducted from the 180-day stay period of any subsequent RJ [L]. If Braskem files an RJ, it gets roughly 120 days, not 180, before the stay lapses and - critically - before creditor plan-proposal rights vest. Management's runway is shorter than it thinks.
7. Valor Econômico, 29 July 2026: "After Another Failed Round of Negotiations, Braskem Moves Closer to Judicial Reorganization," reporting that RJ "would already have the consent of its principal shareholders, albeit unofficially." B3 demanded confirmation or denial. The Company did not deny it [V, bak20260731_6k]. It confirmed only that creditor groups had delivered non-binding proposals "including a possible capitalization and the granting of security interests over assets." Both of the things the Company said in June it would never do - capitalization and collateral - are now on its own disclosed table.

1.6 Can an EJ bind us, and at what threshold?

Yes, and the threshold is >50%. [L, Art. 161–167 LRF as amended by Lei 14.112/2020]

• An EJ plan homologated by the court binds all creditors of the affected class(es) if signed by holders of more than 50% of the credits of each affected species/class (reduced from the pre-2020 three-fifths).
• Art. 163 §7 permits filing with ≥1/3 support and a 90-day period to reach >50%. This is exactly the Company's disclosed timeline [V, june25ocr/011].
• An EJ cannot bind labour claims, tax claims, ACC/export-advance claims, or fiduciary owners/lessors [L, Art. 161 §1 → Art. 49 §3].
• The debtor chooses which classes to affect. The Company's proposal affects "financial creditors (capital markets and banks / financial institutions)" with "equal treatment among all unsecured creditors" [V, june25ocr/009].

The arithmetic of the threshold is the entire ballgame.

If all financial claims sit in one class, senior bonds are 62.7% and we hold a decisive blocking position on our own. If the Company splits the class - bonds separately from the BRL debentures/CRAs, or LC banks into their own class fed by the New LC Facility - then it can assemble >50% in the bank classes and isolate us. Forcing a single financial class is our first tactical objective and a condition precedent to any support agreement.

1.7 What an RJ does to us

[L] Recuperação judicial under Lei 11.101/2005 as amended:

• Stay: 180 days from deferimento, extendable once, less the 60 days already consumed → ~120 days effective.
• Plan: debtor exclusivity to present a plan within 60 days. If the debtor's plan is rejected, or the stay lapses without approval, creditors may present an alternative plan (Art. 56 §§4–7, the 2020 reform), supported by creditors representing >25% of total credits, or >35% of credits present at the AGC. We alone clear both thresholds. This is our cram-up tool and it is the single most important change in Brazilian insolvency law for our purposes.
• Voting: by class. Class III (quirografários) requires both a majority by value and a majority by head count of creditors present. We dominate by value; head count is the risk (thousands of small trade creditors). Bondholders can be counted individually through the indenture/DTC chain if organised - we must organise it early and expensively.
• Cram-down (Art. 58 §1): >50% of all credits present regardless of class, approval by the requisite majority of classes, and >1/3 support within the dissenting class - plus no differential treatment within the dissenting class (§2).
• Tax claims are extraconcursal. Alagoas environmental/remediation obligations largely survive the plan. Labour claims (Class I) get priority up to 150 minimum wages each.
• DIP (Art. 69-A to 69-F): court-authorised DIP financing is extraconcursal and can be secured by the debtor's assets. Every material Braskem asset is unencumbered today - "Except for certain reserve accounts as disclosed in Note 5 (ii), Braskem's borrowings and debentures above consist of unsecured obligations" [V, q2itr note 15]. We can therefore be primed. We can also be the primer. The correct response to being unsecured in a jurisdiction with statutory DIP priority is to be the DIP lender, not to complain about it.
• Timeline: for a group of this scale with 47% Petrobras voting ownership, an Alagoas overhang, Camaçari and Rio employment exposure, a parallel Braskem Idesa Chapter 11, Chapter 15 recognition litigation and cross-border guarantee structure - 24 to 36 months to plan effectiveness and instrument delivery. We model 30 months.

1.8 Chapter 15 / New York law leverage

• All USD bonds are issued by Braskem Netherlands Finance B.V. (or Braskem America Finance Company for the '41s) and guaranteed by Braskem S.A., NY-law, full and unconditional [V, june25ocr/030; q2itr note 15(c)].
• The hybrid '81 is different: "As for the issuance carried out in 2020, in case of default, the guarantee comprises obligation subordinated to all Braskem's current [and future senior debt]" [V, q2itr note 15(c)]. The '81 guarantee is contractually subordinated. We will insist it is classed and treated as junior. We will not let US$247M of structurally junior paper vote in our class.
• Our leverage set: (i) oppose Chapter 15 recognition; (ii) accelerate and sue on the NY guarantee; (iii) attach Braskem America assets; (iv) discovery in aid of execution into intercompany flows to Braskem Idesa and to shareholders; (v) if an EJ is filed, oppose homologation in Brazil and oppose §1521 discretionary relief in the US on the ground that an EJ that impairs NY-law debt without a genuine majority in a properly constituted class fails §1506/§1522 fairness.

2. STRUCTURAL AND LEGAL ANALYSIS OF THE CLAIMS

2.1 Who issues, who guarantees (all from june25ocr/030 and /032) [V]

LC facilities (US$1,308M) [V, june25ocr/032]: Committed Facility IV (Syndicate) 402; Committed Facility V (Syndicate) 392; Bilateral Committed 185; Spot LCs 329. Borrower in every case: Braskem Trading & Shipping B.V. Guarantor in every case: Braskem S.A. (full amount).

2.2 The structural map - who sits where

Braskem S.A. (Brazil) is the operating company for the Brazilian crackers and the ultimate guarantor of essentially everything. It is also the direct obligor on the BRL debentures/CRAs (US$658M) and most bilaterals (US$412M).

Braskem Netherlands Finance B.V. is a pure financing SPV. Bondholders' claim against it is worthless standalone; our recovery is 100% dependent on the Braskem S.A. guarantee, which is a quirografário (unsecured) claim in Brazil, pari with the debentures.

Braskem Netherlands B.V. is not the same entity. It is an intermediate holding/operating entity in the European chain and it is the borrower of US$954M of the RCF and of the Nexi 2017 / SACE 2018 ECA loans. The RCF banks are therefore structurally senior to bondholders with respect to the European asset chain, and additionally hold a full Braskem S.A. guarantee. That is a double-dip: opco claim plus parent guarantee.

Braskem America Inc. holds the US polypropylene business and is the borrower of US$50M of the RCF and of the US$74M Euler Hermes facility - the latter expressly "without guarantee from Braskem" [V, q2itr note 15(i)]. Euler Hermes is therefore structurally senior on US assets with no parent recourse; the RCF again double-dips.

Braskem Trading & Shipping B.V. is the trading entity that buys the naphtha and holds the traded inventory and receivables. All US$1,308M of LCs sit here, with a full Braskem S.A. guarantee. LC banks therefore have (i) a claim against the entity that owns the physical inventory, (ii) the parent guarantee, and (iii) - decisively - an operational chokehold: if they stop issuing, Braskem cannot buy feedstock on credit.

2.3 Are bank and bond claims pari passu?

At Braskem S.A., legally yes; economically no.

• Every claim in the table above (except the hybrid '81) is an unsecured quirografário claim against Braskem S.A., ranking equally [V, q2itr note 15: bonds' guarantees "rank equal in right of payment with all of its other existing and future senior unsecured debt"; all borrowings "consist of unsecured obligations"].
• But the RCF and the ECAs at BNL B.V. / Braskem America Inc. are structurally senior on the assets of those entities. On a going-concern EV of, say, US$7.5B, the US and European businesses are a meaningful minority of value; the RCF's structural position is worth real points.
• And the LC banks are practically senior because their exposure self-liquidates: US$1,274M of the US$1,308M matures by March 2027 [V, june25ocr/031]. If nothing is done, the LC banks are simply paid out at par over eight months while our maturities remain outstanding until 2028–2050. We must not permit that.

2.4 Do LC reimbursement claims now rank as loans? Yes.

Verified twice. R$929M (US$179M) of LC-secured trade obligations "were settled at their original maturity dates by the issuing financial institution directly with suppliers, thereby extinguishing the related trade obligations and giving rise to reimbursement obligations for the Company. The Company's reimbursement obligations were reclassified to loans and financing" [V, q2itr note 14/15]. A further R$703M (US$136M) in July 2026, likewise [V, q2release; q2itr]. Total already converted: US$315M.

The implication is that the LC book is a conveyor belt converting trade payables into pari passu financial debt at par, in cash, ahead of us, week by week. As of 30-Jun-26 there remained R$5.9B (≈US$1,140M) of raw-material purchases up to 360 days backed by LCs [V, q2itr note 14]. Every month of delay in this negotiation transfers value from us to the LC banks.

Demand: a July 1, 2026 record date; all LC reimbursement claims arising thereafter are treated as pari passu restructured claims and subject to the same standstill; any cash paid to LC banks after the record date is credited against their restructured claim.

2.5 Can we prime, or be primed?

• Today, nobody is secured. Every asset - Camaçari, ABC/São Paulo, Triunfo, Rio, the US PP plants, the German assets, inventory, receivables, intercompany notes - is unencumbered [V, q2itr note 15].
• We can be primed in an RJ by a court-authorised, extraconcursal, secured DIP under Art. 69-A/69-B, and out of court by any consensual new-money facility taking liens, since no negative pledge in our documents can survive a court order. This is a genuine and underappreciated risk to our position and it is the reason we must not slow-walk.
• We can prime. As the largest single creditor bloc we can offer the DIP ourselves and take the collateral first. Our new-money offer in Section 5 is calibrated to make ourselves the priming party rather than the primed party.
• In any consensual deal, collateral is the cheapest concession the sponsors can give and the most valuable thing we can take. It costs them no cash. It moves our exit yield materially (Section 3.4 quantifies this at ~16 points of price on the '30s). Collateral is our number-one ask, ahead of coupon.

2.6 The real adverse parties are inside the creditor group

We do not have a unified creditor side. We have four conflicting constituencies and the Company's June proposal is explicitly engineered to exploit that.

(a) LC banks - US$1,308M. The most adverse party. They are short-dated (fully amortising by Mar-27), operationally indispensable, structurally advantaged at BTS B.V., already converting to cash at par, and the Company has offered them a New Committed LC Facility of US$1.5B at SOFR+2.0%, eligibility conditional on committing pro-rata new money [V, june25ocr/010]. Read the Company's own words: "a lender may roll its existing outstanding LC claims into the New Committed LC Facility only if it commits to provide its pro-rata share of the US$200 million additional commitment"; and "Any claims (or portion thereof) of non-committing lenders shall be treated as Financial Unsecured Claims." That is a naked pay-to-play designed to buy the LC banks' EJ votes with par treatment funded by our impairment. We must attack this structure directly.

(b) RCF banks - US$1,004M, maturing 31-Dec-26. Structurally senior at BNL B.V. and Braskem America Inc, plus a full parent guarantee. The Dec-26 spike of US$1,041M on the debt-service chart is theirs [V, june25ocr/005]. They will argue their double-dip entitles them to better treatment. It partly does. Our answer: election. Either waive the Braskem S.A. guarantee and take your structural recovery from the Dutch/US chain, or accept pari treatment with us at the parent. You do not get both.

(c) Brazilian debenture / CRA holders - US$658M, BRL, CDI-indexed, direct claims on Braskem S.A. Retail-adjacent, politically protected, short (Dec-28 to Nov-32) and cheap to buy off with a small cash payment. They are the Company's cheapest route to a class majority in any BRL class. Demand: no class-splitting between BRL and USD financial claims; identical economic treatment; no cash sweeteners for BRL paper.

(d) ECA/MLA lenders - US$401M. SACE, NEXI, Euler Hermes, FDNE, Safra Finem. Sovereign-backed, slow, will not lead, will follow the banks. Euler Hermes (US$74M, no parent guarantee) has a genuinely different structural position and should be treated separately.

(e) Hybrid '81 - US$247M. Contractually subordinated guarantee [V]. Out of the money in our base case. Will support anything. Must be a separate class with strictly junior treatment.

And the fault line inside the bond group is real. The '28s (US$1,189M, maturing in 17 months, trading ~61.5) want short paper and cash; the '50s (US$761M, trading ~43.3) and '41s (US$577M, trading ~54.4) are effectively long-dated and are better off with reinstatement plus warrants. A uniform "extend everything by five years" package systematically transfers value from the '28/'30 holders to the '41/'50 holders. We will therefore insist on differentiated treatment by maturity bucket - no principal extension for the '41s and '50s, front-loaded amortisation and higher cash coupon for the '28s. Anyone in our group who insists on pure pari treatment is asking the near maturities to subsidise them, and the group will not survive it. We say this openly to keep the coalition together.

Coalition assessment [I]: The bond AHG holds US$6,778M of senior bonds = 62.7% of the US$10.8B financial pool and 71.4% of the US$9,497M p.030 pool. That is a blocking position against any single-class EJ. It is not a blocking position if the Company splits classes and buys the banks. Our coalition holds if and only if we (i) force a single financial class, (ii) neutralise the LC pay-to-play, (iii) offer the RCF banks a structurally honest election, and (iv) accept internal differentiation by maturity bucket. All four are Round 1 asks.

3. RECOVERY FRAMEWORK, QUANTIFIED

3.1 The claim pool

We correct our own prior framework here: our internal memo used a US$9.5B claim base. That excludes the US$1,308M LC facility and accrued interest, and it overstates recoveries by roughly 15%. We use US$10.8B for point-in-time recovery and US$11.2B for plan-effectiveness recovery. Being wrong in our own favour is not rigour.

In an RJ, the concursal Class III pool is larger still: add trade payables ~US$2,090M (R$10,822M), leases ~US$786M (R$4,068M), and civil/legal provisions ~US$169M (R$875M) [V, q2itr] → ≈US$14.1B of Class III claims, of which financial+LC is 78%.

3.2 EV grid and recovery per US$ of claim

Enterprise value = multiple × normalised EBITDA (Braskem ex-Braskem Idesa; the June plan explicitly excludes Idesa [V, june25ocr/019]).

Enterprise value (US$B):

Priority reserve ahead of unsecured financial claims:

Recovery per US$ of claim, consensual basis (US$1.0B reserve, US$10.8B pool):

3.3 Reconciliation to market - the honest part

Converting the 30-Jun-26 yields [V, q2release §6.2.1] to prices (semi-annual, actual coupon dates from june25ocr/030):

The market's mark on our senior paper is ~58 cents clean.

Cross-check via total capital structure [I]: senior bonds 6,778 @58 = 3,931; hybrid 247 @45 = 111; other financial 2,472 @~60 = 1,483; LCs 1,308 @~80 (short, self-liquidating) = 1,046; equity 797M shares × R$6.36 ÷ 5.1766 = US$979M (cross-checks to BAK US$2.44/ADR ÷ 2 × 5.1766 = R$6.31); less cash US$795M. Market-implied EV ≈ US$6.75B. On US$1.5B of mid-cycle EBITDA that is 4.5x; on US$2.0B it is 3.4x.

Translation: the market is pricing management's own 2027–28 plan (US$1.5B) at roughly 4.5–5.0x and is giving essentially zero credit to the spread windfall. Our central case and the market's are the same case. The BRKM5 equity at US$979M is a call option, not a valuation.

Does this change our willingness to settle? Yes, and we say so plainly. At 58 cents we can sell. Any deal we sign must be worth demonstrably more than 58 cents of present value on conservative exit-yield assumptions, and we have no fiduciary basis to blow up a deal worth 70+ cents in pursuit of an RJ worth 35. Our rhetoric is hostile; our arithmetic is not sentimental.

3.4 What the packages are actually worth (valuing the '30s per US$100 of face)

We value the 4.5% 2030s, our largest single series, as the benchmark instrument. Exit yield is the key variable; we show 11% / 13% / 15%.

(i) The Company's June proposal [V, june25ocr/010]: +5-year extension (→Jan-2035), 100% PIK toggle Jul-26→Dec-28, −200bp coupon (4.5% → 2.5%), unsecured, no equity. Face accretes to 100 × 1.0125⁵ = 106.4 by Dec-28; then 2.5% cash on 106.4 = 2.66/yr for 6.2 years; bullet 106.4 at t = 8.4y.

Against a 57.6 mark, the Company's June proposal destroyed 4 to 18 points of market value and, at any plausible exit yield for an 8.5x-levered unsecured C-rated Brazilian petrochemical credit, is negative-NPV. That is not rhetoric; that is the arithmetic behind our 19 June rejection [V, june25ocr/013], and it remains the arithmetic today. A coupon reduction, as we said then, "is unheard of in a corporate debt restructuring."

(ii) The original mediator reference case tested in this opening: 5-year extension, 2 years at 4% cash + 4% PIK, thereafter coupon +200bp (4.5%→6.5%), 17.5% warrants, unsecured. The post-run mediator refresh lowers the upfront grant to 15% and makes the 22.5% step-up conditional on cash-conversion, liquidity or PIK misses. Face accretes to 100 × 1.02⁴ = 108.24; cash 4% for 2y; then 6.5% × 108.24 = 7.04/yr for 6.4y; bullet 108.24 at t = 8.4y.

(iii) The single most valuable term is collateral, and it costs the sponsors nothing. Moving the exit yield from 15% to 11% on the same cash-flow package is worth +16 points. The 17.5% warrant is worth ~4 points. We will trade warrants for liens all day long and we want the sponsors to understand that trade exists, because it is the deal-clearing insight in this negotiation.

Warrant valuation [I]: reorganised equity = EV − pro forma net debt. At 5x on US$2.0B: 10.0 − 8.5 = US$1.5B → 17.5% = US$262M = 2.4 points on US$10.8B. At 5x on US$3.0B: 15.0 − 8.0 = US$7.0B → 17.5% = US$1,225M = 11.3 points. At US$1.5B EBITDA the warrants are worthless. Probability-weighted ≈ 4 points at 17.5%, ≈ 9 points at 40%.

4. BATNA, QUANTIFIED

4.1 RJ recovery model

Method: RJ recovery = [EV × (1 − process discount) − RJ priority reserve] × (financial share of Class III = 78%) ÷ financial claims of US$11.0B; then discounted 30 months at 15%.

Process discount (15–25%) is not a haircut on a spreadsheet; it is an operating fact. In this case it is already partly observable: payables down R$2,376M, supplier prepayments up R$1,081M, reverse factoring withdrawn, LC book amortising to zero by Mar-27, utilisation at 70%, Fitch C / S&P D. An RJ filing takes the LC book to zero immediately and forces cash-in-advance naphtha procurement. We use 20% central.

Discount factor at 30 months, 15%: 1.15^(−2.5) = 0.705.

Probability weighting [I] - anchored on management's own plan (1,494/1,524 for 2027–28), the consultants' base case ("normalization of spreads"), the 2022–25 realised average of US$940M, and the structural overcapacity the Company concedes through 2027:

We stress-test the timing: at 24 months instead of 30, expected PV ≈ 38c; at 36 months, ≈ 32c. At a 10% process discount instead of 20%, ≈ 43c. At a 25% discount, ≈ 31c. The RJ range is 31–43 cents. Call it 35.

4.2 The indifference point

• Pure economic indifference vs RJ: ~35 cents of PV. We will not say this in the room, but it is what our own model says.
• Practical floor: the market mark. We can sell our position at ~58 cents today. A deal below the mark is value destruction we cannot defend to our own investors, and it invites tendering-out rather than signing.
• Reservation price: 65 cents of PV valued at a 15% exit yield. Below that we do not sign; we accelerate, oppose Chapter 15 recognition, and prepare a creditor plan under Art. 56 §6.
• Target: 75–82 cents of PV at 13% / ~72 cents at 15%.

We state the uncomfortable conclusion explicitly: our threat to force RJ is credible not because RJ is good for us, but because it is catastrophic for the shareholders. In Case A, an RJ leaves the US$979M equity market cap at exactly zero and destroys IG4's entire control thesis and Petrobras's 36.15% economic stake and its integrated feedstock relationship, over a 30-month judicial process with Alagoas, Camaçari employment, and a parallel Braskem Idesa Chapter 11 running in the background. We lose 23 points. They lose everything. That asymmetry is the leverage, and it is entirely sufficient.

4.3 The affordability arithmetic that decides the deal

The Company will say it cannot pay cash interest. On its own restructuring-case numbers [V, june25ocr/028], CFADS is 885 (YTG-26), 131 (2027), 219 (2028), 571 (2029), 606 (2030). At 4% cash on US$9.5B (US$380M/yr), 2027 and 2028 are negative. That objection is arithmetically real - on their assumptions. Two of their assumptions are ours to change.

(a) Kill the strategic capex. The plan deploys "c.US$900m of strategic capex between 2026 and 2030, mostly related to the Transforma Rio project" [V, june25ocr/019], scheduled 168 / 435 / 249 / 29 / 16 across 2026–30 [V, june25ocr/025]. US$852M of that lands in 2026–2028 - precisely the years in which the Company says it cannot pay us cash interest. Removing it:

Braskem can pay a ~5% cash coupon out of its own cash flow, with no shareholder money at all, simply by not building a growth project with impaired creditors' money. We flagged this in June - "the use of internally generated funds rather than BNDES financing for Transforma Rio" [V, june25ocr/013] - and we now quantify it.

(b) Every dollar of PIK is a dollar of equity the sponsors did not contribute. Two years of PIK versus cash at 4% on US$9.5B = US$760M of cash relief. That is not a technical accommodation; it is a US$760M interest-free, unsecured, subordinated equity contribution from creditors to shareholders. We will price it as such: if the sponsors want US$760M of PIK relief, they contribute US$760M+ of permanent, loss-absorbing capital, or they hand over the equity that US$760M would have bought.

With US$1.3B of sponsor equity injected in 2026, cumulative 2027–28 CFADS becomes 350 + 1,300 = US$1,650M, supporting US$825M/yr of cash interest - an 8.7% all-cash coupon. The PIK question is not an operating question. It is entirely a sponsor-funding question, and we will not allow it to be framed otherwise.

5. ROUND 1 OPENING (HOSTILE ANCHOR) AND TRUE RESERVATION PACKAGE

We present the opening as a menu of three paths, because a menu forces the sponsors to price their own equity, and because an undifferentiated hostile number invites an undifferentiated rejection.

5A. THE OPENING ANCHOR

PATH A - DELEVERAGING (our preferred opening)

Conversion. US$3.0B of financial claims (pro rata across all pari classes, at par) convert into 72% of pro forma fully diluted equity. Remaining claims of ~US$8.2B reinstated on Path B terms below.

Why: pro forma net debt falls to ~US$7.0B → 4.7x at US$1.5B EBITDA, 3.5x at US$2.0B, 2.3x at US$3.0B. This is the only path that produces a solvent balance sheet on management's own plan.

PATH B - REINSTATEMENT WITH REAL BURDEN-SHARING (opening terms)

1. Maturity extension by bucket (no uniform five-year extension - we will not have the '28s subsidise the '50s):

2. Cash vs PIK. 5.0% cash + 3.0% PIK for 24 months only (to Sep-2028), then 100% cash. PIK toggle at the creditors' option, not the borrower's. Any PIK period beyond 24 months triggers automatic warrant step-ups (below). Annual cash interest during relief: ≈US$475M (5.0% × ~US$9.5B).

3. Coupon and step-ups. Post-relief coupon = existing coupon + 300bp (weighted 7.00% → 10.00%), floor 8.5%. +100bp for each 6-month extension of the PIK period. +150bp if Transforma Rio capex is incurred while leverage exceeds 3.5x. +100bp if net leverage exceeds 5.0x at any test date after 2028. Coupon reductions are not negotiable in any direction. There will be no coupon reduction.

4. Fees. Consent/PIK fee 2.5% of restructured claims, paid in kind at closing (≈US$270M). Backstop fee 5.0% on any creditor new money. OID 3.0% on new-money instruments. Full reimbursement of AHG advisor fees and steering-committee member expenses from now through closing and, if the process fails, through any RJ, on a current-pay monthly basis [consistent with our 19 June demand, V june25ocr/014].

5. Collateral and lien ranking (our highest-priority ask):

• First lien, new-money tranche: all Brazilian accounts receivable, inventory, deposit and collection accounts and proceeds; Braskem Trading & Shipping B.V. inventory and receivables; pledge of 100% of the equity of Braskem Netherlands B.V., Braskem Netherlands Finance B.V., Braskem Trading & Shipping B.V., Braskem America Inc. and Braskem America Finance Company.
• Second lien, reinstated legacy claims: the same collateral pool, plus Brazilian fixed assets - Camaçari, Triunfo, ABC/São Paulo and Rio industrial complexes, land, plant and equipment - by way of hipoteca and alienação fiduciária where available; the US PP assets by US-law mortgages and security agreements; the German assets by Dutch/German-law pledges.
• Guarantor structure: upstream, downstream and cross-stream guarantees from every material subsidiary in Brazil, the Netherlands, the US and Germany, expressly including Braskem Trading & Shipping B.V. and Braskem America Inc.
• Excluded: Braskem Idesa and any Mexican asset. We take no Idesa collateral and we permit no Idesa liens over Braskem S.A. assets.
• Negative pledge with no general basket; permitted liens limited to purchase-money and ordinary-course operating liens under US$100M in aggregate.

6. New money. Total package US$3.0B:

• US$1.5B committed LC/working-capital facility. We accept the Company's own sizing [V, june25ocr/010] - ~US$1.3B roll of existing LCs plus US$200M incremental. But: eligibility to roll is unconditional, not conditional on new commitments. The pay-to-play in the June proposal is rejected outright. Pricing SOFR+300 (not +200), first lien on the WC pool, 5-year availability.
• US$1.5B of permanent, loss-absorbing shareholder capital (below).
• US$750M creditor new-money term loan, first lien, SOFR+700, 3% OID, 5% backstop fee, 4-year, offered pro rata to the entire creditor class. We would rather be the priming party than be primed.

7. Required sponsor contributions - and they must be LOSS-ABSORBING. This is the test on which the negotiation turns.

• Petrobras: US$750M. IG4/Shine: US$500M. Minorities: up to US$250M, backstopped pro rata by the two sponsors. Total US$1.5B.
• Form: newly issued ordinary equity subscribed in a rights offering open to all shareholders at a uniform price (Lei 6.404/76 Arts. 170–171: preemptive rights preserved, no unjustified dilution, litigation-proof) [L]. Not preferred. Not convertible. Not a shareholder loan.
• If any sponsor insists on debt form, then it must be: (i) contractually and structurally subordinated to every restructured creditor claim, (ii) PIK-only with zero cash pay, (iii) unable to be repaid, redeemed, prepaid, refinanced or amended while any restructured creditor claim is outstanding, (iv) with no lien, no guarantee, no maturity earlier than 12 months after the final restructured maturity, and (v) mandatorily convertible into equity if leverage exceeds 5.0x at any test date. Anything less than all five and it is ordinary debt, not burden-sharing - and our warrant demand rises from 40% to 55%.
• The test we will apply publicly: does this instrument absorb loss before our claims do, in every scenario, including a subsequent RJ? If the answer is no, it is not capital.

8. Warrants / equity.

• 40% of pro forma fully diluted equity, delivered as penny warrants (strike US$0.01), 10-year, full anti-dilution, cashless exercise, registration rights and B3/NYSE listing.
• Step-up to 55% on any of: (i) PIK extended beyond 24 months; (ii) 2027 or 2028 EBITDA below US$1,300M; (iii) unrestricted liquidity below US$800M at any quarter-end; (iv) failure to complete the full US$1.5B sponsor equity raise by 31-Mar-2027; (v) Alagoas cash outflows exceeding US$200M in any year; (vi) any new-money default.
• Step-down to 25% if sponsors contribute US$2.0B+ of permanent equity, or if US$1.5B of principal is prepaid in cash by 31-Dec-2028.

9. Governance and control.

• Four creditor-designated directors on an 11-member board while net leverage exceeds 4.0x; two while between 3.0x and 4.0x.
• Chief Restructuring Officer with an independent reporting line to the board, appointed by and removable only with creditor consent, for a minimum of 24 months.
• Creditor veto over: any capex above the agreed maintenance budget; any strategic project including Transforma Rio; any related-party transaction of any size; any transaction with Braskem Idesa, Petrobras or IG4 affiliates; any asset sale above US$50M; any debt incurrence or lien; any dividend, buyback or capital return; any amendment to the Idesa arrangements; any settlement of Alagoas claims above US$100M; appointment or removal of the CEO and CFO.
• Monthly reporting (P&L, cash flow, 13-week liquidity, LC and factoring balances, supplier terms, Alagoas cash, Idesa flows) with quarterly advisor diligence sessions and unrestricted access [consistent with our 19 June demand, V june25ocr/014].

10. Cash-use guardrails.

• Minimum liquidity covenant: US$1.25B, tested monthly.
• 75% excess-cash sweep above US$1.25B, applied to the new-money tranche then pro rata to reinstated debt at par.
• 100% sweep of asset-sale proceeds above US$25M.
• 100% sweep of any working-capital release above a US$400M cumulative threshold - this is the deal-contingent US$700M–1.1B identified in §1.3, and it is ours.
• Zero dividends, buybacks, capital returns or management incentive payouts until net leverage < 3.0x for two consecutive quarters.
• No new debt, no liens, no acquisitions, no unrestricted subsidiaries, no investments outside the restricted group, no intercompany transfers outside the ordinary course - binding on non-debtor subsidiaries as well as debtors [our 19 June demand, V june25ocr/014].

11. Mexico / Braskem Idesa - hard ring-fence.

• Zero additional Braskem S.A. money, guarantees, keepwells, letters of comfort, or upstream liens for Braskem Idesa. Not one dollar.
• Preserve and assert the existing US$82M (R$426M) working-capital loan from Braskem to Braskem Idesa, which is secured by Braskem Idesa's assets and matures Dec-2026 [V, q2itr]. It is not to be forgiven, subordinated, converted or extended without our consent, and it must be filed and prosecuted in any Idesa Chapter 11.
• All commercial arrangements with Idesa (ethane, offtake, services, shared functions) to be documented at arm's length, disclosed monthly, and non-amendable without creditor consent.
• Standalone Idesa restructuring milestone: a filed Chapter 11 plan or a signed RSA by 30-Jun-2027, failing which the Braskem S.A. plan permits deconsolidation and abandonment of the Idesa equity at no cost to Braskem S.A.
• Braskem Idesa's US$900M 2029s and US$1.2B 2032s are not our problem and must not become our problem. No substantive consolidation, no cross-guarantees, no cross-default linkage.

12. Alagoas. Safety and remediation obligations sit outside the creditor compromise - they are uncompromisable and politically untouchable, and we do not seek to impair them. But: a ring-fenced, fully budgeted, monthly-reported Alagoas cash line with a hard annual cap of US$175M; any excess is a reporting event and a warrant step-up trigger; any civil settlement above US$100M requires creditor consent; the remaining R$3,247M (US$627M) provision is to be independently reviewed by an adviser we select at Company expense.

13. Minorities / BAK / BRKM5. The rights offering is open pro rata to all shareholders at a single price with statutory preemptive rights preserved. Non-subscribing shareholders are diluted; there is no anti-dilution protection, no top-up, and no adjustment. Sponsors backstop the minority tranche in full. Braskem must maintain its NYSE ADR listing and B3 listing; delisting is a default.

PATH C - WE FUND THE DIP AND FILE

If Paths A and B are refused: we accelerate, oppose Chapter 15 recognition, provide an US$800M priming DIP under Art. 69-A/69-B secured by all unencumbered assets, and present a creditor plan under Art. 56 §6 equitising 100% of the financial claims. Existing equity receives nothing.

5B. TRUE RESERVATION PACKAGE (our economic minimum - not to be disclosed)

We will sign, and only sign, a package worth ≥65 cents of PV at a 15% exit yield. The following clears ~72c at 15% / ~80c at 13% [I], and is what we actually expect to land near:

5C. Why debt-plus-warrants can beat a 70–90% equitisation - and when it does not

We hold this view for four reasons, and we state the condition under which it flips.

1. At US$2.5B+ of EBITDA, reinstated secured debt recovers par and warrants capture the upside. Immediate equitisation surrenders priority and current yield for a claim on the residual - a bad trade when the residual is thin and the priority is about to become secured.
2. Displacing Petrobras and IG4 destroys value we do not control. Petrobras is the naphtha and ethane counterparty and a 47% voting holder; IG4 is the operational turnaround sponsor. A creditor-owned Braskem loses the feedstock relationship, the REIQ/PRESIQ political relationship, and the antidumping protection that Braskem currently enjoys [V, q2itr: definitive PE antidumping duties, 20% import duty maintained to Oct-2026].
3. A creditor-controlled, NYSE- and B3-listed Brazilian petrochemical group is a governance, securities-law, tax and political liability our holders are not paid to carry.
4. We do not want to operate Braskem. Debt plus liens plus warrants gives us yield, priority and upside without that obligation.

The condition under which this flips: if the sponsors will not fund at least US$1.0B of permanent capital and will not grant collateral, then reinstatement is simply an unsecured five-year option written by us for free, and Path A (US$3.0B converted to 72% of equity) becomes strictly better. The sponsors choose which world we are in.

5D. Dilution mechanics - subscribing vs non-subscribing

Assumptions [I]: 797M shares outstanding (from 1H26 EPS of R$5.9882 on R$4,771M attributable income); market cap US$979M; pre-money for the rights offering set at US$750M (below the market cap, because on our central case the equity is worth zero and the market price is option value); sponsor/minority equity raise US$1.3B; creditor warrants at X% fully diluted.

Rights offering issues 100 × (1,300/750) = 173.3 new units against 100 existing → 273.3 units post-rights. A non-subscribing holder retains 36.6% of its prior stake before warrants.

Two things follow, and both are levers.

(a) IG4's control is on the line. Shine I holds 50.11% of the voting capital. If the rights offering is in ordinary (voting) shares and IG4 does not subscribe its full pro rata, IG4's voting stake falls below 50% and it loses control of Braskem. IG4 must either write a cheque or surrender the control premium it bought. We will insist the rights offering is in ordinary shares precisely for this reason.

(b) Percentage dilution is not value destruction. A non-subscribing Petrobras goes from 36.15% of a company whose equity is worth ~US$0–1.0B to 7.9% of a company whose equity, post US$1.3B injection and restructuring, is worth materially more. On a 5x/US$2.5B case with pro forma net debt of US$8.0B, reorganised equity ≈ US$4.5B: Petrobras non-subscribing at 7.9% = US$356M; subscribing at 21.7% = US$977M against a US$470M cheque, i.e. US$507M net - better than today's mark-to-market of ~US$354M (36.15% × US$979M). Subscribing is value-accretive for Petrobras even after 40% creditor warrants. We will put this table in front of them.

5E. Pro forma leverage under our reservation package

Pro forma gross financial debt at the end of the relief period (end-2028) [I]:

This is the strongest argument against our own reservation package and we make it ourselves: at mid-cycle EBITDA the capital structure remains at 7.0x and is not fixed. It is precisely the outcome the Company's own p.028 shows (gross debt rising to US$10,752M by 2029, 8.3–8.5x gross debt+WK/EBITDA in 2027–28). It is survivable only if (i) we are secured, so that a second restructuring in 2031 finds us at the top of the stack, (ii) we hold warrants, so that a recovery accrues to us, (iii) the sweeps are hard, so that any upside deleverages rather than leaking, and (iv) the strategic capex is dead. Take away any one of those four and the package does not clear 65 cents.

Under Path A (US$3.0B converted), pro forma net debt is ~US$7.4B: 4.9x at 1.5B, 3.7x at 2.0B, 3.0x at 2.5B, 2.5x at 3.0B. That is a solvent company. We would rather have Path A. We say so.

5F. Post-deal liquidity and annual cash-interest burden

6. EXPLICIT WALK-AWAY CONDITIONS

We will decline to sign, will accelerate, will oppose Chapter 15 recognition, will oppose EJ homologation, and will fund and file a creditor plan under Art. 56 §6, if any of the following is the final position:

1. Sponsor capital below US$1.0B, or in a form that is not loss-absorbing. Specifically: any sponsor instrument that (a) pays cash interest, (b) can be repaid, prepaid, redeemed or refinanced while our claims are impaired, (c) carries a lien or guarantee, (d) matures inside our restructured maturities, or (e) is not mandatorily convertible on a leverage trigger. Ordinary shareholder debt is not burden-sharing and we will treat an offer of it as a rejection.
2. Any coupon reduction, in any tranche, in any form. Non-negotiable. Our June position stands.
3. Zero-cash-pay. Any structure with less than 4.0% cash interest during the relief period, or a relief period longer than 24 months, or a PIK toggle at the borrower's discretion.
4. No collateral. A final package that leaves all reinstated claims unsecured. Liens are the cheapest concession available and refusing them signals that the sponsors intend to preserve optionality to prime us later.
5. Class-splitting. Any EJ or plan structure that separates BRL from USD financial claims, or carves LC/RCF banks into a favoured class, or counts the subordinated hybrid '81 in our class.
6. The LC pay-to-play survives - i.e. par roll-in conditioned on new commitments, with non-committing lenders demoted to impaired claims [V, june25ocr/010].
7. Warrants below 15% fully diluted, or below 25% if sponsor capital is debt-form.
8. Any Braskem S.A. money, guarantee or lien for Braskem Idesa, or any forgiveness/subordination of the US$82M secured intercompany loan.
9. Transforma Rio or any strategic capex proceeds while our claims are impaired and leverage exceeds 3.5x.
10. Governance: no CRO, no creditor board representation, no veto over related-party and Idesa transactions, or reporting weaker than monthly.
11. Sweeps: less than a 50% excess-cash sweep above US$1.0B, or no sweep of the deal-contingent working-capital release.
12. The package prices below 65 cents of PV at a 15% exit yield on the '30s-equivalent benchmark, inclusive of warrant value.
13. Diligence refused: if the Company continues to refuse the deep-dive advisor diligence on its projections that it offered in the 11 June in-person meeting and has not delivered [V, june25ocr/014]. We have been asked to take a 6.74x-levered credit on faith. We will not.

7. THE SINGLE STRONGEST ARGUMENT AGAINST US - AND OUR ANSWER

The argument

"Your own model says an RJ returns 35 cents. Your bonds trade at 58. A consensual reinstatement-plus-warrants deal on the mediator's terms is worth 65–73 cents. You are therefore threatening to destroy 30–38 points of your own value to chase an outcome worth half as much, and everyone at the table knows it. Meanwhile the facts have moved decisively in the Company's favour: 2Q26 EBITDA was US$1,043M, spreads have held into August, and on a persistent-spread FY26 of US$2.8B your own 5x framework gives US$14B of enterprise value against US$10.8B of claims. You are money-good. Take the reinstatement, drop the equity demand, and stop pretending you have a credible threat."

This is the strongest form of the case against us. It is partly right, and pretending otherwise would make us worse negotiators.

Our answer

(a) One quarter of a war premium is not a capital structure. The 2Q26 PE–naphtha spread of US$773/t is 2.0x the 2022–25 average of US$387/t [V, q2pres], and it exists because of a Middle East conflict that constrained feedstock and product flows [V, q2pres; q2itr]. Braskem's realised EBITDA was 5,024 / 1,796 / 612 / 845 / 508 across 2021–2025 [V, june25ocr/023]: a five-year mean of US$1,757M and a four-year mean ex-2021 of US$940M. Management's own plan - prepared by Lazard, with Cleary and E. Munhoz, for the express purpose of persuading us - says US$1,494M in 2027 and US$1,524M in 2028, and concedes structural overcapacity through 2027 [V]. Capitalising a war premium into permanent enterprise value is precisely the error that built this balance sheet. We will not repeat it, and no counterparty should expect us to.

(b) The windfall's Q2 cash conversion was depressed by a working-capital reset, not necessarily a recurring drain. US$1,043M of EBITDA became US$210M of recurring cash generation and negative US$149M of actual cash movement [V, q2release §4.8.1]. Higher naphtha and product prices, inventory and reduced payment arrangements absorbed cash. If naphtha and inventory stabilize, the operational price-and-volume component should not repeat. The creditor case instead rests on collapsing payables, tripled supplier prepayments, LC runoff and maturities [V, q2itr]. Even at management's own FY26E of US$2,166M, unrestricted cash still goes to negative US$821M by Dec-26 and negative US$1,981M by Dec-27 [V, june25ocr/025], principally because of financial working capital and debt service. A good operating year still does not solve a US$3,686M debt-service wall through Dec-27 [V, june25ocr/005] plus a US$1,326M January-2028 maturity against US$795M of cash.

(c) "Money-good at 4.5x" is not an argument for giving shareholders a free option; it is an argument for us taking the collateral. If the Company genuinely believes it is money-good, then granting us liens and a 4% cash coupon costs it nothing and it should do so today. Refusal to pledge unencumbered assets, in a jurisdiction where a court can authorise an extraconcursal secured DIP over those same assets tomorrow [L, Art. 69-A/69-B], is not a neutral act. It is the preservation of optionality to prime us. We read it as such.

(d) Our threat is credible because of asymmetry, not enthusiasm. We concede the point: RJ costs us ~23 points against the mark. But on our base case an RJ takes the equity to exactly zero - it destroys IG4's entire control investment and its 50.11% voting position, it writes off Petrobras's 36.15% economic stake and puts its integrated Camaçari/Rio feedstock relationship into a 30-month judicial process, and it does so alongside Alagoas, a Braskem Idesa Chapter 11, and Brazilian political exposure over employment at Camaçari and Triunfo. We lose 23 points; they lose 100%. And they have already told the market they know it: on 30 July B3 asked Braskem to confirm or deny that RJ "would already have the consent of its principal shareholders," and the Company declined to deny it [V, bak20260731_6k]. The party that has already conceded its BATNA in the press is not the party with leverage.

(e) And we do not need to win. We need 65 cents. We can get from 58 to 72+ almost entirely through terms that cost the sponsors no cash at all: liens (worth ~16 points of exit-yield compression on the '30s benchmark), hard sweeps, suspension of a US$852M growth project that impaired creditors would otherwise be funding, a 4% cash floor, and 20% of a warrant. We have designed a package the sponsors can afford. If they refuse a package they can afford, then the refusal is about control, not capacity - and control is exactly what an RJ takes away from them.

APPENDIX - KEY VERIFIED FIGURES INDEX