Record Profit, a Covered Dividend, and $1.9 Billion of Revenue Brasilia Has Not Yet Paid

Results vs. Consensus

2Q26 Scorecard

Income Statement, 2Q26 vs. 2Q25

Income Statement, 2Q26 vs. 1Q26

Revenue

The 42.8% sequential increase decomposes cleanly, and the decomposition is the whole story of the quarter. Foreign-market revenue rose 65.0% to $12,914M on a 9.7% increase in external sales volume, which means realisation did roughly six times the work of volume. Crude export revenue alone rose 81.6% to $10,381M on a 12.2% volume increase to 996 Mbpd, an implied realisation of about $114.50 a barrel against a quarterly Brent average of $104.52. In the first quarter the same calculation produced roughly $71.50 against an $80.61 Brent. That eleven-point discount became a ten-point premium in a single quarter, and the reason is exactly what management told us in May: cargoes in transit at the end of March were recognised in the second quarter at second-quarter prices. The backlog cleared, and it cleared into a higher tape.

Domestic revenue is the softer half. It rose 31.7% to $20,693M, but domestic oil products sales volume was 1,742 Mbpd against 1,745 Mbpd in the first quarter and 1,714 Mbpd a year ago, which is to say flat sequentially and up 1.6% year over year. Total domestic sales volume across all products was 2,064 Mboed against 2,071 Mboed a year ago, marginally lower. Every dollar of domestic growth is price, and $1,971M of the sequential domestic increase is the growth in the three subsidy lines. Excluding those lines entirely, domestic revenue rose 19.4% and group revenue was $31,508M, which sits 2.2% above the $30,831M consensus rather than 9.0% above it.

Margins

Gross margin of 58.0% is the highest this company has printed in the period we track, 1,041 basis points above a year ago and 982 above the first quarter. The mechanics are favourable in an unusually literal way: revenue rose 59.8% year over year while cost of sales rose 28.0%, because the two largest cost lines behaved very differently. Production taxes rose 80.3% to $4,607M, tracking Brent as royalties and special participation must. But acquisitions including imports rose only 2.0% to $3,622M, and within that, oil products purchases fell 49.6% to $800M as record refinery utilisation displaced imported barrels. Depreciation rose 17.9% in cost of sales, a marked deceleration from the 26.6% group increase that ate the first quarter's operating leverage.

Below gross profit the picture is less clean. Operating expenses rose 50.1% sequentially against a 71.9% gross profit increase, and the composition matters: Other taxes rose from $474M to $1,184M almost entirely on the crude and diesel export tax, and selling expenses rose 15.2% to $1,746M on freight and logistics for a record export programme. Operating margin still expanded 907 basis points to 42.4%, but the operating expense line is now growing faster than volumes and contains a levy that scales with the export book.

EPS

Reported earnings of $1.62 per ADS and recurring earnings of $1.72 sit in the reverse of their April relationship, and the reversal is instructive. Foreign exchange gains, which contributed $2,118M in the first quarter and $1,735M a year ago, contributed only $379M this quarter as the real appreciated just 4.0% against the dollar rather than the double-digit move of the prior period. Net finance income swung from $1,467M of income to $288M of expense. The reported number is therefore a cleaner read on operations than it has been in either comparison period, which is the correct way to interpret a quarter in which reported earnings rose 68.2% sequentially while recurring earnings rose 144.2%.

One item does deserve a haircut. The effective tax rate of 25.8% is 752 basis points below the first quarter and roughly eight points below the 34% statutory rate, and the gap is interest on capital. The filing discloses a $617M income tax reduction from the first-quarter remuneration approved on May 11 and a further $545M from the 2025 complementary dividends, both recognised in the second quarter. Add $1,162M back and the effective rate is 34.1%, net income attributable is roughly $9,266M and EPS is about $1.44 per ADS. This is a real and permanent Brazilian tax shield rather than an accounting trick, and the same pattern produced an identical 25.8% rate in 2Q25. It is nonetheless a quarter-shape effect: the first half rate is 28.8%, and the full year will land nearer 30% than 26%.

Segment Performance

Exploration & Production

The upstream did in one quarter what the entire first-quarter recap said it had failed to do. Segment revenue rose 42.4% sequentially and 58.2% year over year, gross profit rose 71.1% and 72.2%, and operating income rose 70.2% and 109.1% to $12,456M. In April we wrote that a 16% volume increase delivering a 3% earnings decline was the single most important number in the release. This quarter a 3.4% sequential volume increase delivered a 70% earnings increase, because the realisation gap closed. Segment EBITDA margin of 70% is 7.4 points above a year ago and segment ROCE reached 11.7%.

The operating detail is stronger than the headline. Total own production of 3,336 Mboed is a record, up 14.1% year over year, with pre-salt crude of 2,299 Mbpd carrying it at plus 15.8%. P-79 started on May 1, three months ahead of the 2026 to 2030 business plan and five months ahead of the prior plan, and achieved first gas injection 56 days later, a company record for owned units. Buzios averaged above one million barrels a day for a full month for the first time and hit 1.2 MMbpd on June 26. Mero averaged approximately 740 Mbpd. Ten new producing wells started up, and twenty wells including injectors were tied in.

"In summary, even with the record production of 2.7 million barrels of oil per day, we still have 270,000 barrels per day of capacity to ramp up in the second half."
— Fernando Melgarejo, Chief Financial and Investor Relations Officer

The cost side improved as well, which is not what a ramping upstream usually delivers. Lifting cost in Brazil fell 6.3% sequentially to $6.33 per boe, with pre-salt at $4.45, on higher volumes, fewer maintenance losses and a pre-salt share of the production mix rising from 82% to 83%. That came despite a 4% appreciation of the real, which mechanically raises dollar costs.

Assessment: Bull pillar one is confirmed without qualification, and the earnings conversion that was missing in the first quarter has been demonstrated. The caution is that the demonstration required a $104.52 Brent. At the segment level, lifting cost plus production taxes plus leases rose 14.0% sequentially to $26.56 per boe while lifting cost alone fell, which is the government take scaling with price. The barrels are getting cheaper to produce and more expensive to own.

Refining, Transportation & Marketing

Refining set every operating record available to it and earned less money than in the first quarter. The utilisation factor reached 101.2%, an all-time high that surpasses the 101.1% set in 3Q14, with April and May both at 102.5%. Oil products output of 1,918 Mbpd rose 5.6% sequentially with 68% of the slate in diesel, jet fuel and gasoline. S-10 diesel set a record at 509 Mbpd and jet fuel at 109 Mbpd. Oil products imports fell to 67 Mbpd, the lowest quarterly volume on record, and diesel imports fell 63.3% sequentially to 18 Mbpd. Pre-salt crude reached 73% of processed feedstock.

Segment revenue rose 45.1% and gross profit rose 14.6% to $5,185M. Operating income fell 20.2% to $2,809M and segment net income fell 16.5% to $1,920M. Three things account for the divergence. Selling expenses rose on freight for the record export programme. Other taxes within the segment rose from $136M to $991M on the crude and diesel export levy. And the first-quarter base contained a $414M impairment reversal at the fertiliser unit that did not repeat.

The gross profit itself needs a second haircut. The company discloses that inventory turnover contributed $1,700M in the second quarter and $1,300M in the first, and that excluding it, gross profit would have been $3.5B against $3.2B. On that basis the segment improved 9% sequentially rather than 14.6%, which is a fair reflection of six points of utilisation and a richer slate, and nothing more.

"In refining, we hit a record refinery utilization with a 101% FUT, increasing production by 68% of the yield mix and higher value-added products. In April and May, we came to about 102% FUT, a record for the company."
— Fernando Melgarejo, Chief Financial and Investor Relations Officer

Assessment: The operational half of bull pillar two is now beyond argument: a system running at 101% with record-low product imports is a genuinely different asset from the 65%-to-70% utilisation plant of a few years ago, and it is why a crude spike now produces export margin instead of an import bill. The financial half is the problem. A segment that earns 11% EBITDA margins at a record utilisation factor, with roughly a third of its gross profit coming from inventory appreciation and a new export levy sitting in its cost base, has less earnings power than the headline improvement suggests.

Gas & Low Carbon Energies

The smallest segment continued the improvement it began in the first quarter. Revenue rose 9.1% sequentially to $2,406M, operating income rose 63.1% to $274M and segment ROCE reached 4.1% from 1.1% a year ago. The driver is contractual rather than volumetric: natural gas sales volume actually fell 2.2% to 45 million cubic metres a day on production and pipeline interruptions, while the realised gas price rose 13.7% to $9.97 per MMBtu on the quarterly Brent-linked adjustment in gas contracts. Electricity sales fell 24.4% sequentially, and the average electricity price fell 13.2% to $54.81 per MWh, though the 1,120 average MW of capacity-reserve obligation contracted last year continued to earn a fixed return.

The strategic development is a new mechanism to insulate gas customers from Brent. Petrobras introduced a price band with a floor and a cap for natural gas sales, available by contract amendment, explicitly to blunt the pass-through of Middle East-driven volatility.

Assessment: The segment is doing what we said in April it would do, converting idle thermal capacity and new gas supply into contracted revenue that is less Brent-dependent than the rest of the company. The price band is a small but genuine piece of commercial engineering, and it is also a reminder that this company routinely chooses to give up upside to preserve domestic volume and political position. That instinct is the thesis in miniature.

Revenue by Product, and the Subsidy Line

The revenue-by-product exhibit is where this quarter stops being a straightforward beat. Three separate government fuel-subsidy programmes now sit inside domestic revenue as their own line items, and together they very nearly match the company's entire domestic crude and natural gas revenue combined: $2,099M against $2,257M.

Two readings follow directly. First, the subsidy programmes are 6.2% of total revenue, and their $1,971M sequential increase is 19.6% of the entire sequential revenue increase. Second, and more consequentially, the programme expanded from one fuel to three between March and June. In April the disclosure described a road-diesel subsidy of R$1.50 a litre granted in two tranches. This quarter gasoline and LPG have their own lines. That is a widening of the mechanism, not a wind-down of it.

Operating Metrics

Two rows deserve to be read against each other. The domestic basic oil products price rose 32.0% sequentially in dollars against a 29.7% Brent increase, which on its face refutes the pass-through complaint we made in April. But the domestic sales volume line did not move, and $2,099M of the domestic revenue that produced that realisation came from the treasury rather than from a customer. The price did follow Brent; the payer changed.

Cash Flow, Returns and Balance Sheet

The distribution is covered, for the first time. The board declared R$17.4B of shareholder remuneration, or R$1.34814262 per common and preferred share, which is 45% of the $7,659M of free cash flow the policy formula measures. Post-lease free cash flow of $4,884M covers it 1.4 times. In April we wrote that the declared distribution exceeded post-lease free cash flow and that gross debt had risen in a record quarter; both criticisms are answered this quarter. Payment is in two instalments, in November and December for B3 holders and in December for ADR holders, with the first instalment entirely as interest on capital and the second split between dividends and interest on capital.

One item ADR holders should price. Law 15,270/2025, effective January 1, 2026, imposes a 10% withholding tax on dividends distributed abroad to individuals or legal entities regardless of amount, with limited statutory exceptions, and a companion measure raised the withholding on interest on capital. The gross distribution yield discussed later in this note is therefore not the yield an offshore holder receives.

Key Topics & Management Commentary

Overall Management Tone: Management was more confident than in April and considerably less forthcoming. The prepared remarks were an operational recital of records, framed repeatedly around the argument that the result was earned by the refineries and the reservoirs rather than by the oil price, and that framing was pressed hard enough to become the message. Every subject that connects the record to cash reaching shareholders, the subsidy receivable, the export levy, the working-capital drag and the 2026 debt target, was absent from the script and unprompted in the room. Analyst questioning was cordial and almost entirely forward-looking, so the omissions were never tested.

1. The Catch-Up Arrives at a Ten-Point Premium to Brent

The central prediction of the April note was that the whole export book would reprice in the second quarter, and it did. Crude export volume rose 12.2% to 996 Mbpd while crude export revenue rose 81.6% to $10,381M, which implies a realisation of roughly $114.50 a barrel against a $104.52 Brent. The equivalent first-quarter calculation gives about $71.50 against an $80.61 Brent. The company attributes the difference to production, price and the recognition of exports that were in transit at the end of the first quarter.

The mechanism is worth stating precisely because it cuts both ways. Petrobras recognises export revenue on the trade date, so a rising strip leaves realisations behind and then hands them back a quarter later. The first quarter surrendered roughly eleven points to Brent; the second quarter recovered roughly ten. Over the half, crude export revenue of $16,096M on 942 Mbpd works out near a $94.40 realisation against a $92.57 Brent, a two-point premium rather than a ten-point swing in either direction, and that is the truer picture.

Assessment: The timing thesis is now settled empirically, and it should stop being a source of surprise in either direction. The practical model implication is that quarterly realisations lead or lag Brent by roughly one quarter with a swing of ten points either way, so the right unit of analysis for this company is the half, not the quarter. Anyone extrapolating a $114.50 realisation into the third quarter is extrapolating the tail of a correction.

2. The Subsidy Becomes a Business Line

In April the road-diesel subsidy was a R$740M receivable and a topic the CEO discussed at length. This quarter it is $1,923M of revenue with two siblings, a $159M gasoline programme and a $17M LPG programme, and it went unmentioned. The word "subsidy" does not appear in the call transcript. The closest management came was an oblique reference in response to a direct question about domestic prices sitting below import parity.

"Given that context, our policy is maintained. Our commercial strategy is maintained. Our importing decisions are still based on the criteria of competitiveness and profitability. And we have to say that we have a public policy in force that gives us support for the internal market and that leads the customer perceived prices to be lower."
— Angelica Laureano, Executive Officer for Logistics, Commercialization and Market

The cash consequence is disclosed only in the written management report. Operating cash flow was reduced by $3.2B of working capital, "mainly due to higher accounts receivable, including amounts related to the fuel subsidy program (US$ 1.9 billion), and lower trade payables (US$ 0.8 billion)." The trade and other receivables line consumed $1,951M of operating cash against $245M in the first quarter. In April management told us the March tranche would be collected "within this quarter." It was not; the balance grew by roughly an order of magnitude.

Assessment: This is the most important disclosure in the release and it appears in one sentence of a written exhibit. A company that books 6.2% of revenue as a claim on a finance ministry, expands that claim across three fuels in a single quarter, does not collect the prior tranche and then does not mention any part of it on a ninety-minute call, is not managing that exposure as an exposure. Investors should treat the subsidy line as revenue with a collection risk and a policy duration, and haircut it accordingly.

3. The Export Tax Nobody Mentioned

The one-off events table discloses a $965M charge for "Export tax on crude oil and diesel" in the second quarter, against $122M in the first and nothing in either 2025 period. It is the single largest item in the one-off schedule, it accounts for most of the 149.8% sequential increase in Other taxes to $1,184M, and the bulk of it sits inside the Refining segment, whose own Other taxes line rose from $136M to $991M. It is the reason a segment with record utilisation and higher gross profit reported a 20.2% sequential decline in operating income.

Like the subsidy, it was never discussed. The phrase does not appear in the transcript. Management's only acknowledgement is in the written report's explanation of why net income lagged gross profit, which cites "higher tax expense, especially due to higher expenses related to taxes on crude oil exports."

Assessment: Taken together with the subsidy, the two flows describe a single policy architecture: the state pays Petrobras to hold domestic fuel prices down and taxes Petrobras on the barrels it sends abroad. The net effect this quarter was positive by roughly $1.1B, which is why nobody complained. The structural effect is that a larger share of the company's economics is now set in Brasilia rather than by the crack spread, and that both legs are classified by management as one-off despite one of them scaling directly with the export programme the entire growth plan is built around. An export levy that runs at nearly $1B a quarter is not a one-off; it is a tax on the strategy.

4. Refining Sets Every Record and Earns Less

The utilisation factor of 101.2% is the highest in the company's history, beating the 101.1% of the third quarter of 2014, and it was achieved while holding the high-value slate at 68% rather than by pushing volume into cheap products. Oil products imports of 67 Mbpd are a record low, diesel imports fell 85.2% year over year, and crude imports of 89 Mbpd are the lowest since the pandemic. Refining cost per barrel fell 2.2% sequentially to $3.18.

None of that reached the segment's bottom line. Adjusted segment EBITDA fell 7.4% sequentially to $3,562M and EBITDA margin fell six points to 11%. Freight and the export tax took the gross profit gain, and $1.7B of the gross profit that remained is inventory turnover.

Assessment: The industrial achievement is real and durable, and it is the reason Brazil imported almost no diesel in a quarter when Brent averaged over $100. The financial return on that achievement is being captured elsewhere. This is the clearest illustration in the release of why we hold the equity at a lower multiple than the operating record alone would justify.

5. The $67 Billion Target Disappears

In April the CFO was explicit: convergence to $67B of gross debt during 2026 and $65B by the end of the plan horizon, against a $75B ceiling. This quarter the 2026 figure is gone. It appears nowhere in the call and nowhere in the results deck, whose debt chart now labels its $65B bar as an average of the annual projections for 2027 to 2030.

"We maintain our expectation of converging to $65 billion over the horizon of this plan, a level that optimizes our capital structure."
— Fernando Melgarejo, Chief Financial and Investor Relations Officer

The arithmetic explains the edit. Gross debt fell $408M sequentially to $70,806M, in the quarter that produced $7,659M of free cash flow and $12,250M of operating cash. Reaching $67B by December would now require $3.8B of reduction across two quarters, one of which carries the year's heaviest maintenance programme. The leverage ratios did improve sharply, with net debt to LTM adjusted EBITDA falling from 1.43 times to 1.14, but that is EBITDA doing the work rather than debt.

Assessment: A target that is quietly withdrawn rather than formally revised is worse than a missed target, because it removes the yardstick without acknowledging the miss. No analyst asked about it. This is the specific disclosure we will grade management against next quarter, and the single most concrete piece of evidence that the deleveraging pillar remains contingent on the oil price rather than committed.

6. The Lease Amendment: Trading Reported Debt for Cash

The composition of the debt moved more than the total. Financial debt fell 6.2% to $25,834M, driven by a $1,495M reduction in banking-market obligations, while finance leases rose 3.0% to $44,972M. The company settled $2.9B of loans and financings, including a $1.4B prepayment and the early redemption of $670M of 7.375% global notes due 2027, and raised $614M.

The lease increase is deliberate. Petrobras renegotiated recharter and well-service contracts during the quarter, extending terms in exchange for lower payments, which requires immediate remeasurement of the lease liability upward while reducing future cash outflows.

"Now because the amendments extending the contract terms were signed in the second quarter, we had to reorganize the value of these contracts and lease liabilities immediately. So these increase in the short term, but we reduced our future disbursements and cash flow creating value, and that's the key benefit."
— Fernando Melgarejo, Chief Financial and Investor Relations Officer

Management sized the benefit at over $1B of cash flow savings across 2026 to 2030 and over $400M of debt reduction by 2030.

Assessment: This is good treasury work and the accounting penalty is a fair price for it. It also means the reported gross debt line understates the quarter's underlying deleveraging, and the weighted average maturity extension from 11.33 to 11.92 years at an unchanged 6.8% average cost is genuinely valuable. The honest scorecard: $1.7B of real debt reduction was achieved and $1.3B of it was optically absorbed by the lease remeasurement.

7. Capex to the Top of the Range, Opex Through It

Total capex of $5,299M rose 3.8% sequentially and 19.6% year over year, with 82% in E&P. Cash capex of $4,559M brings the half to $9,072M, or 53.7% of the $16.9B annual plan. Manageable operating expenses of $11.7B for the half are 57.9% of the $20.2B target. The CFO's framing of both was notably softer than in April.

"With regard to cash investments, we expect to end the year at the top of the range. The projection is $16.9 billion with a 5% margin, give or take."
— Fernando Melgarejo, Chief Financial and Investor Relations Officer

"Operating expenses are slightly above plan for this half of the year, pressured by higher freight and logistics, which lead to increased production as well as exchange rate effects. So we totaled $11.7 billion in this half of the year versus a full year plan of $20.2 billion. We're monitoring the situation and expenses may exceed the projection if global market logistics costs and exchange rates remain at the same levels in the next few 6 months period."
— Fernando Melgarejo, Chief Financial and Investor Relations Officer

The two statements are in tension with the arithmetic. If the second half merely matches the first, cash capex ends the year at $18.1B, above even the top of the guided band at $17.7B, and operating expenses end at $23.4B against a $20.2B target. Meanwhile new commitments accumulated during the quarter: the P-81 and P-87 FPSO contracts for the two Sergipe deepwater platforms were signed in May, a 49.99% interest in a Brazilian solar developer was acquired, and the RPBC biorefining project took final investment decision.

Assessment: Bear pillar two escalates. In April the concern was that projects previously outside base capex were being pulled inside it; this quarter the CFO has pre-announced the top of the range on capex and conditionally disowned the opex target, and a new strategic plan is being drafted. The capital envelope is being reset upward at the top of the cycle, which is precisely the pattern this company's credibility discount exists to price.

8. Extraordinary Dividends Ruled Out, and a New Plan Being Written

Asked directly how the incremental cash would be allocated, the CFO restated the same hierarchy as in April and closed the extraordinary-distribution question more firmly than he had three months earlier.

"With regards to extraordinary dividend sharing, we find it very unlikely to look into that possibility even though that is something we would love to do. If we have no investment in sight and our debt is well adjusted, we see no problem with that."
— Fernando Melgarejo, Chief Financial and Investor Relations Officer

He also disclosed that the company is already drafting the next strategic plan, in August, and that its planning view is for Brent to return toward the assumptions embedded in the 2025 to 2030 plan next year.

Assessment: Two things follow. Management is telling shareholders that this quarter's earnings base is not the planning base, which is an honest and analytically useful statement and also a warning against capitalising $19,959M of quarterly EBITDA. And a new plan written in the middle of a $104 tape, by a management team that has just guided capex to the top of its range, is the most likely mechanism by which the windfall is committed before it can be distributed. In April the review was deferred to year end; it has been pulled forward, and not in the shareholders' direction.

9. Lifting Cost Falls While Government Take Per Barrel Climbs 38%

The per-barrel cost ladder Petrobras discloses is the cleanest way to see where the oil price goes. Lifting cost in Brazil fell 6.3% sequentially to $6.33 per boe, and pre-salt lifting cost fell 4.8% to $4.45. Adding leases takes the figure to $9.02. Adding production taxes takes it to $23.88, up 14.9% sequentially and 38.0% year over year. Adding both takes it to $26.56.

In absolute terms, Brazilian production taxes rose 80.3% year over year to $4,605M, split between $3,000M of royalties and $1,595M of special participation. Cash taxes and government take of R$88.6B were paid in the quarter, R$22B more than a year earlier, which management annualised at close to R$88B of incremental national revenue.

Assessment: This is bull pillar three, the falling breakeven, tested for the first time and passing on the half management controls. Lifting cost fell in dollars despite a 4% currency headwind, which is a genuine efficiency result. But the total cash cost of owning a barrel rose 14.0% sequentially, entirely on the government's share. A breakeven that improves on controllable costs while the state's per-barrel claim rises 38% year over year is a breakeven that improves more slowly than the disclosure implies.

10. China Halves as an Export Destination

The export destination mix moved violently. China took 35% of crude exports against 62% in the first quarter and 51% a year ago, a change the company attributes to Chinese measures to curb imports at prevailing prices. India rose from 15% to 22%, Europe from 8% to 18% and the rest of Asia from 8% to 14%. Petrobras secured five new crude customers across Taiwan, Oman, Poland, Sweden and South Africa for Atapu, Buzios, Itapu, Sururu and Tupi grades. Oil products exports remain concentrated, with Singapore at 50% and the United States at 31%.

Assessment: In April we flagged 62% single-buyer concentration as a genuine portfolio characteristic rather than a rounding item. That concentration halved in one quarter without a volume interruption, which is a better outcome than we expected and a real credit to the commercial organisation. It also demonstrates that the concentration was a price-driven placement rather than a structural dependency. Watch whether the diversification survives a normalising Brent, when Chinese buying typically returns.

11. Braskem Escalates to an Injunction With an October Expiry

In April the petrochemical stake was a governance reset with a shareholders agreement expected to close within ninety days and an explicit commitment not to consolidate debt. This quarter the situation has moved materially and management said almost nothing.

"So there is an injunction in place, which is in the public domain, as you mentioned. It ends on October 24. And as the President mentioned as well, on August 13, that company will be sharing their earnings with the market. So this is a very sensitive time. We have a lot of decisions to make."
— Fernando Melgarejo, Chief Financial and Investor Relations Officer

The question that prompted this asked directly about a potential court-supervised reorganisation and about a possible capital injection from Petrobras. Neither was addressed. The CEO declined to elaborate on the grounds that the affiliate reports separately the following week.

Assessment: The unbounded, low-probability liability we flagged in April has become an unbounded, medium-probability liability with a dated deadline attached. There is no figure to model, and management has now twice declined to give the boundary condition that would let one be modelled. In a quarter this strong, an affiliate exposure that cannot be sized is the kind of item that gets ignored until it cannot be.

12. Downtime Deferred Into 2027

Part of the record utilisation has a schedule behind it. Two of the largest planned turnarounds were pushed into next year, which the segment head explained as a deliberate trade against expansion project readiness rather than a stretch of the assets.

"In the specific case of the postponement of REGAP and REPLAN downtimes, we were having ongoing expansion projects. But since they were not yet mature, we decided to postpone them to the beginning of 2027 after reliability analysis, inspections and so on and so forth. We saw that we were able to guarantee the reliability if we did postpone the downtimes to next year and still expand it."
— William Franca, Chief Industrial Processes and Products Officer

He was explicit that no first-half downtime was moved into the second half, that the Cubatao diesel turnaround scheduled for August was always scheduled, and that the remaining second-half work is minor. Separately, the E&P head quantified full-year scheduled upstream downtime at 290,000 barrels a day.

Assessment: The reasoning is sound and the reliability metrics support it, but the effect is to borrow utilisation from 2027. A modelling assumption of 101% utilisation persisting into next year would be double counting: the deferred turnarounds land in the first quarter of 2027 alongside the expansions they were deferred for. The second half should hold near current levels; the following first quarter should not.

Guidance & Outlook

No 2026 target was formally revised, but two were reframed and one was withdrawn.

Implied second-half ramp. Production needs nothing: the first half already ran above the top of the guided band, and 270,000 barrels a day of ramp capacity remains against 290,000 barrels a day of scheduled downtime, so the two roughly cancel and the full year lands near 2.62 to 2.68 MMbpd. Capex is the opposite. Holding cash capex to the guided top of $17.7B requires the second half to spend $8.6B against $9.1B in the first, in a period with two FPSO sailaways scheduled and three Buzios units under construction. Operating expenses would need to fall to $8.5B in the second half to hit the $20.2B target, which management has effectively conceded is not the plan.

Street at: consensus revenue for the quarter sat between $30.8B and $31.5B against a $33.6B print, and the published estimate range going into the quarter spanned $1.25 to $1.52 per ADS on at least two different bases. Post-print, the concentrated Street concern is that full-year capex overshoots the guided envelope by roughly 10%, which would put it near $18.6B.

Guidance style. Unchanged from April in form and softer in substance. Management again declined to raise anything into a strong tape, which remains the right posture for a company whose credibility gap is about capital discipline rather than volumes. But three months ago the unrevised guidance was accompanied by an explicit interim debt target; now it is not, and the two cost lines carry verbal caveats that the printed figures do not. The guidance has less information in it than it did a quarter ago.

What the guidance still does not include. No Brent planning assumption, no revenue or earnings guidance, no capex figure for the projects added during the quarter, and no framework for what would trigger an extraordinary distribution.

Analyst Q&A Highlights

Whether the Production Curve in the Plan Is Too Conservative

The call opened on the gap between the plan's medium-term production curve and what the assets are already delivering, paired with a request for the delivery sequence of the three Buzios units still under construction. Management declined to revise the curve and instead described the reservoir management behind it, then handed to engineering for dates that were more specific than the plan itself.

Q: "But I'd like to see if it's fair to say that the plan curve in the medium and long term does not look very conservative, especially if you look at the highest peak of the curve, and it's at 2.6 million or 2.7 million barrels that you surpassed already."
— Bruno Montanari, Morgan Stanley

A: "So we'll talk about P-80, P-82 and P-83. P-80 and P-82, the sailaway is scheduled for the third quarter of 2026 and production will start in the second quarter of 2027. What we're doing now is we're working strongly towards bringing forward the production of P-80 to the first quarter of 2027. For P-83, the sailaway is scheduled to occur in the beginning of the first quarter of 2027 with production starting on the second half of 2027."
— Renata Baruzzi, Executive Officer for Engineering, Technology and Innovation

Assessment: The answer confirms the curve is conservative without saying so, which has been the pattern for four quarters. The concrete disclosure is that P-80 is being pulled toward the first quarter of 2027, repeating the P-79 playbook. The offsetting disclosure, given later in the same exchange, is 290,000 barrels a day of scheduled downtime for the year, which is why the full-year figure will beat the guide but not by the margin the quarterly run rate implies.

Domestic Fuel Prices Below Import Parity, and What Bridges the Gap

The most consequential question of the call went to why domestic prices sit below import parity again, and how that squares with a commercial strategy said to be based on competitiveness and profitability. The answer restated the policy as unchanged and referred to the bridging mechanism only as a public policy that lowers perceived customer prices, without using the word subsidy, quantifying the $2,099M booked in the quarter, or mentioning the $1.9B receivable it created.

Q: "How is the company assessing the decision to import diesel given the current price scenario vis-a-vis its pricing strategy and its commercial strategy, especially now that with the -- there is a huge volatility? But with the recent increase, we saw, again, local prices below the import parity."
— Monique Greco, Itau BBA

A: "Given that context, our policy is maintained. Our commercial strategy is maintained. Our importing decisions are still based on the criteria of competitiveness and profitability. And we have to say that we have a public policy in force that gives us support for the internal market and that leads the customer perceived prices to be lower."
— Angelica Laureano, Executive Officer for Logistics, Commercialization and Market

Assessment: The exchange is the clearest evidence in the release that the subsidy architecture has been normalised into something management no longer feels obliged to discuss. The question was precise and the answer was a euphemism. Nobody followed up. For a mechanism that now spans three fuels and represents 6.2% of group revenue, that is a disclosure standard that will not survive its first collection problem.

Where the Incremental Cash Goes

A recurring line of questioning asked how a cash flow running well above the plan assumption would be allocated. The response repeated April's three-step hierarchy of investment, then debt, then shareholders, and closed the extraordinary-distribution question more firmly than three months ago while revealing that the next strategic plan is already being written.

Q: "With the very favorable cash generation environment, both because of market circumstances and the company itself. I wanted to understand how that works versus your plan for 2026-2030, for example, thinking about the allocation of this incremental capital that's coming over throughout this year with regards to refining or maybe another ambition Petrobras has in mind?"
— Yuri Pereira, Santander

A: "With regards to extraordinary dividend sharing, we find it very unlikely to look into that possibility even though that is something we would love to do. If we have no investment in sight and our debt is well adjusted, we see no problem with that. The same logic remains since we arrived here, according to which any surplus should be allocated somewhere."
— Fernando Melgarejo, Chief Financial and Investor Relations Officer

Assessment: The ordering is consistent, disclosed and unfavourable to the marginal ADR holder, exactly as it was in April. What is new is the timing. A new strategic plan being drafted in August, in a $104 tape, by a team that has just guided capex to the top of its range, is the mechanism by which "any surplus should be allocated somewhere" becomes a capital budget rather than a distribution.

Braskem, a Possible Court Reorganisation, and a Capital Call

The petrochemical affiliate produced the most pointed question of the session, naming the possibility of a court-supervised reorganisation and asking directly about a capital injection. Management confirmed an injunction with a stated expiry and declined every substantive element of the question, citing the affiliate's own results the following week.

Q: "We're seeing news about a potential legal reorganization. And Petrobras has some active input in these conversations. My question is, how do you see that from Petrobras' side? I understand there's a number of rules with governance and there's great concern about capital allocation. So maybe more discussions about a potential capital injection into Braskem."
— Gabriel Barra, Citi

A: "So there is an injunction in place, which is in the public domain, as you mentioned. It ends on October 24. And as the President mentioned as well, on August 13, that company will be sharing their earnings with the market. So this is a very sensitive time. We have a lot of decisions to make."
— Fernando Melgarejo, Chief Financial and Investor Relations Officer

Assessment: In April the same subject drew an explicit boundary, that debt would not be consolidated and the stake would remain a minority. That boundary was not restated this quarter. A management team that gave a hard constraint three months ago and declines to repeat it, while confirming a dated injunction, has told you something even while saying nothing. This is now the largest unmodellable item on the sheet.

Second-Half Crack Spreads and the Turnaround Calendar

With the system running above nameplate, questioning turned to whether the second half can be held there and what the maintenance calendar does to it. The answers separated cleanly: commercial confirmed that third-quarter seasonality forces diesel imports regardless of utilisation, and the industrial side clarified that two large turnarounds moved into 2027 rather than into the second half.

Q: "You've been running the refining farm at above 102%. And we heard about downtimes as well in the second half. So talking about the refining farm and what can we expect in terms of timing and duration of these downtimes for the second half?"
— Milene Carvalho, JPMorgan

A: "The third quarter is, from a seasonal perspective, a quarter of higher diesel demand. So given that context, our operating planning takes that into account. And obviously, in this quarter, we're going to have to import and the cracking of diesel has led us to attempt to produce as much diesel as we can internally."
— Angelica Laureano, Executive Officer for Logistics, Commercialization and Market

Assessment: The import admission is the same one made in April and it has been consistent since, which is to management's credit given the self-sufficiency framing in the prepared remarks. The genuinely new information is the deferral of the two largest turnarounds into early 2027, which supports second-half utilisation and creates a first-quarter 2027 air pocket that no model built off this quarter's run rate currently contains.

Whether a Four Percent Decline Rate Is Sustainable

A question on reservoir performance drew the most technically substantive answer of the call, and the only one that addressed the durability of the production pillar rather than its near-term trajectory.

Q: "Petrobras' decline is quite low at 4% more or less in the pre-salt. And I'd like to understand how sustainable that is moving forward. If you think about the forecast for the next few years, how do you see this decline behaving in the next few years?"
— Jorge Gabrich, Scotiabank

A: "There is a natural decline, Jorge, and we attempt to fight against it, we work more strongly on four areas. First, 4D seismics, which better understands us -- which better allows us to better understand the reservoirs. We also do intelligent completion."
— Sylvia Anjos, Chief Exploration and Production Officer

Assessment: The four levers described, seismic monitoring, intelligent completion, water injection and infill drilling, are conventional and correct, and the CEO's separate framing that decline has fallen from 12% a year to roughly 4% is the single most valuable long-duration disclosure on the call. It is also unauditable from outside on a quarterly basis. Treat it as the key assumption underwriting the production pillar and grade it over years, not quarters.

Diversification Beyond the Core, and What It Would Cost

A question on the company's public interest in rare metals and frontier basins tested whether the widening of the opportunity set carries any capital commitment. The answer was an unqualified no on commitment and a restatement of the governance process.

Q: "But in addition to Africa, the Mexican Gulf and Latin America, recently, we've also read about rare metals and assessment of the Jupiter basin with the potential offshore exploration, how much do you plan to invest in that segment? I understand the rare metals sort of escape the company's main focus."
— Vicente Falanga, Bradesco BBI

A: "We currently have no commitment in terms of investing in what you mentioned. We have made no commitment in that sense. These are only opportunities that we've been discussing and looking into and assessing both domestically and internationally."
— Fernando Melgarejo, Chief Financial and Investor Relations Officer

Assessment: The right answer, and the same discipline shown in April on Mexico and Venezuela. The pattern across four such questions this quarter is consistent: the company will talk expansively about optionality and will not attach capital to any of it in public. That is reassuring on governance and unhelpful for modelling, and on balance the former matters more.

What They're NOT Saying

1. The word "subsidy," at all. $2,099M of revenue across three programmes, up from $128M across one, and the term appears zero times in the call transcript. The only reference is an oblique mention of "a public policy in force."
2. The $965M export tax on crude oil and diesel. Disclosed only in the written one-off table. Never named on the call, never explained, never given a duration or a rate. It is the largest single one-off item in the quarter and the direct cause of Refining's sequential earnings decline.
3. When the subsidy receivable actually converts to cash. The balance itself is disclosed, but only in note 28.1.1 of the interim financial statements, which puts it at $2,021M at June 30 against $2,227M of net subsidy revenue recognised in the half, discloses a further $844M received in July, and says the remainder is "awaiting documentation analysis by ANP." None of that appears in the release commentary, the deck or the call. What is genuinely absent is a timetable: no expected date for the ~$1.2B still outstanding, no indication of what happens if ANP requires amendments, and no statement of how close the programmes are to the federal budget ceiling that can terminate them early.
4. The $67 billion gross debt target for 2026. Stated explicitly in May, absent from both the call and the deck in August, with no acknowledgement that it was dropped and no analyst question about it.
5. The words "working capital," "receivable" and "free cash flow." None appears in the transcript. The distribution is set by a free cash flow formula and the R$17.4B figure itself was never mentioned on the call.
6. Whether Braskem requires a capital injection. Asked directly for the second consecutive quarter. In May the answer at least drew a boundary on debt consolidation and minority status. This quarter neither was restated.
7. The capex cost of what was added during the quarter. The P-81 and P-87 FPSO contracts were signed in May, a 49.99% solar interest was acquired, and the RPBC biorefining project took final investment decision. Total investment figures appear in the project table but no revision to the annual envelope accompanies them.
8. Any update to the $59 cash breakeven. The most useful disclosure of the April call was not repeated, updated or tested, in the first quarter that provided data against which to check it.
9. The conflict. Brent averaged $104.52 and the word "war" does not appear once in the transcript, having been discussed openly by the CEO in May. The written report refers only to "tensions in the Middle East." Management has chosen to present the record as an operational result rather than a price result, which is a defensible emphasis and a conspicuous one.

Market Reaction

• Pre-print setup: the ADR closed at $18.52 on August 6, the day the results were furnished after the Brazilian close. It entered the print up 56.3% year to date from $11.85 at the end of 2025, up 42.4% over the trailing twelve months and up 11.2% over the trailing thirty days, against a 52-week closing range of $11.54 to $22.03. The S&P 500 was up 12.6% year to date over the same stretch.
• Reaction session: PBR opened at $18.74 on August 7, a gap up of 1.2%, traded between $17.93 and $18.76, and closed at $17.96, down 3.0% or $0.56 on the day. The S&P 500 rose 0.6% in the same session.
• Volume: 28.5 million shares against a 14.5 million thirty-day average, roughly double normal. This was a fully participated session, unlike the 0.9 times turnover that greeted the first-quarter miss.
• Local lines: on B3 the common shares fell 1.69% and the preferred 1.47% in the same session, so the ADR underperformed both by roughly 1.3 to 1.5 percentage points. That gap is currency, not an incremental equity move.
• Call timing: results were furnished after the close on August 6 and the webcast was held on the morning of August 7, so the reaction session incorporates both the print and the call.

A 13.2% earnings beat, a 9.0% revenue beat, records in production, refining utilisation, gross profit and recurring net income, and a distribution above the published Street projection, produced a 3.0% decline on double normal volume. That combination is not a market that missed the numbers. It is a market that read past them.

Three things in the release are visible to anyone who opens the written report rather than the deck. The capital budget is running at 53.7% of plan at the half with the CFO guiding to the top of the range and warning that operating expenses may breach theirs. The extraordinary distribution that a $104 Brent might have funded was closed off in the first Q&A that reached it. And the second half carries the year's heavier maintenance programme against a first half that borrowed two turnarounds from 2027. The stock had already risen 11.2% in the thirty days into the print; the print gave holders the earnings and took back the allocation.

What the tape again did not reward is the operating record. An all-time refining utilisation factor, record oil products output, record-low product imports and a production record were all in the release, and the session went down on twice the normal volume. As in April, this equity is being priced off Brent and off Brasilia rather than off the barrel count. The difference is that in April the market looked through a miss to this quarter's catch-up. This quarter it looked through the catch-up to the capital budget.

Street Perspective

Debate: Is $19,959M of Quarterly EBITDA a Base or a Peak?

Bull view: the bull case on the Street is that the second quarter is the first clean look at what the reshaped asset base earns. Production is up 14.1% with 270,000 barrels a day of ramp still ahead, refining runs above nameplate, lifting cost is falling and the import bill has nearly vanished. On that reading the quarter is a base that grows.

Bear view: the bear camp points out that the same quarter contains a Brent 54.1% above a year ago, $1.7B of inventory turnover, $2.1B of subsidy accrual and a tax rate flattered by roughly $1.2B of interest-on-capital deductions, and that management's own planning view has Brent returning toward plan assumptions next year.

Our take: the bears have the arithmetic. Strip the four items and the quarter is still excellent and materially smaller. The right base is the first half, not the second quarter: $31,696M of adjusted EBITDA excluding one-offs on a $92.57 Brent, which annualises near $63B and is roughly 20% below annualising the quarter alone.

Debate: Does the Covered Dividend Change the Capital Return Story?

Bull view: the optimists note that the single loudest criticism of the first quarter has been answered. R$17.4B was declared at exactly 45% of free cash flow, it was covered 1.4 times by post-lease free cash flow, it exceeded at least one published Street projection, and net debt fell 2.7% at the same time. The formula works when the cash is there, which was the whole question.

Bear view: the skeptics answer that a formula producing a bigger number in a bigger quarter is not a change in policy, that the extraordinary distribution was ruled out in plainer language than in May, that a new strategic plan is being drafted at the top of the cycle by a team guiding capex to the top of its range, and that offshore holders now face a 10% withholding on the dividend component that did not exist a year ago.

Our take: the bulls win the quarter and the bears win the year. The distribution pillar has genuinely improved and we are upgrading its status accordingly. But the ordinary distribution rising mechanically with free cash flow was never in dispute; what was in dispute is whether a windfall reaches holders, and management has now answered that twice, more firmly the second time.

Debate: Is the Subsidy Plus Export Tax Architecture Net Positive?

Bull view: a growing consensus view is that the net flow is positive and that is what matters. $2,099M in and $965M out leaves Petrobras roughly $1.1B ahead, it keeps domestic market share while protecting the consumer, and it removes the political pressure that historically forced genuinely below-parity pricing at the company's own expense.

Bear view: the other side observes that the inflow is an uncollected receivable subject to administrative process and annual fiscal choice, while the outflow is a cash tax collected at the border on the export programme that carries the entire growth plan; that the subsidy expanded from one fuel to three in a single quarter; and that management discussed neither leg on the call.

Our take: the bears, and by a wider margin than in April. In May the concern was that a commodity margin had become a sovereign receivable. This quarter that receivable tripled in scope and went uncollected, and a levy appeared on the other side of the trade. The correct treatment is to model the export tax as a recurring cost of roughly $0.9B a quarter at current export volumes, and to apply a collection haircut to subsidy revenue rather than carrying it at par.

Debate: Does a De-Rated ADR at Three Times EBITDA Already Discount All of This?

Bull view: the bull case is that at $17.96 the ADR trades near three times enterprise value to trailing adjusted EBITDA and under four times annualised first-half recurring earnings, with a gross distribution running near 9% at the first-half pace, having fallen 12.4% since the last print while earnings more than doubled. On any normalised measure that is a very low price for record execution.

Bear view: the bears reply that the multiple is low because the earnings are not the holder's. The state sets the domestic price and pays the difference on its own schedule, taxes the exports, takes 38% more per barrel in production taxes than a year ago, withholds 10% on dividends sent abroad, and controls the board that is writing a new capital plan this month. A three-times multiple on those cash flows is not obviously a discount.

Our take: this is the crux and it decides the rating, as it did in April. The valuation is genuinely low and the de-rating since May is real, which is why we are not downgrading a company that just printed its best quarter. But two of the three upgrade signposts we published failed and a new form of state capture appeared, which is why we are not upgrading either. The equity is cheap for reasons that this quarter reinforced rather than resolved.

Model Update & Valuation Framework

Valuation framework. At $17.96 per ADS across the 6,444 million ADS-equivalent shares implied by the 12,888,732,761 weighted-average shares in note 26.5, the equity is capitalised near $116B on a common-share-equivalent basis, before the discount at which the preferred line trades. Adding $60.4B of net debt gives an enterprise value near $176B. The disclosed 1.34 times gross-debt-to-LTM-adjusted-EBITDA ratio on $70,806M of gross debt implies trailing adjusted EBITDA near $52.8B, so the ADR trades at roughly 3.3 times enterprise value to trailing adjusted EBITDA. First-half recurring net income attributable of $15,608M annualises to $31.2B, which puts the equity near 3.7 times. The two declarations of the first half total roughly $5.3B, or about $0.82 per ADS, which annualises near 9% gross at the first-half pace and materially less after Brazilian withholding.

Valuation impact. Our fair-value range moves to $18 to $22 per ADS from the low-$20s framework we set in April, and the stock trades just below the low end of it. The range narrows and shifts down for a specific reason: the first quarter's discount was a timing artefact that everyone could see would reverse, so we discounted it heavily. This quarter's discount is not a timing artefact. It is a capital allocation posture that management restated more firmly, a new export levy that is not modelled as recurring, a subsidy receivable that did not collect, and a debt target that was withdrawn. Those compress the multiple rather than deferring the earnings. Against that, the operating asset is demonstrably better than it was three months ago and the stock is 12.4% cheaper. A stock trading fractionally below the bottom of a fair-value range is exactly what a Hold with upgrade optionality looks like.

Thesis Scorecard Post-Earnings

The pillars below are the standing thesis established in April, graded against what this quarter's print and call revealed. They are not re-derived.

Overall: thesis unchanged in direction and stronger in evidence on both sides. The operating pillars are confirmed without qualification and the distribution pillar has been repaired. The two pillars that determine whether any of it reaches an offshore holder, capital discipline and deleveraging, both deteriorated, and the state's claim on the company took a new and larger form that management declined to discuss. Petrobras is a better business than it was in April and no more clearly an investable one.

Action: Hold. In April we named three upgrade signposts. The subsidy receivables were not collected in full and instead grew to $1.9B; gross debt did not print below $70B at the half, closing at $70.8B; and the ordinary distribution did step up in line with the free cash flow formula. One of three is not an upgrade, and the two that failed are the two that matter for cash reaching holders. Revised signposts for an upgrade: subsidy receivables collected and disclosed with a balance by the third quarter; the 2026 capital budget confirmed inside the guided band when the new strategic plan is published; and gross debt below $69B at the third quarter. Signposts for a downgrade: a new plan that lifts the 2026 to 2030 capex envelope by more than 10%; a Braskem capital commitment of any size; the export tax being extended or raised; or a negotiated de-escalation that returns Brent to the plan assumption before the subsidy receivables are collected.