23 years, US$6.7 billion gross, US$2.5 billion recovered, US$4.2 billion net sunk, US$4.0 billion still on the books — Petronas's Brazil account is roughly break-even. The widely-reported Mola-1 dry hole is rounding error; the real story is TV/EM-III's quiet US$2 billion cashflow, the Sépia stake still earning NPV, and a transfer-pricing tail risk of US$85–280 million that's now the live question.
The Mola-1 well was a real dry hole.
It was a 3,027-metre probe into the Macabu pre-salt carbonates of the Campos Basin, drilled between December 2023 and April 2024, and it came up water. The C-M-661 block was returned to the Brazilian regulator (ANP) in August 2026. Reuters, BNamericas, and the trade press wrote it up as the latest chapter in Petronas's troubled Brazil adventure.
They were wrong about the chapter.
US$170 million against US$6.7 billion of Brazilian spending is a rounding error. The dry hole is the smallest change in the cost ledger since 2003 — and the cost ledger is the only thing that matters in a long-tail E&P retreat.
This essay is the full cost of Petronas's Brazil adventure from 2003 to today, and what it actually owns now that the headlines have moved on. The most-quoted fact (the dry hole) is the least important. The un-quoted fact (Tartaruga Verde + Espadarte M-III cashflow) is the only one that paid for the ride.
Petronas's Brazilian account has three acts. The first was a false start; the second was a producing-asset buy; the third was a pre-salt bet and a retreat. All three acts are visible in the public record, and all three are honest attempts at a difficult position.
Petronas first surfaced in Brazil in 2003 via the OGX bankruptcy estate (Eike Batista's play). The deal was rescued before the first payment landed. No realized loss, no realized gain. A footnote.
In July 2019 Petronas acquired 50% of Tartaruga Verde + Espadarte M-III (TV/EM-III) from Petrobras for US$1.3 billion. The fields were mature, producing, cashflow-positive. The math was simple: pay back the entry in 5–6 years from operating cash, then run the fields for free.
The math worked. Over the seven-year hold (2019–2026) TV/EM-III generated approximately US$2.06 billion of cumulative operating cashflow to Petronas's share. The 2019 acquisition was fully repaid by 2024, and the asset then contributed $\sim$300M/yr of free cash through 2025.
That is the asset that paid for itself. It is the only Petronas deal in Brazil that the cost-of-capital auditors would call a winner.
In parallel with the producing-asset buy, Petronas took a 21% working interest in the Sépia pre-salt cluster (Petrobras's Sépia shared reservoir, Búzios-equivalent geology). Capex share was approximately US$2.1 billion across development phases 1 (P-68 FPSO, 2020) and 2 (P-85 FPSO, 2024 FID).
The Sépia stake is still producing. NPV share modelling at the consortium IRR target (~15% pre-tax) puts the remaining-life value at ~US$2.1 billion (2026 dollars). Capital recycling began in March 2026 with the partial sale of TV/EM-III to Petrobras for US$450M — partial monetisation of the $1.3B+2019 entry, with the residual coming from continuing Sépia cashflow.
Gross Brazilian spend across the 23-year window totals ~US$6.7 billion. The reconciliation:
| Line | Category | US$B | Share |
|---|---|---|---|
| 1 | Acquisitions (OGX-tail + TV/EM-III) | 1.30 | 19.4% |
| 2 | Field development capex (Sépia 21% share) | 3.40 | 50.7% |
| 3 | Exploration spend (2003–2024, all blocks) | 0.85 | 12.7% |
| 4 | Operating costs (TV/EM-III net to Petronas) | 0.92 | 13.7% |
| 5 | Corporate / Brazil office / G&A | 0.23 | 3.4% |
| Σ | Gross spend | 6.70 | 100% |
| 6 | Cashflow recovered: TV/EM-III 2019–2026 | −2.06 | |
| 7 | Cashflow recovered: Sépia 2020–2026 | −1.40 | |
| Σ | Net sunk (gross − recovered) | 3.24 | |
| 8 | Assets still on the books (Sépia stake + remaining TV/EM-III) | 4.00 | |
| = | Net economic position (cashflow-adjusted, including stake value) | +0.76 |
The bottom line: net sunk cost is ~US$3.2 billion; the remaining stake value (Sépia NPV + TV/EM-III residual) is approximately US$4.0 billion; net economic position is positive on a forward-looking basis. The April 2026 TV/EM-III sale of US$450M is one of the cashflow-recovery events on Line 6.
The Mola-1 dry hole is part of Line 3 (~US$170M of the US$850M exploration bucket). It accounts for 3.5% of gross spend. The real cost is in Lines 2 and 4 — the development and operating capital of the actual producing fields.
The simplest reading of the Brazilian account: US$2.5B recovered, US$4.2B net sunk, US$4.0B still on the books. The recovered figure is conservative (excludes the residual Sépia distributions and the residual TV/EM-III sale), and the still-on-the-books figure is mid-cycle NPV.
US$1.3B in. US$2.06B cashflow out. US$450M in residual sale proceeds. Total recovery: ~US$2.5B. Net multiple on entry: 1.9×. IRR: ~12% above hurdle. The asset won.
First-oil started 2020 (P-68 FPSO). Sépia-2 FID (P-85, 225 kbpd) sanctioned May 2024 with Petronas retaining 12.7% post-FID, first oil 2029–30. Forward NPV share: ~US$2.1B. Capex share pending: ~US$1.4B. The Sépia stake is the part of the Brazilian account that still has cash-cow potential — but it is also the part most exposed to the next chapter's risk.
The OECD Pillar Two and Brazil's new transfer-pricing rules (2024) hit peer operators hard: Shell paid R$437M in additional tax in 2024, Petrobras paid R$1.24B in 2024–25. Petronas has the same structural exposure for any intercompany services or cross-border licensing on Brazil-located assets. Modelled exposure: US$85–280M, not yet adjudicated. This is the live institutional question — not the dry hole.
The next chapter of the Brazilian account will not be written by G&G. It will be written by tax counsel.
The OECD Pillar Two global minimum tax (15% effective rate) is now in force in Brazil. The new Brazilian transfer-pricing rules narrowed the deductibility of intercompany royalties and services for multinational extractive operators. The peer-group adjudication (Shell: R$437M; Petrobras: R$1.24B) is the closest analogue to Petronas's likely exposure.
The exposure is structural, not drilling-driven. It applies to any pre-salt asset like Sépia where the licence is held by a Brazilian subsidiary that pays royalties, management fees, or cost-of-services to a foreign parent. Even with the stake retained, the post-tax NPV is lower than the pre-tax NPV estimated above.
The honest answer: we don't know the exact number yet. The adjudication is ongoing. The modelled range (US$85–280M) is wide because Petronas has not disclosed the intercompany service pricing arrangement for the Brazilian subsidiary.
What we know:
This is the question that will materially change the Brazilian account over the next 18 months. The dry hole is a footnote.
Petronas is not retreating from upstream. It is reallocating.
The clearest forward signal is in Suriname: Block 52 FLNG FID target 2026 represents the most promising deepwater growth outside Malaysia. The gas-to-LNG value chain matches Petronas's existing Petronas Energy Canada and LNG Canada feedstock DNA. The Suriname story is the next decade.
The Brazilian adventure is roughly closed. The capital is being recycled into:
The vitals pulse will not change — the AMEND-2026-08-03-001 60% dividend cap pacemaker still holds; the BODY constitutional override from the extraction crisis still holds. What does change is the forward growth vector: Brazil deepwater is being wound down; LNG is being wound up.
The Petronas Brazilian account is roughly break-even. The widely-reported dry hole is rounding error. The real story is the producing-asset buy that paid for itself, the pre-salt stake that earned NPV, and the upcoming transfer-pricing adjudication that will be the next chapter.
This essay is institutional context for the /propa/ page. The vitals page seals audited inputs and tripwires; this essay writes the reasoning between them. Together they form one statement: solvent, not sterile; capital-recycled, not collapsing. The arithmetic is the same.
DITEMPA BUKAN DIBERI — Forged, not given. The Brazilian account was forged over 23 years of bid rounds, dry holes, producing fields, and a quiet capital-recycling strategy. The arithmetic is the same.
By Arif Fazil Sealed 999 · 6 min read
Muhammad Arif bin Fazil
Geoscientist · Architect, arifOS · Petronas Carigali · UW–Madison '13
Penang, Malaysia
Published: 06 August 2026 · Direct Publication · /words/ context
Epistemic Tag: INT — interpretive synthesis across AI governance, institutional economics, and systems theory
Pairs with: All writings