Mozambique LNG took its final investment decision on 18 June 2019 [1] and had been under way for 22.3 months when TotalEnergies, as operator, declared force majeure on 26 April 2021 and withdrew all project personnel from the Afungi site [2]. The consortium decided to lift the force majeure on 7 November 2025, 54.4 months later, and on 29 January 2026 announced the full restart of onshore and offshore work, with the project about 40% complete and first LNG expected in 2029 [3]. The halt was 2.4 times as long as the work before it. Counting the restart from the lift, the project's months outside force majeure since sanction do not catch up with its months inside it until 12 July 2028; counting from the full restart, until 25 December 2028. For nearly the whole run to first LNG, the halt is the larger part of the project's life since sanction.
Nothing published describes the state of Area 1's subsurface record, which wells' logs are final, or which reservoir model the development plan rests on, and this piece makes no claim about any of it. What is public is the clock, and the clock is enough to frame a question that any team restarting after a stop this long has to answer before it interprets anything: what exists, in what state, and is the current version known?
The clock, to the day
The dates come from the operator's and the press's statements at the time, and the months are calendar days divided by 30.44.
| Event | Date | Months since the previous row |
|---|---|---|
| Final investment decision, two trains, 12.88 Mtpa [1] | 18 June 2019 | |
| Total completes its purchase of Anadarko's 26.5% and becomes operator [4] | 30 September 2019 | 3.4 |
| Force majeure; all personnel withdrawn from Afungi [2] | 26 April 2021 | 18.9 |
| Consortium decides to lift force majeure [3] | 7 November 2025 | 54.4 |
| Full restart announced, about 40% complete [3] | 29 January 2026 | 2.7 |
| First LNG expected [3] | 2029 |
Two details in that table matter for anyone who inherits the subsurface side. The operator changed 3.4 months after the investment decision, as part of Total's purchase of Anadarko's African assets, so operatorship had changed hands once before the halt, and the current operator had run the project for 18.9 months when it stopped. And "halted" means site work. TotalEnergies said that almost all engineering and procurement of main equipment was carried out during the force majeure [3], and by the restart more than 4,000 workers had been mobilised across onshore and offshore activities [5]. A count of months outside force majeure is a count of calendar time, and it says nothing about how much subsurface work went on inside it.
What a long stop can do to a record
A project stopped for four and a half years need not lose a single file. What it can lose is the knowledge of which file is current: which of several log versions was final, which pick was moved and why, which deliverables a contractor issued as final and which as drafts. That knowledge often lives with people more than in the archive, and on any project stopped this long a restart team should expect that some of those people have moved on, and check rather than assume it.
Whether any of that is true of Mozambique LNG is a question for the operator; we found no public statement on it. For a restart in general, it turns into checks with yes-or-no answers. Was the archive migrated during the stop, and was each file's version history carried across? Is each well held under one name, or under the names, datums and formats of every contractor that touched it? How many candidate reservoir models exist, and is it recorded which of them was the basis of design? A schedule that assumes the subsurface was settled when work stopped is assuming the answers.
An inventory counts before it interprets
The first useful output of a restart is an inventory: for each well and survey, what exists, what state it is in, and whether the current version is known. The rows below are invented, to show the shape; they describe no real project.
| Asset (invented) | Logs, digital | Logs, image only | Completion report | Picks with provenance | Current version known |
|---|---|---|---|---|---|
| W1, exploration | yes | partial | yes | 2 of 5 | no |
| W2, appraisal | yes | no | yes | 5 of 5 | yes |
| W3, development | partial | yes | draft | 0 of 5 | no |
| 3D survey | n/a | n/a | n/a | n/a | yes |
| Reservoir model | n/a | n/a | n/a | n/a | three candidates |
The last column is the one that decides the order of the work. An image-only log is a well-defined job: digitise it, and confirm it against whatever digital version exists, a problem we have written about at the scale of national data platforms. Three candidate reservoir models with no record of which was the basis of design is a different kind of gap, because it decides which of the other gaps matter.
One cell decides which wells matter
The exhibit puts the clock above and an invented inventory below: seven wells, W1 to W7, three candidate models, each built on its own subset of the wells, and four decisions that rest on whichever model was the basis of design. Toggling a well marks its final log version as unknown.
Clock: dates from public statements, months are days divided by 30.44, and "worked" is time outside force majeure, counted from the lift on 7 November 2025. Engineering and procurement continued during force majeure. Inventory: invented wells, models and decisions, to show the logic; it describes no real project.
It opens on the draft's case: three candidates, no record of the basis, and an image-only log on W3. All four decisions are in doubt, and 6 of the 7 wells have to be reconciled, because any of the three models might be the one the decisions rest on. Record that the basis was M2 and the picture changes: 4 wells need reconciling, W3 was not one of them, and its log, still unknown, reaches no decision at all. Mark W4 unknown instead and all four decisions are back in doubt, because M2 used it.
A late find behaves the same way. Let a fourth model turn up from a contractor's archive. If the basis of design is recorded, nothing changes, because the fourth model is one more candidate that the decisions do not rest on. If it is not recorded, the fourth model adds W7 to the wells that must be reconciled, and the work grows by a well that may turn out to be irrelevant.
The point is about order. The wells worth reconciling first are those the basis of design used, and until the basis is known, that set is the union of every candidate's wells. Settling which model was the basis is a records question, answerable from approvals, meeting minutes and the development plan's own references, and it is cheaper to answer before interpretation resumes than after, when every new interpretation has been built on an unstated assumption about it. We have not measured that cost on a restart and do not put a number on it here.
In the scheme we have designed for this, each inventory row is a record with a hash, and each later interpretation is stored with the inventory state it started from. When a fourth model turns up, the inventory changes and every interpretation that depended on the old state is flagged. EarthScan's platform already has the pieces underneath: a digitiser for logs that exist only as images, bulk loading of well files in mixed formats, and a knowledge graph that holds each well under every name, datum and vendor format it has carried. The inventory method on top of them is designed, and it has not been run on a restarted project.
Three projects on one basin, three record problems
The rest of the Rovuma basin did not stop, and its projects pose different questions of their records.
Mozambique LNG's is a reconciliation problem: a record that crossed an operator change 3.4 months after the investment decision and then a 54.4-month halt, inherited by a team that is restarting construction toward first LNG in 2029.
Coral Norte's is a continuity problem. Eni and its Area 4 partners took the investment decision on 2 October 2025 for a floating unit described as an identical twin of Coral Sul, with first LNG scheduled for the second quarter of 2028 [6]. Coral Sul, fed by six subsea wells, has produced since its first cargo on 13 November 2022 [6][7]. Its record has been in use the whole time; the question for its twin is whether what was learned on the first unit is carried across to the second.
Rovuma LNG's is a fold-in problem. ExxonMobil plans 18.6 Mtpa from twelve modular trains, with start-up targeted for 2031 [8]; its force majeure was lifted on 20 November 2025 [9], and its scope includes 18 wells [10]. In August ExxonMobil said it was on track to take the final investment decision before the end of the year [11], and in August and September it signed letters of intent for engineering with McDermott [8] and with Saipem and Jan De Nul [10]. As of 19 October 2026 no decision had been announced. Whenever it comes, the record problem is the one the decision starts: 18 wells' results arriving during the build, each of which has to land in an inventory that says which model it updates.
The three need one inventory format, asking the same three things of each well and model, and three different first months.
Key takeaways
- Mozambique LNG worked 22.3 months from its investment decision on 18 June 2019 to force majeure on 26 April 2021, then spent 54.4 months in force majeure until the lift on 7 November 2025: 2.4 times as long.
- Counted from the lift, its months outside force majeure since sanction catch up with its months inside on 12 July 2028; counted from the full restart of 29 January 2026, on 25 December 2028. First LNG is expected in 2029.
- Engineering and procurement of main equipment continued during the force majeure, so the halt was a halt of site work, and the clock says nothing about the state of the subsurface record.
- On the illustrative inventory, an unrecorded basis of design puts all 4 decisions in doubt and widens the wells to reconcile from 4 to 6 before any log is checked; recorded, an unknown log on a well the basis did not use reaches no decision.
- A fourth model turning up changes nothing when the basis of design is recorded, and adds a well to the reconciliation list when it is not.
- Coral Norte, Mozambique LNG and Rovuma LNG pose continuity, reconciliation and fold-in problems of their records on one basin.
Limitations
The clock is calendar arithmetic on public dates. It counts calendar time: engineering and procurement went on during the force majeure, and nothing here measures how much subsurface work did. The projection past 19 October 2026 assumes the restart continues without another stop.
Nothing in this piece is evidence about Mozambique LNG's subsurface data, its staff, its archive or its reservoir models. The passages on what a long stop can do describe what a restart team should check on any project, and assert nothing about this one.
The inventory, its seven wells, its candidate models, which wells each model used and the four decisions are invented to show the logic. On a real project decisions rest on more than one model and on wells directly, so the all-or-nothing count in the exhibit is a simplification; the ordering argument survives it, the exact counts do not.
The inventory method is designed and has not been run on a restart. We have no measured figure for how long it takes or what it saves, and the argument that the basis of design is cheaper to settle early is reasoning that we have not measured.
The Rovuma LNG decision was pending when this was written. Its dates and scope are ExxonMobil's stated targets as of September 2026.
References
- Journal of Petroleum Technology. Anadarko sanctions Mozambique LNG project. 18 June 2019, for the decision and the two trains of 12.88 Mtpa: jpt.spe.org.
- Mozambique LNG. Total declares force majeure on Mozambique LNG project. 26 April 2021, for the withdrawal of all project personnel from Afungi: mozambiquelng.co.mz.
- Hydrocarbon Processing. TotalEnergies and Mozambique announce the full restart of the $20 B Mozambique LNG project. 29 January 2026, for the lift on 7 November 2025, 40% progress, first LNG in 2029 and engineering and procurement during force majeure: hydrocarbonprocessing.com.
- Offshore. Total confirmed as new operator of Mozambique LNG. 30 September 2019: offshore-mag.com.
- World Oil. Mozambique LNG restarts onshore, offshore activities after force majeure lift. 29 January 2026, for more than 4,000 workers mobilised: worldoil.com.
- Journal of Petroleum Technology. Eni inks FID on Coral Norte FLNG for Mozambique Area 4. 8 October 2025, for the decision on 2 October, 3.55 Mtpa, first LNG in the second quarter of 2028 and Coral Sul's six subsea wells: jpt.spe.org.
- LNG Prime. Eni's Coral Sul FLNG off Mozambique sends first cargo. 13 November 2022, for 3.4 Mtpa and the project's sanction in 2017: lngprime.com.
- World Oil. ExxonMobil advances Rovuma LNG with McDermott engineering award. 7 August 2026, for the letter of intent, 18.6 Mtpa, twelve modular trains and start-up in 2031: worldoil.com.
- Journal of Petroleum Technology. ExxonMobil lifts force majeure on Mozambique megaproject. 21 November 2025: jpt.spe.org.
- ExxonMobil Mozambique. ExxonMobil Moçambique Limitada and Area 4 co-venturers select Rovuma LNG EPC contractor and service providers. 10 September 2026, for the letter of intent with Saipem and Jan De Nul and the 18 wells: corporate.exxonmobil.com.
- Green Building Africa. ExxonMobil targets final investment decision on Mozambique Rovuma LNG project before year end. 13 August 2026: greenbuildingafrica.co.za.
