Round 1 - PETROBRAS position paper

Project Catalyst / Braskem S.A. restructuring Prepared by the Petrobras negotiating team · Information cutoff 2026-08-15 Status: independent opening. Prepared without sight of the management, IG4/Shine I or creditor position papers.

Post-run correction, 16 August 2026. Visual verification of the June plan corrected July operational working capital from US$87M to US$57M. The H2 release is therefore US$518M, not US$548M. Q2's US$547M outflow is also a mix of a price/inventory reset and persistent financing pressure, not a recurring quarterly burn. The refreshed mediator estimate uses 15% upfront warrants, stepping to 22.5% if cash-conversion, liquidity or PIK milestones fail.

Evidence convention. Every figure is labelled [V] = verified against a primary document in .context (file/page cited), or [I] = inference/model built by us on top of verified inputs. Where OCR of the Lazard deck is ambiguous we say so.

FX convention. Closing rate 30-Jun-2026 US$1 = R$5.17 [I, derived from four verified pairs in q2release.txt / q2itr.txt: R$929m=US$179m (5.190); R$703m=US$136m (5.169); R$507m=US$98m (5.173); R$196m=US$38m (5.158)]. Average 2Q26 ≈ R$5.04 [I, from R$4.4bn = US$869m Brazil segment EBITDA]. The Lazard deck uses USD/BRL ≈ 4.99 at 30-Apr-2026 [V, june25ocr/030].

1. Independent situation assessment

1.1 What we know is true

Ownership and control (exact, [V] q2itr.txt note 24.1 and note 1).

Shine I closed its purchase from Novonor on 3-June-2026 [V bak20260605_6k.htm]. From that date Braskem is under joint control of Shine I and Petrobras, and "decisions relating to Braskem's relevant activities became subject to consensus among the controlling shareholders" [V q2itr.txt note 1]. The ultimate joint controllers are named as the Federal Government of Brazil and Shine Equity LP [V, same]. The new Shareholders' Agreement was signed 23-April-2026 and became fully effective 3-June-2026 [V bak20260423_6k.htm, bak20260605_6k.htm].

Authorized capital [V q2itr.txt 24.1] - this is a materially under-appreciated fact.

Balance sheet and debt.

• Corporate gross debt 30-Jun-26 US$10.3bn; adjusted net debt US$9.5bn; corporate leverage 6.74x [V q2release.txt §4.9.1]. From 2Q26 gross debt now includes lease liabilities, applied retrospectively [V, footnote 2].
• Lazard debt schedule at 30-Apr-2026 totals US$9,497m of principal carrying US$665m of annual interest - a blended 7.00% [V june25ocr/030]. The gap to the US$10.3bn June figure is leases, accrued interest, FX and the US$179m of LC reimbursement obligations reclassified into loans and financings in Q2 plus US$136m more in July [V q2release.txt, q2itr.txt note 14].
• Shareholders' equity is negative R$13,087m consolidated / R$12,586m parent = negative US$2.53bn / US$2.43bn [V q2itr.txt note 1; converted at 5.17].
• Fitch C, S&P D, both 26-June-2026 [V q2release.txt §4.9.3].
• BRKM5 R$6.36 · BAK US$2.44/ADR (1 ADR = 2 PNA) [V q2release.txt]. Implied market capitalisation R$5,070m = US$980m [I]. Petrobras's 288,188,691 shares are worth R$1,833m = US$355m [I].

Full debt stack [V june25ocr/030, 30-Apr-2026, US$m]. Every guarantor line reads "Braskem S.A." or a Braskem entity. There is no Petrobras guarantee anywhere in the capital structure.

Line-item OCR sums to ~US$9.70bn against the deck's stated US$9,497m total; the ~US$207m residual equals the 2081 hybrid, suggesting the total may exclude it. We use US$9,497m / US$665m as the verified anchors and flag the ±US$207m. [I]

• LC facilities outstanding US$1,308m at 30-Apr-26 across Committed Facility IV (954 borrower Braskem Trading & Shipping BV, guaranteed in full by Braskem S.A.), Committed Facility V, bilateral committed and spot LCs [V june25ocr/031-032].
• 3Q26 scheduled debt service US$878m - bonds & debentures 188, bilateral & ECA 118, LCs 572 [V june25ocr/005].
• Management's June plan showed cash of US$918m / unrestricted US$795m at 30-Jun-26 [V june25ocr/024]. Braskem therefore cannot pay 3Q26 debt service out of cash. That is the entire reason the injunction exists.

Operating.

• 2Q26 recurring EBITDA US$1,043m (margin 24.2%); 1Q26 US$192m; 2Q25 US$74m [V q2pres.txt]. Segments: Brazil/S. America 869, USA & Europe 147, Mexico 57 [V q2release.txt].
• Brazilian complex utilisation 70% (2Q25 74%, 1Q26 69%) [V q2pres.txt].
• 2Q26 cash bridge: EBITDA 1,043 → working capital −547 → operating cash 385 → capex, leases, Alagoas −48, interest, tax → recurring cash generation ~210 → decrease in cash position −15 [V q2pres.txt slide 12].
• Trade credit is collapsing. Advances to suppliers R$1,625m vs R$544m a year earlier (+R$1,081m ≈ US$209m of cash tied up) [V]; reverse-factoring availability reduced [V note 14]; US$179m of LCs cash-settled by issuing banks directly to suppliers in Q2 and US$136m in July, both reclassified into financial debt [V]; suppliers' negative balance US$1,606m at 2Q26 close [V q2pres.txt].

Legal status.

• 60-day stay of enforcement and attachments granted by the 2ª Vara de Falências e RJ de São Paulo on 25/26-June-2026 in favour of creditors invited to the Wind Mediation Chamber [V q2release.txt §4.9.2]. Chapter 15 filed 26-June; US preliminary automatic stay granted 30-June "for the same period" [V].
• As at 30-Jun-26 management asserts no event of default or acceleration right had been triggered [V]. In July, after cure periods expired on R$507m (US$98m) of unpaid obligations, the Company was in default under certain financial instruments [V q2itr.txt].
• From July the Company no longer has the unconditional right to defer settlement beyond 12 months; defaulted balances reclassify to current liabilities [V both].
• Auditors: material uncertainty / substantial doubt on going concern for Braskem S.A. and separately for Braskem Idesa [V q2itr.txt review report and note 1].
• Creditor groups have delivered indicative, non-binding proposals that include "a possible capitalization and the granting of security interests over assets" [V q2release.txt, q2itr.txt, bak20260731_6k.htm].
• Valor Econômico, 29-July-2026: Braskem moving closer to judicial reorganization (RJ) with unofficial controller acquiescence. Braskem's 31-July response to B3 neither confirmed nor denied, saying only that "no decision has been reached regarding the terms of a potential Restructuring or any additional measures, whether judicial or otherwise" [V bak20260731_6k.htm]. That is a non-denial and we should read it as one.

Mexico. Braskem Idesa borrowings on the consolidated balance sheet R$14,156m = US$2,738m; Braskem Idesa equity negative R$2,467m; total liabilities R$23,399m [V q2itr.txt lines 648, 829]. Missed 2029 notes coupon 18-Nov-2025 and 2032 notes coupon 20-Feb-2026 [V]. Negotiating reorganisation "through judicial measures (e.g., Chapter 11)… which may have potential impacts on the Company as well on the shareholding control of Braskem Idesa" [V]. Braskem's own exposure: term loan US$180m committed / US$129m disbursed (US$34m due Dec-26, US$95m due 2029) plus US$82m of WC loans due Dec-26 secured by Braskem Idesa's assets [V].

Alagoas. Total provision recognised R$18.2bn; disbursed R$14.6bn; R$1.2bn reclassified to other payables; realised AVP R$873m; net provision balance ≈R$3.2bn (US$619m) at 30-Jun-26, 32% current / 68% non-current [V q2pres.txt slide 13]. 99.9% of properties relocated; 19,140 of 19,205 proposals accepted [V]. State of Alagoas agreement (10-Nov-2025): R$1.2bn, R$139m paid, balance over 10 variable annual instalments "adjusted mainly after 2030, taking into account the Company's ability to pay" [V]. Live TCU and CVM proceedings relating to the geological event [V q2release.txt]. 2026 Alagoas cash outflow in management's plan: US$260m [V, sum of june25ocr/024 monthly row].

1.2 The August ~25 cliff, and what happens after it

The stay runs 60 days from 26-June-2026 → expiry on or about 25-August-2026 [I, date arithmetic on the verified 60-day grant]. Today is 15-August. There are roughly nine or ten days left. The Chapter 15 preliminary stay is expressly coterminous [V].

On expiry, absent action:

1. The July payment defaults become enforceable. Cross-default and cross-acceleration cascade across effectively the entire US$9.5bn stack - the bonds are cross-guaranteed by Braskem S.A. and the ECA/bilateral documents will contain standard cross-default triggers [I on the documentation, [V] on the fact of default].
2. Attachment risk returns. Braskem's operating cash and receivables become seizable in Brazil; the Chapter 15 protection over US assets lapses pending final recognition.
3. LC issuing banks continue to cash-settle at maturity and convert trade obligations into accelerated financial debt - US$315m has already happened in Q2 and July alone [V]. There is US$572m of LC service in 3Q26 [V].
4. Suppliers tighten further. Advances to suppliers have already tripled [V]. At 70% utilisation, further tightening cuts runs.

The realistic branches are: (a) extend the tutela cautelar; (b) file the Recuperação Extrajudicial (EJ) - management's plan required only 1/3 creditor support to file, providing a standstill, and a majority to confirm [V june25ocr/011]; or (c) file RJ.

This distinction is the single most important legal fact in the negotiation for Petrobras. In an EJ, only the adhering financial-creditor classes are bound, equitisation is impossible without shareholder approval, and Petrobras's 47.03% voting block plus the consensus mechanism in the Shareholders' Agreement is decisive. In an RJ, if the debtor's plan is rejected, creditors may present an alternative plan which, under Law 11,101 as amended by Law 14,112/2020, may provide for capital increase and equitisation without the approval of shareholders [I - this is our reading of Brazilian law, not a document in the record; it must be confirmed by counsel before Round 2]. Related-party creditors - which is what Petrobras is, as a trade creditor - do not vote in the creditors' meeting [I, same caveat].

Therefore: the creditors' entire path to forced equitisation runs through RJ, and Petrobras's leverage falls off a cliff the day an RJ is filed. Every hour we spend outside RJ is an hour in which we hold a veto. We should price that honestly rather than pretend our veto is durable.

1.3 The spread question - and why we must not accept the June forecast as fact

The Q2 presentation was filed 14-August-2026 but states on its own disclaimer page that it "was updated as of June 30, 2026" [V bak20260814_6k.htm]. Its forward spreads are a June-vintage third-party consultant forecast, not observed August data.

PE–Naphtha, Brazil, US$/t [V q2pres.txt]:

PP–Naphtha Brazil: 293 → 457 → 329e → 386e → 267e. PE–Ethane Mexico: 824 → 1,425 → 844e → 805e [V].

Note the internal inconsistency the creditors will not point out but we should: the forecast 3Q26e of 316 is below the 2022-25 average of 387 and 43% below the ten-year average of 555, i.e. management's own consultants are forecasting a below-trough quarter immediately after a US$773 quarter. That is a forecast, and a conservative one. Public Platts reporting for the week ended 22-July showed Americas PE prices stable to rising with broadly stable CFR Brazil grades, and our principal observes spreads held into mid-August [stated as fact by the principal; not independently verifiable in this record].

Calibration [I]. Management's own June monthly plan implies 3Q26 EBITDA of 263+166+157 = US$586m at a PE spread of 316, against US$1,043m actual at 773. Attributing across a ~938kt/quarter Brazilian resin book (763kt domestic + 175kt export, [V q2pres.txt]) of which PE ≈ 55%: each US$100/t of persistent PE–naphtha spread ≈ US$52m of Brazil EBITDA per quarter (~US$200m/yr). Reconstructing 3Q26 top-down from 2Q26 gives ~US$624m against management's US$586m - our sensitivity is calibrated to within 7%.

Case A - persistent spreads.

We adopt A2 = US$2.9bn as the working persistent case. Note A1 = US$3.3bn matches the top of the mandated range and requires PE spreads at 773 every month to December - possible, not central.

Case B - normalization. Management's plan, verified line by line at june25ocr/023: 2026E 2,166 · 2027E 1,494 · 2028E 1,524 · 2029E 1,710 · 2030E 1,866 · 2031E 2,304 · 2032E 2,334 · 2033E 2,621 · 2034E 2,830 · 2035E 3,030. Historic for context: 2021A 5,024 · 2022A 1,796 · 2023A 612 · 2024A 845 · 2025A 508.

Two observations the creditors will use, and one they will not. They will use 2027E 1,494 and 2028E 1,524 to prove insolvency. They will not volunteer that the same plan has EBITDA back above US$2.3bn by 2031 and above US$3.0bn by 2035 [V]. The plan that proves insolvency in 2027 proves solvency in 2031. That is precisely a five-year option, and precisely what a five-year maturity extension buys.

1.4 Does H2-26 deliver cash? - the required test

Management's own status-quo monthly plan, Jul–Dec 2026 [V june25ocr/024, our summation]:

Two facts that reframe the entire negotiation:

1. Management's own plan shows H2-26 recurring operating cash flow of +US$978m and working capital reversing +US$518m POSITIVE, versus the −US$547m consumed in Q2 [V]. The business generates cash in H2 even on the pessimistic spread deck.
2. The negative US$821m year-end cash position is caused entirely by refinancing items - US$860m of LC runoff, US$325m of cash interest, US$1,242m of amortizations of which US$1,004m is the RCF on 31-Dec-26. It is not caused by operating losses.

Persistent case A2 (+US$703m of H2 EBITDA vs plan) [I]:

Answer to the test: YES - in the persistent-spread case H2-26 delivers cash after working capital, roughly US$155m to US$673m of CFADS against planned CFADS of about negative US$30m. But it comes nowhere near the US$1,242m of scheduled amortizations, of which US$1,004m is the RCF. Persistent spreads solve the operating problem. They do not solve the maturity problem. Extension is required regardless of spreads.

That sentence is our opening statement at the table.

1.5 The conclusion management's own deck reaches, and everyone is ignoring

Management's restructuring-proposal cash flow - 5-year extension, 100% PIK to Dec-2028, −200bp, US$1.5bn LC facility, and zero new money - produces unrestricted cash of US$1,602m (YE26) · 1,695 (YE27) · 1,900 (YE28) · 2,105 (YE29) · 2,221 (YE30) [V june25ocr/028].

New money is not a liquidity requirement. It is a consent price. Maturity extension plus the LC roll alone fixes the liquidity. Whatever Petrobras pays, it is paying for creditor signatures, not to fill a hole. That is the frame we must hold all negotiation, and it is verifiable from the creditors' own cleansing materials.

2. Incentives, legal and governance constraints

2.1 What Petrobras is trying to achieve

1. Preserve joint control and Brazilian downstream/petrochemical integration.
2. Preserve the offtake for domestic naphtha. Braskem purchased R$5,098m (≈US$1.01bn) from Petrobras and subsidiaries in 6M26 [V q2itr.txt note 8], i.e. ~US$2.0bn/year. Sales to Petrobras were R$212m. Net balances are small (R$281m asset / R$257m liability, parent) [V] - large flow, small stock exposure.
3. Avoid consolidation, avoid guarantees, avoid Alagoas contamination, avoid Mexico contamination.
4. Preserve optionality on the 2029-2035 recovery in management's own plan.

2.2 Binding constraints

(a) Consolidation. Petrobras accounts for Braskem under the equity method because control is joint and by consensus [V q2itr.txt note 1]. If Petrobras exceeded 50% of voting capital, or if the consensus mechanism were removed, IFRS 10 would force consolidation of US$10.3bn of Braskem corporate gross debt plus US$2.74bn of Braskem Idesa debt onto Petrobras's balance sheet [V both figures]. This is a hard, non-negotiable cap: Petrobras will not go above 47.03% of voting capital.

(b) Equity-method carrying value. With Braskem's consolidated equity at negative US$2.53bn [V], under IAS 28 Petrobras will have ceased recognising further losses once its carrying amount reached zero, absent legal or constructive obligations [I - Petrobras's own accounts are not in this record]. Implication: Petrobras's accounting downside from a total wipeout is already close to zero. Any new money is genuinely new cash at risk, and it will be judged as such. We should not pretend otherwise internally.

(c) State-controlled company governance. Petrobras is subject to Lei 13.303/2016 (Lei das Estatais), which requires related-party transactions to be on arm's-length terms, documented, approved through the statutory committees and disclosed [I on the citation; the fact of TCU and CVM scrutiny of Braskem matters is [V] q2release.txt]. Any support to Braskem is a related-party transaction between two listed companies under joint control, requiring:

• an independent fairness opinion / valuation on the terms;
• approval by Petrobras's Board with abstention of conflicted members;
• disclosure under CVM Res. 44 and the related-party policy;
• resilience to TCU review - and the TCU is already engaged on Alagoas-adjacent matters [V].

The practical consequence is decisive: the more a Petrobras contribution looks like a commercial term on an existing arm's-length contract, and the less it looks like a rescue capital injection, the faster and more durably it can be delivered. This is not squeamishness. It is a delivery constraint on the negotiation calendar, and we have nine days.

(d) Lei 6.404/76 Art. 170 §1. The issue price in a capital increase must be fixed by reference to net worth, market quotation, or earnings prospects, without unjustified dilution of existing shareholders. Braskem's net worth is negative. Its earnings prospects are the disputed variable. The only defensible, observable reference is market quotation - US$980m of market capitalisation. A Petrobras representative cannot vote for a capitalisation at a pre-money materially below that without exposing himself personally. This is a legal constraint, not a preference, and it is the most useful sentence we own.

(e) Lei 6.404/76 Art. 171 - preemptive rights. Petrobras and every minority holder have statutory preemptive rights in any capital increase. Petrobras will not waive them.

(f) Authorized capital. Any issuance beyond the headroom in §1.1 requires a bylaw amendment, which requires an EGM and, under the April-2026 Shareholders' Agreement, prior consensus of the controlling shareholders [I on the SHA's specific matter list; [V] on the general consensus requirement]. Practically, Petrobras holds a veto over the size of any equitisation outside RJ.

(g) The April-2026 Shine I / Petrobras Shareholders' Agreement. Signed 23-Apr-2026, effective 3-Jun-2026, "regulating, among other things, the exercise of shared control" [V bak20260423_6k.htm]. Petrobras simultaneously declined to exercise pre-emptive and tag-along rights under the prior agreement [V, same] - which tells us such rights existed and very likely exist in the new one, constraining any Petrobras exit. Preserving this agreement intact is the second hard red line after the 50% voting cap.

(h) Preferred-share voting contingency. Under Lei 6.404/76 Art. 111 §1, preferred shares acquire voting rights if the fixed or minimum dividend is not paid for three consecutive years. Braskem's PNA carry a 6% priority non-cumulative dividend [V q2itr.txt 24.2] and Braskem has not been paying. If PNA acquire votes, all 345m existing PNA plus any new ones vote, which would swamp both controllers. This risk exists today independent of this transaction, but issuing warrants over PNA increases the stakes. Counsel must confirm the status of the three-year clock before Round 2. [I]

2.3 What Petrobras must not do

No guarantee. No keep-well. No comfort letter. No cross-collateralisation. No Alagoas assumption. No Mexico assumption. No funding ahead of creditor delivery. No majority voting stake. Each of these is grounded in (a)–(h) above, not in preference.

3. Negotiating leverage - an honest inventory

What we actually control

1. The consensus veto under the April-2026 Shareholders' Agreement over relevant activities, including - we assume - capital increases, bylaw amendments, liens over core assets, and the decision to file EJ or RJ [V on the consensus requirement; [I] on the matter list]. This is real and it is our principal asset - but only outside RJ.
2. 47.03% of voting capital, sufficient to block any bylaw amendment (which requires a qualified quorum) and to block the authorized-capital increase needed to deliver a large equitisation outside RJ.
3. The naphtha relationship. ~US$2.0bn/year of purchases [V]. Petrobras is the counterparty Braskem cannot replace at scale or on short notice, and - critically - the one supplier that can expand trade credit while every other supplier is contracting it [V: advances to suppliers tripled to R$1,625m].
4. Credibility and balance sheet. We are the only party at the table whose signature makes a facility bankable without security.
5. Time, for nine more days.

What we do not control - and must stop pretending we do

1. We are not a financial creditor. Petrobras holds no Braskem bond, debenture, RCF or bilateral participation. We have no seat, no vote and no claim in the creditor negotiation. Our R$257m trade payable balance is a related-party claim which, in an RJ, would very likely be disenfranchised from voting [I].
2. We cannot stop an RJ filing in practice. Management and the Board can file; Valor reports controller acquiescence is already assumed [V the report, [V] Braskem's non-denial]. And once in RJ, a creditor alternative plan can equitise without our vote [I]. Our veto is a pre-RJ asset with a decaying value, and the decay accelerates on 25-August.
3. We cannot force IG4/Shine I to fund. We can only decline to move first.
4. We cannot deliver the LC banks. The US$1.3bn roll is the crux of the plan and it depends on trade-finance banks who are not in the AHG and who have been quietly exiting by cash-settling at maturity [V US$315m already].
5. We cannot credibly threaten to walk. A sale of 36.15% of a distressed, negative-equity, defaulted issuer, subject to ROFR/tag under the SHA and to a TCU-compliant competitive divestment process, is not executable inside this negotiation window. Bluffing exit would be seen through immediately and would cost us more than it gains. We will not do it.
6. Our accounting downside is already taken. Sophisticated creditors will work out that our carrying value is ~zero and will argue our marginal loss from a wipeout is only the US$355m of market value. They will be roughly right.

Net assessment: we hold a strong hand this week and a weak hand next month. Our strategy must be to convert the veto into a signed deal before it expires, not to husband it.

4. BATNA, quantified

Baseline. Petrobras stake at market: 288,188,691 × R$6.36 = R$1,833m = US$355m [V/I]. Equity-method carrying value ≈ zero [I].

Braskem enterprise/equity value grid [I] (EV/EBITDA at 5.5x, a mid-cycle commodity chemicals multiple; corporate net debt US$9.5bn [V]):

At management's own 2027-28 EBITDA the equity is out of the money, which is exactly consistent with negative book equity of US$2.53bn and a US$980m market capitalisation against US$9.5bn of net debt. We must state this internally without flinching.

(i) Braskem files RJ

Deduct RJ-specific costs to Petrobras outside the stake:

• Trade claim haircut. ~US$50-58m of related-party receivable [V R$257-301m] crammed into the plan at 30-50% NPV → −US$25 to −40m.
• Commercial disruption. Post-petition sales move to cash-in-advance or 7-day terms, tying up Petrobras working capital; run cuts from 70% utilisation reduce naphtha offtake. Petrobras's incremental margin on domestic naphtha vs the export alternative: ~US$150m/yr [I]. 30% probability of a 40% multi-year volume loss, NPV at 10% over 5 years → −US$68m.
• Reputational/political/TCU cost: material, unquantified.

RJ BATNA ≈ US$277m − US$32m − US$68m ≈ US$175-190m, plus unquantified strategic cost.

Note the discipline this imposes: our RJ BATNA (~US$185m) is barely half today's market value of the stake (US$355m). Filing costs us roughly US$170m of stake value plus the commercial damage. That is what we are defending - and it is not a large number relative to the sums being discussed. We must not overpay to avoid it.

(ii) Creditors equitise; Petrobras diluted below joint control

Model: US$4.0bn of claims converted at a pre-money equity value P.

Remaining debt after a US$4bn conversion is US$6.3bn; at 1.5bn EBITDA that is 4.2x - a genuinely viable capital structure, which is why the equity has value again.

The honest and uncomfortable conclusion: a large equitisation at a fair pre-money can leave Petrobras better off in value (US$409m) than a no-equitisation PIK deal that leaves US$10.3bn of debt against US$1.5bn of EBITDA (equity ≈ zero). The fight is not over the principle of equitisation. It is over the pre-money and over control. Add a strategic control premium of US$100-200m [I] for retaining the SHA veto, board seats and the ability to protect the naphtha relationship.

(iii) Petrobras walks away entirely

• No buyer at scale for 36.15% of a defaulted, negative-equity issuer with a US$980m market capitalisation. The block would clear at a 30-50% discount to screen: US$180-250m.
• Subject to ROFR/tag under the SHA [I, strongly implied by [V] the April-2026 waiver] and to a TCU-compliant competitive process - not executable within this negotiation window, or within twelve months.
• Forfeits the naphtha relationship's strategic anchor and Petrobras's declared downstream strategy.

Walk-away ≈ US$180-250m nominal, ~US$0 executable now. This is not a credible BATNA and we will not present it as one.

4.1 What the BATNA implies for our funding ceiling

Value of a consensual deal to Petrobras [I], valuing the equity as an option on survival to management's own mid-cycle:

Gain over RJ BATNA: 709 − 277 = US$432m. Plus protection of the commercial/strategic relationship (§4(i)): US$100-300m, midpoint US$200m.

Rational maximum Petrobras contribution ≈ US$600-650m of value, of which only US$150-250m should be genuinely loss-absorbing. Absolute ceiling US$750m, and only against verified IG4 and creditor burden-sharing. Every dollar above that destroys Petrobras shareholder value relative to letting Braskem file.

4.2 The commercial/feedstock relationship, valued separately

This is why Petrobras is at the table at all, and it is also why our best instrument is trade credit rather than cash.

5. Round 1 opening package - "Petrobras Structural Bridge"

Headline: US$1.5bn LC facility + US$400m of Petrobras facilities + US$150m rights offering + no principal haircut + 7.5% creditor warrants stepping to 22.5%, in exchange for a full maturity term-out and 50/50 cash-PIK.

We open by withdrawing management's two indefensible asks: the 200bp coupon reduction and the flat 100% PIK to Dec-2028. The AHG called the coupon cut "unheard of in a corporate debt restructuring" and said it "calls into question the seriousness" of the approach [V june25ocr/013]. They are right. We buy credibility with that concession on the first minute.

5.1 Maturity extension by bucket

Weighted-average extension across the impaired stack ≈ 3.9 years, versus management's undifferentiated 5.0 [I]. We deliberately do not extend the 2041s and 2050s: they create no liquidity benefit and extending them is exactly what the AHG condemned as "maturity extensions that are unnecessary to address the Company's liquidity and capital structure needs" [V june25ocr/013]. Conceding it costs us nothing and buys enormous credibility.

Result: no corporate maturity before January 2031, other than ECA amortisation. The Dec-26 RCF cliff and the Jan-28 bond cliff [V, US$1,326m of Jan-28 service, june25ocr/005] both disappear.

5.2 Cash vs PIK, and for how long

• 50% cash / 50% PIK for 24 months (Oct-2026 → Sep-2028), 100% cash thereafter.
• PIK toggle up to 100% at the Company's option in any quarter in which unrestricted liquidity is below US$1.2bn, at a +100bp toggle premium on the PIK'd portion. Use of the toggle for more than four quarters fires the warrant step-up (§5.8).
• LC facility and ECA/MLA: always cash-pay. These are the trade lifelines; PIK-ing them destroys the facility and the ECA relationships needed for Transforma Rio.
• Bond 2081 hybrid: 100% PIK for 60 months, no toggle premium, no consent fee. It is deeply subordinated and must absorb loss ahead of senior unsecured. We will not allow the hybrid to free-ride on senior concessions.

Sanity test against management's own numbers: 2027 recurring operating cash flow US$556m; 2028 US$689m [V june25ocr/025], less strategic capex ~US$225m/yr → CFADS US$331m / US$464m. Our 50% cash construct costs ~US$380m/yr (§5.14). It does not clear in 2027 without the toggle. We will say so out loud rather than be caught. That is precisely why the toggle and the warrant step-up exist - they are the risk-sharing device, and they fire in exactly the state of the world in which creditors are proved right.

5.3 Coupon and step-ups

• No coupon reduction. Withdrawn.
• Existing coupon +100bp from closing.
• Step to +200bp from month 25 (when full cash-pay resumes).
• Step to +300bp from month 49 if net leverage exceeds 4.0x.
• PIK'd interest accrues at cash coupon +100bp.
• LC facility: Term SOFR + 300bp all-in including issuance and fronting fees (management offered SOFR+200 [V june25ocr/010]; that was not clearing).

Blended contractual rate moves 7.00% → 8.00% → 9.00% → 10.00% on the US$9,497m base [V base and rate].

Internal movement authority: to +150/+250/+350bp in Round 2.

5.4 Fees

Total cash fee outlay at closing ≈ US$60m, inside management's own US$100m transaction-fee envelope [V june25ocr/028 note]. The advisor-fee concession directly answers an explicit AHG demand [V june25ocr/014] and costs us almost nothing.

5.5 Collateral and lien ranking

The creditors' July non-binding proposals expressly seek "security interests over assets" [V ×3 sources]. Management's June position - "no encumbrance of fixed assets, all debt retains unsecured status" [V june25ocr/009] - has already failed. We propose a waterfall, not a blanket lien:

Tier 0 - new money, super-senior. First-priority lien on financed inventory, related receivables and proceeds only, securing the new US$1.5bn LC facility and the Petrobras facilities. Borrowing-base governed: 85% advance on eligible receivables, 65% on eligible finished-goods inventory, monthly certificates, dominion on default.

Tier 1 - legacy financial unsecured. No fixed-asset lien at closing. Instead:

• a negative pledge with a hard-capped permitted-lien basket of US$500m plus the Tier 0 borrowing base;
• share pledges over non-core subsidiaries: Braskem America Inc., Braskem Netherlands/Europe B.V. equity, Braskem Green, Wise Plásticos, the TQPM interest;
• expressly excluded: Braskem Idesa equity (already encumbered and in default) and the Brazilian crackers.

Contingent second lien - automatic, no further shareholder action. If at 30-June-2028 net leverage exceeds 4.5x, OR unrestricted liquidity falls below US$800m for two consecutive quarter-ends, holders of extended claims automatically receive a second-priority lien on the Brazilian crackers (Camaçari, Triunfo, ABC/Capuava, Duque de Caxias) and unaffected real property, documented and pre-signed at closing, held in escrow by a security agent.

Why a first lien at closing is refused, on grounds not of preference but of law and economics:

1. It strips the only collateral capable of supporting the LC and trade facilities that are Braskem's sole incremental funding source - US$1,308m of LCs are 100% of the marginal liquidity [V], and trade credit is already collapsing [V].
2. Granting security for antecedent unsecured debt while payment defaults are live [V: default occurred in July] is exposed to avoidance as an ineffective act in a subsequent RJ [I on the specific provision; the principle is not seriously disputed]. A lien that can be unwound is worth less to creditors than a contingent lien that cannot.
3. It is uncompensated. Creditors are not offering new money for it. New money gets Tier 0. Consent gets contingent Tier 2.

5.6 New money - amount, instrument, and who pays

Total package capacity: US$2.05bn, of which only US$350m is genuinely new external cash.

Petrobras - US$400m of facilities plus US$54m of equity:

(a) US$250m Supplier Credit Facility ("naphtha term extension"). Petrobras extends payment terms on its own feedstock sales to Braskem to 120 days, capped at US$250m outstanding, priced at SOFR + 400bp, documented as a commercial amendment to existing supply contracts. Pari passu with Tier 0 on the inventory/receivables borrowing base; ordinary-course by nature.

• Sizing: US$250m = 46 days of Braskem's US$1.97bn/yr of purchases from Petrobras [V]. Fully self-liquidating; unwinds automatically if volumes fall.
• Why this instrument: it is the highest-value intervention available to any party. Braskem's advances to suppliers have tripled to R$1,625m and reverse factoring is contracting [V]. Petrobras is the one supplier large enough to move the aggregate the other way, and restoring the largest supplier's terms anchors the rest. US$250m of Petrobras supplier credit is worth more to Braskem's liquidity than US$250m of cash.
• Why it clears governance: it is a commercial term on an existing arm's-length contract at a market rate - deliverable in days, not months, and defensible to the TCU and CVM with a standard fairness opinion.

(b) US$150m Borrowing-Base WC Revolver. 3-year, SOFR + 450bp, first-priority on financed inventory/receivables/proceeds only. Drawable only after (i) an RSA signed by ≥66⅔% of financial unsecured claims by value; (ii) the extensions are legally effective (EJ homologated or equivalent); (iii) the US$1.5bn LC facility is committed and closed; (iv) IG4/Shine I has funded in cash. Not one dollar before creditors deliver a durable extension - otherwise Petrobras money simply pays legacy creditors.

(c) US$150m rights offering, Petrobras subscribes its pro-rata 36.15% = US$54m. Petrobras does not backstop the minority portion in Round 1.

Petrobras Round 1: US$454m of exposure - US$204m of cash, of which only US$54m is loss-absorbing, plus US$250m of self-liquidating trade credit.

IG4 / Shine I - demanded US$275m, of which ≥US$150m loss-absorbing, plus a US$100m backstop of the minority portion of the rights offering. Rationale, stated plainly at the table: Shine I closed its purchase of control on 3-June-2026 - twenty-two days before the cleansing materials were published and the injunction was sought [V both dates]. It is the newest money in the structure and the only shareholder that voluntarily assumed this risk with full and current information. Petrobras will not fund ahead of, or in larger loss-absorbing size than, the party that just bought control.

Creditors - the US$200m incremental LC tranche backstopped by AHG members and banks, plus the US$1.3bn roll [V structure from june25ocr/010]. Additionally, an option (not an obligation) to fund US$275m of new money as a first-out super-senior tranche ranking with Tier 0, in exchange for an incremental 2.5% of fully-diluted warrants.

Eligibility linkage retained from management's proposal [V june25ocr/010]: a lender may roll its existing LC claims into the new facility only if it commits its pro-rata share of the US$200m incremental; non-committing lenders' claims are treated as financial unsecured claims. This is the correct incentive and we keep it.

5.7 Permanent capital vs repayable shareholder debt

Final-package capacity for permanent capital: US$150-250m, contingent on IG4 matching at ≥US$150m and on creditors taking contingent warrants rather than immediate conversion.

Petrobras will not provide a deeply subordinated shareholder loan. At negative book equity of US$2.43bn [V], subordinated related-party debt is (a) recharacterisation bait in any later RJ and (b) attackable by CVM as a disguised capital contribution made without independent valuation. If subordinated support becomes necessary, it must take the form of non-voting redeemable preferred shares with a fixed 6% priority dividend - an instrument Brazilian corporate law already recognises [V the 6% priority dividend construct, q2itr.txt 24.2] and which carries no votes.

5.8 Warrants and triggers

Round 1: 7.5% fully-diluted penny warrants to financial unsecured creditors, issued as warrants over Class A preferred shares (PNA) - non-voting. Strike R$0.01, 10-year, cashless exercise, standard anti-dilution, no board rights, no voting, no information rights beyond those in §5.10.

Contingent step-up to 15% if any of:

1. 2027 recurring EBITDA < US$1,300m;
2. unrestricted liquidity < US$800m at any two consecutive quarter-ends;
3. the PIK toggle is used at 100% for more than four quarters;
4. net leverage > 5.5x at 31-Dec-2028.

Second step-up to 22.5% if net leverage > 5.5x at 31-Dec-2029 and the Company has not refinanced ≥US$2.0bn in the capital markets.

Structural undertaking - and this is a real concession we should be loud about. A 22.5% PNA warrant requires 276.6m new PNA against a post-rights headroom of only 116.2m [I, computed from [V] authorized capital]. It cannot be delivered without a bylaw amendment, which requires our vote. So: at closing, Petrobras will vote to increase authorized capital to 1,600,000,000 shares (of which up to 1,000,000,000 PNA), so that the full contingent step-up is deliverable without any further shareholder action. This directly answers the AHG's core complaint that shareholder promises are unenforceable, and it costs Petrobras nothing in the good state of the world.

Warrant exercise is blocked until the Tier 0 facilities are repaid, or with Petrobras consent - exercise cannot be used to manufacture a governance event.

Internal movement authority: to 15% immediate / 22.5% contingent in Round 2; in extremis 20% immediate / 25% contingent - but only if there is no immediate debt-to-equity conversion, no creditor board control, and IG4 has funded ≥US$275m.

A strategic observation for our own side. We NPV'd the package for a 2028 bondholder at 100 par: 50bp cash consent + 100bp PIK fee, coupon 4.5%→5.5%→6.5%, 50% cash for two years, bullet 2033 at ~105.6 with PIK accretion. PV ≈ 72.5 cents at a 14% discount rate, ≈ 85.6 cents at 10%. The 7.5% warrant strip is worth US$75-128m across US$9.5bn of claims = 0.8-1.3 cents; even at 22.5% it is only 2.4-4.0 cents. Warrants are cheap for us and close to worthless to them. The real currency is coupon, cash-pay percentage, collateral, extension length and sponsor cash. We should trade warrants generously and defend coupon, cash-pay and collateral to the last. Expect creditors to see this too and to push hard on collateral and principal reduction rather than equity - our contingent second lien is the designed pressure valve.

5.9 Immediate and contingent dilution - full arithmetic

Rights offering: US$150m = R$777m at R$5.00/share (a 21.4% discount to the R$6.36 screen) = 155.4m new PNA. New total 952.6m shares.

The critical structural point: because the warrants are over non-voting Class A preferred, Petrobras's voting stake stays at exactly 47.03% at every level of economic dilution up to 22.5%. We can concede up to roughly a quarter of the economics without touching control, without exceeding 50% voting, and without triggering consolidation. This is our largest and cheapest concession, and we should deploy it deliberately.

Percentage dilution ≠ value destruction, and we will make the distinction explicitly (as the protocol requires):

• Theoretical ex-rights price = (797.21 × 6.36 + 155.4 × 5.00) / 952.6 = R$6.138.
• A non-subscribing minority loses R$6.36 − R$6.138 = R$0.222/share = 3.5% of value - not the 15.7% headline percentage dilution.
• A minority who sells their rights (R$1.138 per new share × 0.195 rights/share = R$0.222) loses nothing.
• Cumulative percentage dilution for a fully non-participating minority through the 22.5% warrant case: 29.53% → 19.30%, a 34.6% relative reduction. Cumulative value impact is far smaller and, in the warrant cases, is a transfer that only occurs in states of the world where the equity would otherwise be worth little.

5.10 Control and governance

Preserved without amendment: the April-2026 Shine I / Petrobras Shareholders' Agreement and its consensus mechanism [V q2itr.txt note 1]. Non-negotiable - it is the documented basis for equity-method accounting.

Creditors receive:

• 2 of 11 Board seats, independent, from a slate of 5 proposed by the AHG, vetted for independence under CVM standards;
• observer rights at the Finance Committee;
• monthly liquidity and covenant reporting, quarterly management calls, annual budget delivery, and the ongoing diligence access they requested [V june25ocr/014];
• support for an independent Chief Restructuring / Transformation Officer reporting to the Board, scope limited to liquidity, working capital and the restructuring, jointly selected by the controllers with a creditor consultation right.

Creditors do not receive: any veto over the business plan, capex approval rights, CEO/CFO appointment rights, or any right that impairs "consensus among the controlling shareholders."

Cash-use guardrails - conceded early and in full, because the AHG demanded them [V june25ocr/013-014] and they cost Petrobras nothing; indeed as a shareholder we want them:

• No dividends, buybacks or capital reductions until net leverage < 3.5x for two consecutive quarters and unrestricted liquidity > US$1.5bn.
• No related-party leakage. No new intercompany loans, guarantees, upstreaming or asset transfers to Braskem Idesa, Novonor, Shine I, Petrobras or their affiliates outside the ordinary course without AHG-majority consent. Petrobras accepts being bound on identical terms to everyone else, including on its own supplier-credit facility, which will be documented on published arm's-length terms with an independent fairness opinion. We volunteer this; it is the cheapest credibility available to us.
• Strategic capex capped at US$200m/yr through 2028. Management's plan carries c.US$900m of strategic capex 2026-30, mostly Transforma Rio [V june25ocr/019]. Transforma Rio proceeds only with committed BNDES or third-party financing, never with internally generated funds - this answers the AHG's specific, named objection [V june25ocr/013] at zero cost.
• Excess cash sweep: 50% of unrestricted cash above US$1.25bn, semi-annual, applied first to Tier 0 then pro rata to extended claims at par. (Movement: to US$1.0bn.)
• Asset-sale sweep: 100% of net proceeds above US$50m/yr, same waterfall.
• Minimum unrestricted liquidity covenant US$800m, tested quarterly, with the PIK toggle as the release valve.

5.11 Minority, BAK and BRKM5 participation

• Minorities hold 29.53%: ADRs 13.24%, NSP Investimentos 3.14%, other 13.15% [V]. NSP is itself in judicial reorganisation and will not subscribe [I].
• Full statutory preemptive rights (Art. 171) for every holder at the same price as Petrobras, 30-day exercise, plus a sobras (leftover) round.
• ADR mechanics. A registered US rights offering is slow and expensive. Standard solution: rights trade on B3 as BRKM5 direitos; the depositary sells ADR holders' rights into the B3 market and remits net proceeds, with a Rule 144A / Reg S placement of the sobras. This is a genuine execution constraint on timing and we flag it rather than assume it away.
• Consequences of non-subscription: quantified in §5.9. Percentage dilution 29.53% → 24.90% after the rights issue; value loss to a non-subscriber only 3.5%, and zero if rights are sold.
• No squeeze-out, no delisting, no tender offer. Shine I represented on 3-June-2026 that it does not intend to cancel Braskem's public registration within one year [V bak20260605_6k.htm]. Petrobras supports maintaining the listing and the ADR programme.
• Rights offering structured entirely in Class A preferred, with preemptive rights allocated on total capital. This preserves the 47.03/50.11 voting balance automatically and avoids the ON authorized-capital constraint (only 84.0m ON of headroom [V], against ~88.1m ON that a proportional offering would require).

5.12 Mexico - Braskem Idesa, ring-fenced

1. No Petrobras guarantee, indemnity, keep-well, comfort letter or cross-collateralisation of any kind for Braskem Idesa. Ever. If Mexico enters the Braskem S.A. perimeter, Petrobras's contribution goes to zero. We state this in Round 1 so it cannot be traded later.
2. Braskem S.A. is prohibited from advancing further cash to Braskem Idesa beyond the already-committed undrawn ~US$51m of the existing term loan [V US$180m committed / US$129m disbursed], and only matched dollar-for-dollar by Grupo Idesa.
3. No Braskem S.A. guarantee of Braskem Idesa Chapter 11 exit financing.
4. No Braskem Idesa claim is treated in the Braskem S.A. plan. Separate estates, separate creditors, separate proceedings.
5. Braskem should be prepared to accept deconsolidation of Braskem Idesa through its Chapter 11 - Braskem Idesa's own filings contemplate impacts on its shareholding control [V]. Deconsolidation removes R$14,156m = US$2,738m of debt and negative R$2,467m of equity [V] from Braskem's consolidated balance sheet and is, on these figures, accretive to Braskem S.A.
6. Braskem's intercompany claims (~US$211m) should be subordinated but not written off, and Braskem must assert its lien over Braskem Idesa assets securing the US$82m WC facility [V].

5.13 Alagoas treatment

1. Not restructurable. The obligations arise from judicially ratified agreements with the MPF, MPE, DPU, DPE, the Municipality of Maceió and the State of Alagoas [V q2release.txt]. Any attempt to impair them would fail, reopen settled litigation and be politically fatal. Braskem's own 2H26 priorities list "fulfil the commitments established in the agreements related to the Alagoas Case" as a non-negotiable value [V].
2. Carved out of everything. Alagoas payments are permitted at all times regardless of default, and are excluded from every cash sweep, liquidity covenant, restricted-payment basket and collateral package.
3. Ring-fenced reserve. R$500m (~US$97m) held in a segregated account, excluded from the liquidity covenant and from creditor collateral. (Braskem already maintains Alagoas-specific cash balances and reserve accounts [V q2pres.txt].)
4. No Petrobras guarantee, indemnity, contribution or credit support of any kind toward Alagoas. This is the clearest of our red lines. Petrobras has spent six years avoiding any assumption of Alagoas liability, and there are live TCU and CVM proceedings relating to the geological event [V q2release.txt]. Any Petrobras support characterisable as funding Alagoas would create an independent TCU exposure for Petrobras's own directors.
5. Sizing for the record: US$260m of Alagoas cash outflow in 2026 [V, sum of june25ocr/024]; remaining provision R$3.2bn = US$619m, of which 32% current [V]; ultimate cost may differ materially [V, management's own caveat].

5.14 Annual cash-interest burden

Coverage, on management's own cash-flow lines [V june25ocr/025]:

In the persistent case, H2-26 recurring operating cash flow of US$1,133-1,681m [I §1.4] against half-year cash interest of ~US$190m clears with a very wide margin.

Honest statement for the table: this package is affordable in the persistent-spread case and requires the toggle in the normalization case. That is the structure of the deal. The warrant step-up fires in exactly the state where the toggle is used, so creditors are compensated precisely when they are proved right.

5.15 Expected post-deal liquidity

Built off management's own restructuring-proposal cash flow [V june25ocr/028: unrestricted cash 1,602 / 1,695 / 1,900 / 2,105 / 2,221 for YE26-30], adjusted for the fact that our package pays cash interest where management paid none [I]:

Minimum liquidity covenant US$800m; toggle trigger US$1.2bn. Both cases clear the covenant; the normalization case approaches the toggle trigger by 2029, which is the intended and disclosed behaviour.

5.16 Pro forma leverage

Corporate gross debt 30-Jun-26 US$10.3bn; adjusted net debt US$9.5bn; reported leverage 6.74x [V] (implying LTM EBITDA of US$1.41bn - 9.5/6.74).

Pro forma at YE2028 under our package [I]: gross 10.3 + PIK accretion (380 × 2 = 760) + PIK'd extension fee 95 − rights proceeds applied 150 = US$11.0bn; net US$9.6bn.

We must read this table honestly. At US$1.5bn of EBITDA, 7.3x gross is not refinanceable in 2031-2034 and a maturity-extension-only deal merely relocates the problem. At US$2.5bn and above the structure is money-good, and creditors receive par plus a step-up coupon plus warrants for free. The entire negotiation is a wager on where 2029-2031 lands in this table. Our package is designed so that creditor compensation rises exactly as the outcome deteriorates - 7.5% → 15% → 22.5% warrants, +100 → +200 → +300bp coupon, and an automatic second lien on the crackers at 4.5x. That is the economically correct answer to the AHG's demand for positive-NPV compensation, and it is what we should say when they demand cash today.

6. Explicit rejection terms - what Petrobras vetoes, and on what grounds

7. The strongest argument against us - and our answer

7.1 The argument, stated at its strongest (we should expect the AHG to make it roughly this well)

"Your own management's business plan says 2027 EBITDA is US$1,494m and 2028 is US$1,524m. Against US$10.3bn of corporate gross debt that is 6.9x. Your own status-quo plan shows unrestricted cash at negative US$821m by December 2026 and negative US$1,981m by 2027. Your own restructuring plan only works because it pays creditors nothing in cash for thirty months and cuts the coupon by 200bp. Braskem's book equity is negative R$12.6bn and its market capitalisation is US$0.98bn against US$9.5bn of net debt - the market has already told you the equity is worthless. Petrobras is the only party here with an investment-grade balance sheet, and it is offering the least loss-absorbing capital of anyone: US$54m of real equity, a secured revolver that ranks ahead of us, and US$250m of payment-terms extension on its own naphtha sales - which is not a contribution at all, it is Petrobras protecting its own US$2.0bn of annual revenue. If the equity is out of the money, the economically correct answer is that the creditors own this company, and Petrobras's insistence on preserving 36% of a zero is destroying value for everyone - including Petrobras."

That argument is substantially correct on its arithmetic. Our answer must concede the arithmetic and contest the inference.

7.2 Our answer, in four parts

(1) We concede the arithmetic; we contest the date. At US$1.5bn of EBITDA the equity is out of the money. We agree, and we will say so. But the same plan the creditors are quoting shows EBITDA at US$1,710m in 2029, US$1,866m in 2030, US$2,304m in 2031, US$2,621m in 2033 and US$3,030m in 2035 [V june25ocr/023]. The document that proves insolvency in 2027 proves solvency in 2031. Equitising today at a US$0.98bn market capitalisation does not transfer a zero - it transfers the entire recovery, for free. The correct instrument for a disagreement about the future is a contingent claim, not a permanent one, and we have offered one: warrants stepping 7.5% → 15% → 22.5% precisely as the creditors' case is proven right, plus a coupon stepping +100 → +200 → +300bp, plus an automatic second lien on the crackers at 4.5x. If creditors are right, they are paid. If they are wrong, they are not paid for being wrong.

(2) The near-term evidence is observable, not forecast, and it is on our side. The 3Q26e PE-naphtha spread of US$316/t that anchors the pessimistic case comes from a deck filed 14-August but which states on its own face that it "was updated as of June 30, 2026" [V bak20260814_6k.htm]. It is a June-vintage consultant forecast. The actual 2Q26 spread was US$773/t against a ten-year average of 555 [V]. That forecast has 3Q26 running 43% below the ten-year average and 18% below even the depressed 2022-25 average - a below-trough quarter immediately following a US$773 quarter. Platts reported Americas PE stable to rising through late July with broadly stable CFR Brazil grades, and spreads held into mid-August. If the persistent case holds, 2026 EBITDA is US$2.6-3.3bn rather than US$2.17bn, pro forma net leverage is 3.8-4.8x rather than 6.4x, and the insolvency argument evaporates for two years. Creditors are asking to be paid today, in permanent equity, for a forecast that the market has already begun to falsify. We will pay for that risk in a contingent instrument. We will not pay for it in permanent capital.

(3) The need is predominantly a refinancing need, not evidence of recurring US$547m quarterly operating burn - and the numbers are theirs, not ours. Management's own status-quo plan shows H2-2026 recurring operating cash flow of +US$978m and working capital reversing +US$518m positive after the −US$547m consumed in Q2 [V june25ocr/024, q2pres.txt]. The negative-US$821m year-end cash balance is driven mainly by US$860m of LC runoff, US$325m of cash interest and US$1,242m of amortizations of which US$1,004m is the RCF on 31 December. And management's own restructuring case, with zero new money, produces US$1,602m of unrestricted cash at YE26 rising to US$2,105m by 2029 [V june25ocr/028]. Extension addresses the maturity wall, while replacement trade finance addresses the structural working-capital constraint. The question is the fair price of that support.

(4) On the charge that our supplier credit "isn't a contribution" - they are right, and that is exactly why it is worth more than a guarantee. Yes, the US$250m naphtha term extension protects Petrobras's own US$2.0bn of annual sales. We will say so openly. That is precisely why it is credible and durable: it is the one form of support Petrobras will still be providing in 2031, because Petrobras wants to. A guarantee is support a sponsor spends the next five years looking for a way out of; a trade line is support that is renewed every month because it serves the provider. Braskem's advances to suppliers have tripled to R$1,625m, reverse factoring is contracting, and US$315m of LCs have already been cash-settled by banks and converted into financial debt in Q2 and July alone [V]. Trade credit is the binding constraint on this company, and Petrobras is the only participant who can move it. Meanwhile the burden-sharing test cuts the other way from where the AHG is pointing it: Shine I closed its purchase of control on 3-June-2026, twenty-two days before the cleansing materials were published and the injunction was sought [V both dates]. The newest, best-informed money in this structure should be the most loss-absorbing money. Petrobras will not fund ahead of it, and Petrobras's US$275m demand of Shine I - with at least US$150m loss-absorbing - is not a bargaining position. It is the condition.

7.3 What we should privately concede is true

For internal discipline, so we are not surprised:

• Our accounting carrying value is ~zero; our marginal loss in a wipeout is the US$355m of market value plus ~US$100m of commercial NPV. Creditors will work this out.
• Our RJ BATNA of ~US$185m means our defended value is roughly US$170m of stake value plus strategic option - not a large sum against a US$9.5bn negotiation. We must not overpay to avoid RJ.
• Our veto is a pre-RJ asset that decays sharply after 25-August and is largely worthless inside an RJ where a creditor alternative plan can equitise without our vote [I].
• A large equitisation at a fair pre-money (≥US$1.0bn) can leave Petrobras better off in value than a no-equitisation PIK deal (US$409m vs an option-weighted US$709m only if the company survives; US$409m vs ~US$0 if it does not). We should not reflexively refuse equitisation. We should refuse cheap equitisation, and we should refuse loss of control. Those are different objections and only one of them is legally grounded.

8. Round 1 package on one page

Prepared 2026-08-15. Nine days to the stay expiry. Every material figure verified to .context primary sources or explicitly marked as our inference.