The digital mirror of the Norwegian Continental Shelf
BUSINESS DEVELOPMENT · NORWEGIAN CONTINENTAL SHELF
NCS Business Development Book
The shelf's undeveloped inventory, re-solved from the ground up. Every discovery
screened against every reachable host, every break-even computed rather than quoted, every input
traceable to the source it came from — and every chart on this page live.
Reflect. Navigate. Decide.Speil is a mirror, peil is the
bearing you take off it. The shelf shown back to you on one consistent basis — and a direction
you can act on. A new edition every month.
Edition 2026-Q3 pilotPublished 26 July 2026Cases analysed 93 (76 with a break-even)Transactions 20Basis Sodir open data · AFP engine
This is a sample of the monthly edition.
Every figure below is real and computed by the same build off the same
screening revision as the subscriber edition — nothing here is illustrative.
What is short is the quantity: each table shows its first rows and then
says how many it is holding back, the seat chapter opens on Equinor Energy AS alone rather
than all 22 offered, and the CSV exports are not included.
The full edition carries all 93 screened cases, the
20-transaction record, every seat netted to its own working interests,
and the exports. Request access.
THE ARGUMENT
Most shelf research tells you what the operator plans. This tells you what the
asset is worth.
Why this holds
The conventional way to build a shelf inventory is to read each operator's impact assessment,
take the stated capital cost, and report the break-even that falls out. It is fast, it is
verifiable, and it inherits every assumption the operator made — including the concept they had
already chosen before the document was written. You end up with a well-sourced description of
other people's decisions.
This book takes the other route. Each discovery is put through a development-planning engine:
candidate concepts are generated against every host within tie-back reach plus a standalone
development, wells and templates are placed, the flowline network and subsea layout are solved,
a production profile is built and run through the Norwegian fiscal regime, and the break-even
oil price is found by bisection. What comes out is not a restatement — it is an independent
valuation, reproducible from the revision identifier in the footer.
Two things follow that matter commercially. First, when an operator's chosen concept is not
the best one available, that shows up as a gap you can quantify rather than an assumption you
inherit. Second, because the same machinery values every discovery on the same basis, the
results are comparable across the whole shelf — which is what a supply curve and a
capacity-constrained portfolio actually require.
419
MMboe clears $40/bbl
22% of screened volume, 10 of 76 cases
$77 vs $108
median break-even, researched vs generic
the gap curation closes — read both, never one
16.2 bn
NOK optimal portfolio NPV
capacity-constrained, not a sorted list
27
hosts with a queue
more discoveries want the slot than the portfolio can seat
1 · Where the shelf stands, and where the activity is
419 MMboe of the 1,910 screened (22%) clears a $40/bbl hurdle. What binds is not geology but processing capacity: 67 discoveries are queuing for a slot across 27 hosts.
76
discoveries with a solved break-even
419
MMboe under a $40 break-even
10
cases clearing the hurdle, of 76
16.2 bn
NOK optimal portfolio NPV
16
cases on researched cost anchors
The Norwegian shelf is not short of undeveloped barrels. It is short of barrels
that clear a hurdle rate, and — more acutely every year — short of the spare processing capacity
that makes reaching them cheap. Those two constraints, not the subsurface, are what the screening
in this edition keeps running into.
Why this holds
That statement is worth being precise about, because it is easy to say and usually said
loosely. The shelf's remaining discoveries are mostly small, and small discoveries have almost no
tolerance for a standalone development. Their economics live or die on whether an existing host
has room, how far away it is, and what the owner of that host will charge to let the volumes
through. All three of those are commercial variables. None of them appear in a resource report.
This is why a screening exercise built on volumes alone systematically misranks the shelf: it
measures the thing that is abundant and ignores the things that are scarce.
Three structural reads carry across the whole inventory this quarter.
Value is concentrated, and the tail gives it back
A minority of discoveries carry nearly all the positive value. Summed across the full
screened set the loss-making tail is large enough to pull the portfolio total negative. This
is why a shelf-average break-even flatters the opportunity and a supply curve does not — and
why "how many barrels" is the wrong first question.
Host ullage is the binding input
Spare capacity at existing hubs, not resource size, decides which discoveries develop
cheaply and which wait. Where several discoveries compete for one host's free capacity the
ranking between them changes depending on who books the slot first — so the value of a
licence position partly depends on a queue nobody publishes.
Paying to expand a host usually destroys value
The intuitive fix for a full hub is to fund an expansion. It repeatedly fails: the
expansion capital has to be carried by the marginal barrels it unlocks, and those are the
barrels least able to carry it. There is a threshold cost above which waiting beats paying,
and it is computable.
Where the activity is
Activity on this shelf has concentrated into a small number of areas where a host, a cluster of
discoveries and a decision window happen to coincide.
More on this
The table below ranks areas on three
independent measures — how much screened value sits there, how much drilling has actually
happened recently, and how much of the deal flow touched it. An area that scores on all three is
where the shelf is genuinely busy; one that scores only on value is a backlog rather than a
programme.
Area
Cases
MMboe
Clear $40
Value bn NOK
E&A wells
Deals
North Sea
52
959
6
13.7
72
16
Barents Sea
16
766
3
4.4
17
2
Norwegian Sea
25
201
1
1.6
19
9
North Sea carries the most screened value (13.7 bn NOK across 52 cases, of which 6 clear a $40 hurdle). Drilling activity is concentrated in the same area, so value and effort are aligned there. Deal flow touched North Sea most often (16 of the recorded transactions).
What is coming next
The near-term agenda is set less by geology than by dates. Licence expiries force
drill-or-drop decisions, host cessation dates close tie-back windows that cannot be reopened, and
first-oil years cluster where hosts have room. This is the calendar across all three of those
triggers, nearest first — the full licence-expiry register, every screened discovery on it and out
to the last date rather than only the urgent front, is in chapter 7.
Year
Trigger
Asset
What it forces
NPV MNOK
2026
Licence decision
35/12-2 (Grosbeak)
PL925 runs out — drill, commit, trade or relinquish
2,638
2026
Licence decision
34/4-15 S (Dugong)
PL882 runs out — drill, commit, trade or relinquish
754
2026
Licence decision
31/1-2 S (Røver Nord)
PL923 runs out — drill, commit, trade or relinquish
349
2026
Licence decision
35/6-3 S (Ofelia)
PL929 runs out — drill, commit, trade or relinquish
-240
2026
Licence decision
35/10-9 (Heisenberg)
PL827 S runs out — drill, commit, trade or relinquish
-744
2027
Licence decision
30/2-5 S (Atlantis)
PL878 runs out — drill, commit, trade or relinquish
-596
2027
Licence decision
6407/8-4 S (Galtvort)
PL1223 runs out — drill, commit, trade or relinquish
-823
2027
Licence decision
3/7-8 S (Trym Sør)
PL147 runs out — drill, commit, trade or relinquish
-1,760
Showing the 14 nearest of 38 dated triggers across the screened set.
Showing 8 of 14 dated triggers. The monthly edition carries the rest of this table. About this sample
How to read every number in this book.
Break-evens are post-tax, real, at the discount
rate disclosed in chapter 10, under the Norwegian petroleum tax regime including the special tax
and uplift, with gas valued against oil on an energy-equivalent basis. Each case reports the
best concept found across all hosts within fluid-dependent tie-back reach plus a standalone
development. Point estimates are P50; a P90–P10 band accompanies every case. Volumes are the
regulator's booked recoverable figures unless a research anchor is cited.
2 · The undeveloped inventory and its break-even curve
The cheap end is thin: 10 of 76 cases clear $40/bbl. Researched cases sit at a $77 median break-even against $108 for cases on generic costs — read the two apart, never blended.
This is the chapter the rest of the book exists to support. Every discovery in the
public resource register with a booked volume has been screened, sorted cheapest-first, and
plotted against cumulative volume. The result is a supply curve for the undeveloped shelf.
The screened shelf
Every discovery in this edition, positioned on the public register's own geometry, over the
minimum-economic-field-size surface: at each cell, how small a discovery could be and still pay
for a tie-back to its best reachable host. Pale ground is cheap ground. Blank sea is beyond every
host's reach, where nothing short of a standalone development works.
More on this
The surface is an oil tie-back
threshold, so it is drawn only from hosts that can actually take oil. The two onshore gas
terminals are excluded from it: they are legitimate destinations for gas and no route at all for
oil, and including them painted cheap tie-back ground across the mainland behind them. Colour the
discoveries by any of the three questions in the control above — the key beside the map restates
what the colours mean each time you change it.
Research-anchored Engine default inputs Proxy valuation $35–40/bbl hurdle band
419 MMboe clears a $40/bbl hurdle (22% of screened volume, 10 of 76 cases)
1,910 MMboe screened across 76 non-sanctioned discoveries
$77median break-even of the 16 research-anchored cases
$108median of the 60 cases on generic cost assumptions — conservative by construction
Every bar is one discovery the AFP engine screened against all reachable hosts: its width is recoverable volume, its height the post-tax break-even oil price of the winning concept. The axis stops at $100/bbl: 37 small cases break even above it and are shown faded in the off-scale band, since the exact figure stops mattering once a project cannot clear any plausible price. Colour shows what the number rests on — see the provenance appendix.
17 screened cases are not on this curve. They carry 16 MMboe between them and do not break even below $400/bbl — the ceiling the engine searches to — so there is no price to plot them at. They are mostly sub-2 MMboe accumulations whose tie-back cost cannot be carried by the volume at any price (the worst screens at -4,414 MNOK). Every one is in the CSV in the data appendix with its basis recorded, because a curve that quietly omits them would read as the whole shelf.
Read it as a procurement decision. Moving left to right, each step is the next-cheapest barrel
available to the shelf and the price it needs to be worth developing.
More on this
The band marks where
operators say their portfolio hurdles sit. Volume to the left of the crossing competes for capital
today; volume to the right needs a cost reduction, a host that does not yet have room, or a
different price environment. The curve's shape matters more than any single point on it: a steep
curve means the cheap barrels are nearly exhausted and the next tranche costs materially more,
which is exactly the position a mature shelf reaches.
Two medians, and why they are so far apart
The curve reports the research-anchored cases and the generic-cost cases
separately, and the gap between them is wide enough to look like an error. It is not. Both
medians are drawn from the same screened set; they differ in what the capital estimate rests on.
Where no public cost anchor exists the engine falls back on a parametric estimate deliberately
set conservative for a new-build, so those cases break even high by construction — a
screening floor a project must beat to be interesting, not a forecast of what the operator would
actually spend.
The bridge below walks the distance one filter at a time, so the span can be read as the
thing it is rather than as a range of opinion.
The span is a difference in what is being compared, not a disagreement about the shelf. Every case here comes from the same screened set. The single largest step is matching them on size, worth $51/bbl on its own — so most of the headline gap is a comparison between populations that do not resemble each other, while matching them on concept barely moves it. The $20/bbl still standing after the like-for-like filters is the parametric cost estimate itself, which applies a ×1.35 capital uplift, a $40/bbl screening price, a 12% hurdle, and fixed facility operating cost to any development with no public cost anchor. That is deliberate: it makes an unresearched case a floor to beat rather than a forecast to trust. Each bar is a median of a different, smaller population, so the steps do not add up to the gap and are not meant to — separating the cost levers would need the screen re-run with each one switched off.
The bridge as numbers
Population
Cases
Median $/bbl
All 60 generic-cost cases
60
$108
… of comparable size (7–468 MMboe)
35
$58
… and a tie-back, not a standalone build
34
$58
The 16 research-anchored cases
16
$77
Why the book refuses to publish a single blended median.
A single shelf-wide median would be dominated by the
generic tail and would mislead in both directions at once: too pessimistic about the
un-researched assets, and too flattering to the researched ones by dragging them up into an
average. Coverage is therefore the product. Each edition moves discoveries from the generic
group into the anchored one by researching their public cost basis, and the supply curve is
where that work becomes visible.
Is that number normal? The shelf's own track record
A break-even is an assertion until it is placed against something. So here is the whole
sanctioning history of the Norwegian shelf on the same axes as our forward book: for every field
sanctioned since 1990, what it actually needed to work, computed from the regulator's own published
investment and production series — then, to the right of the rule, our screened inventory placed at
the year its licence forces a decision.
The shelf's own track record is in the edition. Every field sanctioned since 1990 at what it actually needed to work, computed from the regulator's published investment and production series, with the screened book placed against it at the year each licence forces a decision. About this sample
Realised, 84 sanctioned fields Screened, research-anchored Screened, engine default inputs▲ tie-back · ■ standalone · marker area is life volume
$31median realised break-even across 84 fields sanctioned since 1990
$44median for the 30 sanctioned from 2010 onwards — the shelf got more expensive, not less
$77median of the 16 screened cases with a researched cost anchor — the like-for-like comparison
$103median of the 57 on generic cost assumptions — a screening floor, not a forecast
Left of the rule: what 84 NCS fields actually needed, computed from Sodir's own investment and production series on the same tax, discount rate and $14/boe operating cost as the screened cases — not the break-even anyone quoted at PDO, because Sodir's investment series runs the whole field life and carries every later infill campaign with it. Right of the rule: our screened inventory, placed at the year its licence forces a decision. Compare like with like. The 16 screened cases with a researched cost anchor sit at $77, against $44 for what the shelf has actually delivered since 2010 — a real gap, and the first thing to argue about. The other 57 cases carry the engine's parametric CAPEX and median $103; that number is a floor each project must beat to be interesting, not a claim about what it would cost, and pooling it with the anchored cases would misstate both. 27 case(s) break even above the $120 ceiling; they are drawn hollow in the hatched lane above the axis, not at the top of it, and every one is named with its true figure in the table below. 3 field(s) whose remaining life is mostly forecast rather than reported are excluded. Break-even is quoted per boe: a gas-weighted field's figure is an oil-equivalent, not an oil price.
The second price line is not a real oil price. It restates Brent against the cost base the shelf actually spends in: converted to kroner at that year's Norges Bank rate, brought to 2025 on the SSB consumer index, and returned to dollars at today's rate. It therefore moves with the krone as well as with inflation, which is the point — a 2008 barrel bought materially less Norwegian development than its dollar figure suggests, and the cost outcomes of that vintage show it. Consumer prices are a proxy for offshore cost inflation, not a measure of it. The optional band on the 5 largest forward cases is the resource band the screen already sweeps — the break-even at 1.50× and 0.60× the booked volume, taken as the inverse scaling that follows from revenue moving with volume while capital does not. Capital is held at its point estimate, so this is not a full uncertainty range: on a researched case the cost side is usually the larger question, and on an unresearched one the parametric estimate dominates everything else.
The 27 case(s) above the $120 ceiling, with the break-even the lane does not show
Field
Side
Year
Break-even $/boe
MMboe
Concept
24/6-1 (Peik)
Screened
2043
$393
1.5
Tie-back
6406/12-G-1 H (Fenja Nord)
Screened
2039
$365
3.7
Tie-back
35/8-3 (Aurora)
Screened
2030
$329
4.2
Tie-back
30/5-3 S (Corvus)
Screened
2031
$323
2.7
Tie-back
7122/7-8 (Goliat Nord)
Screened
2042
$295
1.9
Tie-back
24/9-10 S (Caterpillar)
Screened
2032
$286
2.1
Tie-back
2/4-17 Tjalve
Screened
2048
$278
2.6
Tie-back
35/11-31 S (F-Sør)
Screened
2040
$259
2.6
Tie-back
25/1-14 G (Sigma)
Screened
2030
$259
4.2
Tie-back
31/1-4 (Ringand)
Screened
2028
$258
4.4
Tie-back
15/3-4 (Sigrun)
Screened
2032
$253
2.6
Tie-back
6506/9-2 S (Fogelberg)
Screened
2028
$251
4.2
Tie-back
7120/1-3 (Gohta)
Screened
2043
$240
26.3
Standalone
24/9-15 S (Froskelår Nordøst)
Screened
2032
$232
2.2
Tie-back
6406/11-2 S (Vidsyn)
Screened
2039
$223
7.8
Tie-back
34/12-1 (Afrodite)
Screened
2039
$216
7.5
Tie-back
6407/1-7 (Solberg)
Screened
2028
$205
4.8
Tie-back
6407/1-6 S (Rodriguez)
Screened
2028
$204
4.8
Tie-back
6407/1-8 S (Sierra)
Screened
2028
$197
4.8
Tie-back
6407/7-8 (Noatun)
Screened
2034
$167
7.8
Tie-back
6406/3-10 A (Åre)
Screened
2051
$162
6.9
Tie-back
Hanz
Realised
2018
$138
5.7
Tie-back
25/2-4 Lille-Frigg
Screened
2052
$137
13.2
Tie-back
6507/4-2 S (Adriana Sabina)
Screened
2032
$130
7.6
Tie-back
35/11-24 S (Swisher)
Screened
2035
$129
7.3
Tie-back
15/3-12 S (Sigrun Øst)
Screened
2032
$128
6.0
Tie-back
6507/8-9 (Carmen)
Screened
2045
$127
4.6
Tie-back
Realised break-even by field (84 fields, 14 $/boe opex, 8% discount to sanction year)
Field
Sanctioned
First oil
Concept
Life MMboe
Reported
CAPEX bn NOK
Gas share
Break-even $/boe
Pre-tax
Tyrving
2022
2024
Tie-back
23
60%
5.4
3%
$47.0
$44.2
Duva
2019
2021
Tie-back
48
86%
5.7
41%
$31.0
$29.6
Ærfugl Nord
2019
2021
Tie-back
14
100%
1.8
84%
$30.4
$28.7
Fenja
2018
2023
Tie-back
21
87%
10.9
5%
$92.7
$85.9
Hanz
2018
2024
Tie-back
6
93%
5.4
37%
$137.5
$121.9
Nova
2018
2022
Tie-back
64
50%
12.5
12%
$47.5
$44.6
Skogul
2018
2020
Tie-back
12
95%
2.2
8%
$39.6
$37.4
Bauge
2017
2023
Tie-back
9
75%
3.5
15%
$79.1
$73.5
Byrding
2017
2017
Tie-back
5
100%
1.0
0%
$40.5
$39.3
Oda
2017
2019
Tie-back
31
98%
5.2
0%
$39.0
$36.8
Trestakk
2017
2019
Tie-back
41
89%
6.6
0%
$38.8
$36.6
Dvalin
2016
2020
Tie-back
63
63%
15.6
96%
$57.0
$53.3
Utgard
2016
2019
Tie-back
17
92%
2.4
45%
$33.4
$31.7
Johan Sverdrup
2015
2019
Standalone
1,882
79%
130.1
3%
$25.7
$24.7
Maria
2015
2018
Tie-back
63
80%
17.8
6%
$62.0
$57.9
Aasta Hansteen
2013
2018
Standalone
405
92%
36.8
98%
$31.5
$30.0
Gina Krog
2013
2017
Standalone
173
91%
34.3
58%
$49.4
$46.3
Ivar Aasen
2013
2016
Standalone
155
96%
32.1
15%
$49.1
$46.1
Bøyla
2012
2015
Tie-back
38
76%
7.3
8%
$49.7
$47.6
Edvard Grieg
2012
2015
Standalone
322
97%
31.9
8%
$30.8
$29.4
Martin Linge
2012
2021
Standalone
115
89%
58.7
56%
$105.6
$97.7
Svalin
2012
2014
Tie-back
60
98%
6.2
0%
$30.7
$29.3
Atla
2011
2012
Tie-back
11
100%
1.4
84%
$38.2
$36.5
Hyme
2011
2013
Tie-back
22
96%
4.5
12%
$56.0
$52.5
Knarr
2011
2015
Standalone
72
100%
14.2
4%
$47.8
$44.9
Skuld
2011
2013
Tie-back
33
100%
11.3
7%
$75.3
$70.1
Valemon
2011
2015
Standalone
114
93%
26.9
87%
$55.0
$51.4
Visund Sør
2011
2012
Tie-back
72
88%
6.4
57%
$30.7
$29.2
Gudrun
2010
2014
Standalone
263
97%
27.7
38%
$32.7
$31.1
Marulk
2010
2012
Tie-back
71
99%
4.4
81%
$26.1
$25.1
Goliat
2009
2016
Standalone
145
96%
57.0
0%
$89.4
$82.9
Trym
2009
2011
Tie-back
39
92%
3.0
70%
$31.7
$30.2
Alve
2007
2009
Tie-back
116
84%
5.3
64%
$24.7
$23.8
Gjøa
2007
2010
Standalone
513
98%
33.8
54%
$29.9
$28.5
Morvin
2007
2010
Tie-back
101
100%
8.0
28%
$33.0
$31.6
Skarv
2007
2013
Standalone
625
94%
61.4
64%
$36.3
$34.3
Vega
2007
2010
Tie-back
273
98%
13.2
50%
$24.4
$23.5
Volund
2007
2009
Tie-back
86
99%
6.0
12%
$27.8
$26.6
Yttergryta
2007
2009
Tie-back
18
100%
1.4
65%
$32.4
$31.1
Rev
2006
2009
Tie-back
23
100%
4.4
75%
$56.1
$52.4
Blane
2005
2007
Tie-back
5
98%
0.6
0%
$39.0
$38.0
Ringhorne Øst
2005
2006
Tie-back
94
92%
1.6
2%
$16.7
$16.6
Vilje
2005
2008
Tie-back
92
99%
3.1
4%
$22.0
$21.3
Volve
2005
2008
Standalone
72
100%
4.7
7%
$27.0
$25.9
Alvheim
2004
2008
Standalone
432
96%
41.2
15%
$32.1
$30.5
Gimle
2004
2005
Tie-back
32
92%
2.6
23%
$26.1
$25.6
Ormen Lange
2004
2007
Tie-back
1,906
97%
63.5
94%
$20.9
$20.3
Tyrihans
2004
2009
Tie-back
505
96%
22.7
44%
$24.7
$23.9
Urd
2004
2005
Tie-back
55
98%
9.9
3%
$45.5
$42.7
Snøhvit
2003
2007
Tie-back
751
88%
25.7
81%
$20.8
$20.3
Kristin
2002
2005
Standalone
488
88%
35.1
45%
$31.3
$29.8
Skirne
2002
2004
Tie-back
83
100%
2.7
84%
$22.1
$21.4
Fram
2001
2003
Tie-back
384
98%
15.1
23%
$22.6
$22.0
Mikkel
2001
2003
Tie-back
349
98%
8.2
57%
$18.1
$17.7
Sigyn
2001
2003
Tie-back
126
100%
3.5
38%
$19.0
$18.7
Glitne
2000
2001
Standalone
56
100%
3.0
0%
$21.5
$21.1
Kvitebjørn
2000
2004
Standalone
922
99%
25.6
70%
$20.3
$19.7
Tambar
2000
2001
Standalone
84
97%
5.0
0%
$21.9
$21.3
Tune
2000
2002
Tie-back
153
100%
4.7
84%
$20.5
$20.1
Sygna
1999
2000
Tie-back
72
99%
2.5
0%
$19.5
$19.1
The comparison is the point. If our screened cases cluster where the shelf has historically
delivered, the estimates are ordinary and the argument moves on to which ones to chase. If they sit
materially above it, either the remaining inventory really is harder than what came before — which
is what a mature shelf looks like — or our cost assumptions are too conservative. Both readings are
actionable; neither is available from a supply curve alone.
What the realised numbers are, and are not.
These are not the break-evens operators
quoted at the time of PDO. Those figures are somebody's compiled analysis and are not ours to
republish. What is ours is the arithmetic: the regulator publishes annual investment and annual
production per field, and reserves booked against each, which is enough to ask what flat realised
price would have zeroed the whole-life cash flow discounted back to sanction.
Because the investment series runs the entire life of a field, it carries every infill campaign
and redevelopment that followed first oil. So a realised figure here is a whole-life outcome
plotted at its sanction vintage, not a snapshot of what was believed at sanction. The two diverge
most for the long-lived fields that have been redeveloped twice — which is worth knowing, and is
exactly the risk a forward estimate cannot see in itself.
Three assumptions carry the result and none of them are observable: unit operating cost, which is
not published per field; the decline extrapolation used to complete a producing field's remaining
life; and the exchange rate applied to convert published krone investment into a dollar price.
Fields whose remaining life is mostly extrapolation rather than reported production are excluded
rather than shown faint. The table view carries every input.
Break-even against return
Two projects can share a break-even and deserve very different amounts of
management attention. Break-even says whether a project survives a low price; return on
capital says whether it is worth doing at all. The shaded corner is where both answers are
yes.
Bubble area is recoverable volume; the shaded corner is sub-$40 break-even with an IRR above 15%. Darker bubbles are research-anchored. 12 of 73 cases are named — the research-anchored ones and then the largest, as far as the space allows without labels overlapping; the rest carry their figures on hover and all of them are in the inventory table.
Capital intensity benchmark
Capital against volume on log-log axes, with a fitted shelf trend. A project
sitting well above the line carries a cost premium for its size — sometimes justified by water
depth or fluid, often the first sign that a concept is wrong for the asset.
CAPEX (bn NOK, log) against recoverable volume (log); dashed line is the fitted shelf trend (exponent 0.07). Points well above it carry a cost premium for their size.
Are these costs normal for this shelf?
The benchmark above is internal: it fits a trend through our own screened cases and reads each
one against the others. That is the right way to spot the odd case out, and the wrong way to spot a
cohort that is uniformly optimistic — if every estimate in this book were a third too light, the
fitted line would move with them and every point would still sit comfortably on it.
So here is the same question asked from outside. The regulator publishes annual investment per
field; deflate it by the consumer price index and set it against the volume each field was booked
at, and you have what the shelf has actually charged to develop a barrel, over three decades and
every concept it has tried. The screened cases go on top, unadjusted.
The sample stops here.
You are part-way through chapter 2 of 10,
The undeveloped inventory and its break-even curve. The rest of it — starting with the answer to the
question just asked — and the 8 chapters that follow are in the
edition, and are shown here only in outline:
The host atlas — click any hub and read its book
Top candidates — cash flows, capital and concept choice
Optimal portfolio under capacity constraints
Transactions, corporate positions and who is moving
Drilling inventory, results and licence deadlines
The seat view — assets, net value and minimum economic field size
How to improve value — the levers, ranked
Economic basis, assumptions and provenance
Every one of them is computed on the same basis as what you have just read,
off the same screening revision. Request access.
Subsea tie-backStandalone developmentAFP estimate11 point(s) off scale, drawn hollow at the frame edge
Realised development capital against recoverable volume for the 106 NCS fields that started production since 1990 — Sodir investment reporting, deflated to 2025 kroner by SSB consumer prices. Red diamonds are the 6 largest screened cases in this edition, on the same axes. The frame is cut at the cohort's 90th percentile so the undeveloped end is legible; the handful of giants beyond it, Johan Sverdrup among them, are pinned hollow to the edge and counted under the chart rather than dropped. The shelf has spent a median of 89 NOK/boe, but that rate does not transfer to a screening case unaltered: volume explains only 6% of the variation in what a development cost (R²=0.06), and the screened book is an order of magnitude smaller per case than the cohort it is drawn against. Compared like for like — each case against realised fields within a factor of 3 of its own volume, which 78 of the 93 screened cases have enough peers for — the median screened case carries 1.64× the capital per barrel that the shelf actually spent at that size, and only 12 of 78 come in cheaper than their peers. The screening basis is therefore conservative against the realised record, not optimistic, which is the direction an inventory this early ought to err in.
Benchmark data — realised capex, volume and unit cost
Field
Start
Capex BNOK
MMboe
NOK/boe
Type
Wisting
—
105.9
468
226
AFP estimate
Peon
—
6.1
172
35
AFP estimate
Grosbeak
—
3.0
87
34
AFP estimate
Alta
—
5.9
79
75
AFP estimate
Garantiana
—
5.4
56
95
AFP estimate
Sigma NE
—
7.6
48
159
AFP estimate
Johan Sverdrup
2019
126.3
1,856
68
standalone
Johan Castberg
2025
83.7
559
150
standalone
Åsgard
1999
81.9
2,440
34
standalone
Martin Linge
2021
74.9
189
396
standalone
Goliat
2016
70.0
242
289
standalone
Skarv
2013
69.8
429
163
standalone
Heidrun
1995
56.7
1,391
41
standalone
Aasta Hansteen
2018
48.7
290
168
standalone
Sleipner Øst
1993
48.1
734
66
standalone
Snorre
1992
47.7
1,622
29
standalone
Gjøa
2010
43.4
343
127
standalone
Gina Krog
2017
39.3
225
175
standalone
Kristin
2005
34.4
478
72
standalone
Ivar Aasen
2016
34.3
157
218
standalone
Ormen Lange
2007
32.2
2,499
13
tie-back
Edvard Grieg
2015
30.7
183
167
standalone
Valemon
2015
29.9
211
142
standalone
Draugen
1993
29.0
863
34
standalone
Gudrun
2014
28.9
123
236
standalone
Sleipner Vest
1996
24.7
953
26
standalone
Alvheim
2008
24.5
184
133
standalone
Breidablikk
2023
21.5
192
112
tie-back
Tyrihans
2009
21.1
0
—
tie-back
Visund
1999
20.0
582
34
standalone
Gullfaks Sør
1998
19.2
575
33
tie-back
Knarr
2015
19.2
59
327
standalone
Brage
1993
19.1
333
57
standalone
Norne
1997
17.2
673
26
standalone
Veslefrikk
1990
16.8
382
44
standalone
Oseberg Sør
2000
16.5
418
40
standalone
Maria
2018
16.5
197
84
tie-back
Gyda
1990
15.8
292
54
standalone
Kvitebjørn
2004
15.7
465
34
standalone
Jotun
1999
15.0
165
91
standalone
Skuld
2013
13.9
92
151
tie-back
Nova
2022
13.6
79
172
tie-back
Snøhvit
2007
13.3
1,181
11
tie-back
Fenja
2023
13.3
97
138
tie-back
Vega
2010
13.0
76
171
tie-back
Njord
1997
12.8
232
55
standalone
Tommeliten A
2023
12.8
134
95
tie-back
Morvin
2010
12.2
87
141
tie-back
Dvalin
2020
12.1
118
102
tie-back
Hugin Satellitter
1995
11.6
0
—
standalone
Brynhild
2014
11.5
20
573
tie-back
Troll
1990
11.4
10,135
1
standalone
Huldra
2001
11.3
109
104
standalone
Oseberg Øst
1999
8.8
194
45
standalone
Fulla
1994
8.4
76
111
tie-back
Trestakk
2019
8.3
74
112
tie-back
Statfjord Nord
1995
8.0
274
29
tie-back
Vigdis
1997
8.0
319
25
tie-back
Varg
1998
7.8
76
102
standalone
Bøyla
2015
7.7
23
328
tie-back
Solveig
2021
7.6
58
131
tie-back
Tordis
1994
7.5
418
18
tie-back
Statfjord Øst
1994
7.0
266
26
tie-back
Duva
2021
6.8
88
76
tie-back
Svalin
2014
6.7
76
88
tie-back
Oselvar
2012
6.6
54
122
tie-back
Hyme
2013
6.5
25
256
tie-back
Halten Øst
2025
6.2
96
65
tie-back
Fram
2003
6.2
98
63
tie-back
Urd
2005
6.1
66
93
tie-back
Tune
2002
6.1
124
49
tie-back
Rev
2009
6.1
41
150
tie-back
Oda
2019
5.9
47
125
tie-back
Hanz
2024
5.9
19
306
tie-back
Tyrving
2024
5.7
27
210
tie-back
Visund Sør
2012
5.6
98
57
tie-back
Alve
2009
5.5
58
94
tie-back
Jette
2013
5.2
12
418
tie-back
Marulk
2012
5.2
77
67
tie-back
Volund
2009
4.9
51
96
tie-back
Volve
2008
4.7
86
55
standalone
Trym
2011
4.4
36
122
tie-back
Embla
1993
4.4
115
38
standalone
Bauge
2023
4.2
73
58
tie-back
Skirne
2004
3.7
52
72
tie-back
Fram H-nord
2014
3.6
11
345
tie-back
Yme
1996
3.6
50
72
standalone
Sigyn
2003
3.6
103
35
tie-back
Gaupe
2012
3.4
32
107
tie-back
Symra
2026
3.2
60
54
tie-back
Vilje
2008
3.2
52
60
tie-back
Mikkel
2003
3.1
265
12
tie-back
Sygna
2000
3.0
69
43
tie-back
Skogul
2020
2.8
9
300
tie-back
Vale
2002
2.5
27
96
tie-back
Verdande
2025
2.4
36
68
tie-back
Tambar
2001
2.4
55
44
standalone
Yttergryta
2009
2.3
14
164
tie-back
Ærfugl Nord
2021
2.1
28
73
tie-back
Utgard
2019
2.1
28
75
tie-back
Atla
2012
2.1
11
190
tie-back
Hod
1990
2.0
64
31
standalone
Glitne
2001
1.6
46
34
standalone
Byrding
2017
1.3
9
142
tie-back
Blane
2007
0.9
5
175
tie-back
Islay
2012
0.9
1
931
tie-back
Ringhorne Øst
2006
0.8
52
16
tie-back
Flyndre
2017
0.8
3
279
tie-back
Mime
1990
0.8
3
239
tie-back
Gungne
1996
0.7
97
7
tie-back
Gimle
2005
0.4
32
11
tie-back
Enoch
2007
0.3
3
135
standalone
Why this is read size-matched, and not off the trend line
Two obvious ways to summarise this chart both give a confident answer that is wrong, so neither is
used. Reading each case against the fitted line is the first: a straight capex-against-volume fit
over the realised cohort explains about six per cent of the variance, because what a development
cost on this shelf is driven by water depth, concept and vintage far more than by how big it was.
Almost anything falls below a line that flat, and nothing is learned from the fact that it does.
Reading them against the raw cost-per-barrel distribution is the second, and it is the more
dangerous of the two because it looks robust. The undeveloped inventory is an order of magnitude
smaller per case than the fields that have already been built — a median of single-digit millions
of barrels against roughly ninety. A small development spreads the same fixed cost of a template,
an umbilical and a control system over far fewer barrels, so on cost per barrel the screened book
looks several times more expensive than the shelf purely because of what size it is. That is an
artefact of the comparison, not a finding about the estimates.
What is left after both traps is the like-for-like read: each case against the realised fields
closest to it in size. That comparison is the one quoted under the chart, and it is also the only
one that can be argued with on its own terms — the peer set is public, the band is stated, and the
cases with too few peers to compare are counted rather than quietly averaged in.
Volume, capital and break-even in one view
The same inventory ordered by size rather than by cost, with the three numbers that decide a
project held on one row. Big and cheap is the combination worth acting on, and it is only visible
when volume, capital and break-even line up against each other.
Research-anchoredEngine default inputsProxy valuationCAPEX and break-even panels — basis
1.64 bn boe across the 34 largest cases drawn here (85% of the 1,926 MMboe screened)
241 bn NOK over the 34 of 34 rows carrying a CAPEX estimate, of which 7 rest on a researched cost anchor — not a shelf total
$47median break-even of the 7 research-anchored rows
$54median of the 27 rows on generic cost inputs — a screening floor, not a forecast
One row per discovery, ordered by recoverable volume and held in that order across all three panels, so a project can be read straight across. Colour carries the operator on the left and the input basis on the right — the two questions that decide whether a number is a fact or a floor. The 59 smaller cases not drawn hold 285 MMboe between them and are listed in full in the inventory table. Bars past their axis are drawn faded — 1 on resources (axis stops at 181); 1 on capex (axis stops at 15); 3 on break-even (axis stops at 100) — and the value column still prints the real figure. An axis is only cut where one project runs more than twice the next.
Table view — inventory panel
Project
Operator
MMboe
CAPEX bn NOK
Break-even $/bbl
Basis
Wisting
Equinor Energy
467.9
105.9
88.4
Research
Peon
Equinor Energy
172.0
6.1
41.1
Proxy
Grosbeak
Equinor Energy
87.3
3.0
25.6
Research
Alta
Aker BP
79.2
5.9
38.7
Engine
Garantiana
Equinor Energy
56.4
5.4
41.2
Engine
Sigma NE
Aker BP
47.7
7.6
54.4
Engine
Kjøttkake
Aker BP
44.2
2.9
32.4
Engine
Omega
Aker BP
37.0
5.7
54.3
Engine
Slagugle
ConocoPhillips Skan…
35.8
3.2
34.8
Engine
Zagato Sør
Vår Energi
35.3
2.5
31.3
Engine
Kayak
Equinor Energy
34.6
2.8
40.5
Engine
Ringhorne Nord
Vår Energi
33.0
2.8
36.2
Engine
Mistral Sør
Equinor Energy
32.4
4.4
55.6
Engine
Dugong
Vår Energi
32.4
2.7
39.3
Engine
Elgol
Vår Energi
31.9
2.9
36.7
Engine
Bergknapp
Harbour Energy
30.6
3.3
46.1
Engine
Røver Nord
Equinor Energy
30.1
3.2
46.8
Research
Kveikje
Equinor Energy
27.0
3.1
48.2
Research
Gohta
Aker BP
26.3
14.3
240.3
Engine
Norma
DNO
26.2
3.9
58.8
Engine
Carmen
Wellesley Petroleum
25.8
4.2
64.0
Engine
Othello
DNO
25.2
4.7
55.6
Engine
Bestla
—
23.8
2.8
19.0
Research
Busta
Equinor Energy
23.1
3.4
59.0
Engine
Neiden
Aker BP
22.0
3.3
61.1
Engine
Toppand
Equinor Energy
19.2
3.3
66.5
Research
Frøy
Aker BP
18.5
3.5
57.6
Engine
Countach S3
Vår Energi
18.4
3.1
55.2
Engine
Ragnfrid
Equinor Energy
17.6
5.4
105.8
Engine
Beta
Vår Energi
17.4
3.5
66.1
Engine
7220/7-CD-1 H
Equinor Energy
16.9
2.0
50.1
Engine
Skruis
Equinor Energy
16.7
2.4
57.5
Engine
Cerisa
Vår Energi
16.1
2.5
22.1
Research
Lille-Frigg
Aker BP
13.2
5.5
136.8
Engine
The inventory
Sorted by break-even. The basis column is the one to read alongside every number: it says
whether the case rests on a researched cost anchor, on the engine's own parametric estimate, or on
the proxy valuation used when a full layout solve could not close.
#
Discovery
Area
Winning host
MMboe
CAPEX bn NOK
Break-even
IRR
First oil
Basis
1
31/7-1 Bestla
North Sea
23.8
2.8
$19.0
22.0%
2029
Research
2
36/7-5 S (Cerisa)
North Sea
16.1
2.5
$22.1
17.7%
2029
Research
3
35/12-2 (Grosbeak)
North Sea
87.3
3.0
$25.6
22.7%
2029
Research
4
7122/8-3 S (Zagato Sør)
Barents Sea
35.3
2.5
$31.3
18.0%
2029
Engine
5
35/10-15 S (Kjøttkake)
North Sea
44.2
2.9
$32.4
17.7%
2029
Engine
6
6507/5-10 S (Slagugle)
Norwegian Sea
35.8
3.2
$34.8
17.2%
2029
Engine
7
25/8-23 S (Ringhorne Nord)
North Sea
33.0
2.8
$36.2
15.2%
2029
Engine
8
7122/9-2 (Elgol)
Barents Sea
31.9
2.9
$36.7
14.8%
2029
Engine
Showing the 40 lowest break-evens of 93 screened cases; the full table ships in the data appendix. 2 cases carry a gas-basis break-even: the liquids are a by-product, so the figure is per barrel of oil equivalent. A further 17 screened cases do not break even below $400/bbl; they carry no price to sort by and are listed in the CSV appendix rather than here.
Showing 8 of 40 screened cases. The monthly edition carries the rest of this table. About this sample
3 · The host atlas — click any hub and read its book
31 of 46 hosts have curated spare capacity, and 27 of them have a queue. Whoever controls a slot at the right hub controls which neighbours can develop at all.
On a shelf this mature, a host is not infrastructure — it is a commercial position.
Whoever controls spare capacity at the right hub controls which nearby discoveries can be
developed, on what timetable, and at what tariff. The atlas is built to make that position
legible.
Click any host. You get its remaining spare oil and gas capacity and the tariff basis we
screened against, its actual throughput history from the public register — a declining curve is
precisely the ullage that makes a tie-back possible — the discoveries whose best concept is that
host, which of them the capacity-constrained portfolio actually selects, and any recorded
transaction touching the asset.
More on this
Where the number of contenders exceeds the number of slots, that
host has a queue, and the difference between being first and third in it is the whole value of a
licence position.
The capacity numbers are the softest in this book, deliberately so.
Spare capacity is a
negotiated quantity, not a published one. What is shown is screening-grade guidance compiled from
public development plans, operator disclosure and observed production decline. A licence holder
with access to the host's actual commercial terms should replace these first — they are the
single input most likely to change a ranking, which is also why we show the basis note on every
host rather than burying it.
Every host on one page
The same capacity, tariff basis and cessation dates the atlas shows one hub at a time, tabulated
across the whole shelf. These are the assumptions every tie-back in this book is screened against,
and the assignment in chapter 5 runs against this table.
Host
Spare oil kbbl/d
Spare gas MSm³/d
Tariff $/bbl
Ullage opens
Cessation
Basis
30
2.0
$8.0
2025
2035 ⚠
Central Aker BP tie-back hub (Vilje/Volund/Bøyla) public PDO tariffs (screening-grade)
20
0.5
$10.0
2027
2045 ⚠
Older FPSO redeveloped (Balder Future); extended life (screening-grade)
38
2.0
$9.0
2027
2032 ⚠
Fixed jacket (OKEA op); nameplate oil 120 / gas 85 mmcfd / water 115 / 40 slots; free-oil constrained by mature-facility liquid/water handling not nameplate; screening tariff arm's-length (Bestla actual = NKr5/bbl proc intra-OKEA); WoodMac Brage Asset Rpt Sep-2023 + GEM
30
0.5
$8.0
2025
2040 ⚠
Mature FPSO, significant spare oil as base declines; lifetime extension to ~2040 (screening-grade)
35
1.0
$7.0
2025
2038
Spare ullage as field declines (Solveig/Rolvsnes PDO) (screening-grade)
40
3.0
$9.5
2026
2036 ⚠
Gjøa SEMI host (Vega/Duva/Nova/Cara) — public PDO/tariff guidance (screening-grade)
Mature platform with declining base (screening-grade)
Screening-grade capacity and tariff guidance compiled from public plans for development and operation, operator disclosures and field production history. Each row is an assumption a licence holder should replace with its own commercial terms. 27 cessation date(s) marked ⚠ are contested — the curated year shown disagrees by more than two years with a projection from that host's own production decline. Both candidates are retained; neither has been adopted. Because a cessation year truncates a tie-back's producing life outright, this is the single assumption on this page most able to move a queued case from clearing to failing. The flag is unfiltered: it fires on any material difference, and the projection is an oil-rate exhaustion date rather than a facility cessation, so most marks are a property of the method and not evidence the curated year is wrong. Three cases dominate — the projection hit its 25-year modelling horizon without reaching the cease floor; the host's oil is already below that floor; or the hub is carried by gas, where an oil floor says nothing about cessation. Read ⚠ as "this date has not been reconciled", not as "this date is wrong".
Showing 8 of 31 hosts. The monthly edition carries the rest of this table. About this sample
4 · Top candidates — cash flows, capital and concept choice
12 cases carry a full stored cash flow. For each, the gap between the winning concept and the runner-up is usually worth more than tightening the volume estimate — host choice is the lever.
A break-even is a summary statistic, and summary statistics hide the things that
kill projects. This chapter opens the top candidates up: the year-by-year cash flow the valuation
actually rests on, where the capital goes, which concepts the engine compared and by how much it
preferred the winner, and the provenance of every input that moved the answer.
Each card below is one candidate, with the shape of its cumulative discounted cash flow across
field life.
More on this
Read the line, not the endpoint. Every card dips first —
that is the capital phase, and the depth of the dip is money at risk before anything comes back —
then climbs as production pays it off. Where it crosses back through zero is payback. Two projects
with the same break-even can have very different dips, and the deep one needs a balance sheet the
shallow one does not.
What you are looking for is not the highest ending value but the shallowest dip for the value it
returns: that is capital efficiency, and it is what survives a downturn.
Cash flow, capital and concepts, case by case
Pick a candidate. The bars split each year into what pays in and what pays out — revenue
against capital, operating cost, host tariff, carbon and tax — and the line is cumulative
discounted value, so where it crosses zero is payback and where it ends is net present value.
More on this
A
long flat start followed by a steep climb is a tie-back; a deep trough is a new-build carrying its
own facilities.
Why the concept comparison table matters more than it looks.
The gap between the selected
concept and the runner-up is the value of getting the host choice right, and it is frequently
larger than the uncertainty band on the volume. That has a direct commercial implication: effort
spent negotiating access to the better host usually beats effort spent appraising the reservoir
more precisely. It is the cheapest value uplift available on a mature shelf, and it is
commercial work, not technical work.
5 · Optimal portfolio under capacity constraints
Solved jointly under capacity limits, the shelf supports 24 developments worth 16.2 bn NOK — not the same set a sorted list would pick, because hosts are shared.
Why this holds
Ranking projects one at a time answers the wrong question when they share infrastructure. Two
discoveries can each look attractive against the same host when only one can have the capacity,
and a list sorted by net present value will happily recommend both. The portfolio below is the
answer to the joint problem: assign discoveries to hosts so as to maximise total value subject to
each host's free oil and gas capacity, solved as a mixed-integer programme rather than picked off
a sorted list. The capacity and tariff assumptions it runs against are tabulated host by host in
chapter 3.
The difference between this and the sorted list is the cost of contention, and it is a real
number. Where it is large, the shelf is capacity-rationed and the commercial priority is securing
access. Where it is small, capacity is not binding and the priority is cost.
16.2 bn
NOK portfolio NPV
24
developments selected
87 bn
NOK capital
13
hosts carrying load
Discovery
Assigned host
MMboe
Oil kbbl/d
Gas MSm³/d
NPV MNOK
Break-even
P90 NPV MNOK
35/12-2 (Grosbeak)
87.3
12.3
0.20
2,638
$25.6
1,097
7220/11-1 (Alta)
79.2
17.4
0.28
1,646
$38.7
-370
31/7-1 Bestla
23.8
11.8
0.51
1,308
$19.0
111
35/10-15 S (Kjøttkake)
44.2
8.8
0.14
1,244
$32.4
66
34/6-2 S (Garantiana)
56.4
15.3
0.25
1,241
$41.2
-429
6507/5-10 S (Slagugle)
35.8
10.1
0.16
1,073
$34.8
-19
7122/8-3 S (Zagato Sør)
35.3
7.7
0.13
1,066
$31.3
56
25/8-23 S (Ringhorne Nord)
33.0
6.8
0.11
812
$36.2
-124
Showing 8 of 24 developments. The monthly edition carries the rest of this table. About this sample
Break-even is the price test; the P90 column is the separate volume-and-cost test (P90 ×0.60 on volume, ×1.50 on capital). A development can pass one and fail the other, so both are printed rather than blended into a single robustness grade — on this revision 20 of the 24 seated developments are net-present-value positive in the base case and negative in the P90 case.
Where the next area development is
The assignment above is a snapshot of one constraint set. The question that follows it is where
the constraint is worth attacking — which hubs have a queue behind them large enough that solving
the cluster together beats applying for a slot one discovery at a time.
Why this holds
An area development is not a bigger tie-back. It is a different
commercial object: several licences, usually several operators, agreeing a shared route and a
shared schedule before any of them has sanctioned anything. It is expensive to organise and it is
the only move that works when the binding constraint is a hub's free capacity rather than any one
discovery's cost.
Ringvei Vest is the worked precedent. The Troll B area carries several discoveries that
each screen poorly against a dedicated route and considerably better against a shared one — the
joint solution lands roughly a third above the sum of the standalone cases. Nothing about that
result is exotic: the shared flowline corridor is what makes the marginal members viable, and the
marginal members are what make the corridor worth laying. That reciprocity is the test. Where it
holds you have an area development; where one member carries the corridor alone you have a tie-back
with passengers.
The counterweight is the Johan Castberg satellite work, and it is the more common outcome.
There, paying to expand the host destroyed value rather than creating it: the expansion capital has
to be carried by the marginal barrels it unlocks, and those are the barrels least able to carry it.
The break-even expansion cost is computable — around 3,300 MNOK on that case — and above it waiting
for natural ullage beats paying for capacity. So the table below ranks queues by what is waiting,
not by what a joint solution would be worth: the first number is available from the screen, the
second needs the area study the first one justifies.
Host
Contenders
Not seated
Queued MMboe
Queued NPV MNOK
Median break-even
Cheapest
Waiting
4
3
185.1
2,709
$58
$41
Cerisa West, Ofelia, Peon
2
1
17.4
-125
$66
$66
Beta
3
1
11.3
-278
$85
$85
Iving
2
1
4.6
-369
$127
$127
Carmen
2
1
2.6
-699
$278
$278
Tjalve
2
1
0.4
-719
—
—
DELTA
4
1
13.2
-895
$137
$137
Lille-Frigg
7
4
40.6
-1,738
$59
$58
Iskrystall, Kramsnø, Neiden, Skruis
17 host(s) carry more than one screened contender with at least one left unseated by the capacity-constrained portfolio. Queued NPV is the sum of the unseated cases valued standalone against that host — it is the prize an area solution has to beat, not a claim that a joint development would realise it. A negative figure says the queue is not worth unlocking at current assumptions, which is itself a finding: the constraint there is cost, not access.
Showing 8 of 10 clusters. The monthly edition carries the rest of this table. About this sample
6 · Transactions, corporate positions and who is moving
20 transactions recorded, 7 with a disclosed value and 3 stated as pure swaps — so only 1 supports a defensible dollars-per-barrel multiple. On this shelf, comparables are the scarce input.
Two independent records sit in this chapter and they answer different questions.
More on this
The curated
transaction table records deals as they were announced: who bought what, and where a
party chose to disclose it, for how much. The ownership-change feed beneath it records interests
as they actually moved in the licence register. A deal appearing in both is confirmed
against open data; a transfer appearing only in the register is activity nobody put a price on,
which on this shelf is most of it.
20
transactions recorded
7
with a disclosed value
3
stated as no-cash swaps
10
value withheld
9/20
corroborated in the register
Announced transactions
Every row is sourced to a primary public announcement — a company release or an exchange
notice — linked in the final column. The table runs south to north: North Sea first, then the
Norwegian Sea, then the Barents.
More on this
Where a consideration was not disclosed the cell says so: no
value is estimated, and no figure is carried over from a third-party valuation service. Implied
dollars per barrel appear only where a volume basis exists, and are flagged when that volume is
derived from the public reserves register rather than stated in the announcement.
A deal's province is matched from the assets it names against the discoveries this book has
screened, so a package naming nothing screened carries no province and sorts last rather than
being guessed at. A deal spanning provinces sorts by its southernmost.
Announced
Buyer
Seller
Type
Assets
Consideration
Implied $/boe
Source
2026-06-30
Vår Energi ASA
Equinor ASA
swap
Vår Energi receives 32.5% and operatorship of the Peon gas discovery (PL269 and the PL318 group); Equinor receives 5% Fram, 40% of Mulder and Grønngylt, 15% of PL090JS and 10% of PL925 Grosbeak
DNO receives 5% of the Gjøa field and the Gjøa Nord discovery (PL153) plus a balancing payment; Vår Energi receives 15% Ringhorne North (PL956) and 5% Nova (PL418)
Aker BP receives 19% across the Ringvei Vest licences (PL090JS, PL248I and PL925 Grosbeak, PL248C Swisher, PL630 Toppand, PL923 Røver Nord and Sør) and 38.16% of UK P2343; Equinor receives 7.5% of Wisting (PL537)
DNO receives 19% Atlantis and 10% Afrodite (gas-condensate discoveries near Kvitebjørn); DNO transfers its Røver, Mistral, Tyrihans East and Bergknapp interests plus the Sjørøver exploration licence
Showing 8 of 20 transactions. The monthly edition carries the rest of this table. About this sample
How to read this — the in-situ position and what the deals imply
1 of these 20 transactions supports a defensible $/boe. That is the finding, not a gap in the table: the consideration is withheld on 10, 3 moved no cash at all, and several of the rest disclose only one leg of the price. 6 of them are swaps. A swap is a statement that both sides preferred concentration to cash, which is what a shelf does when operators are rationalising positions around infrastructure they already control rather than entering or leaving. It also means no price is set. Of the 19 whose assets match something this book has screened, 16 touch the North Sea, against 2 in the Barents Sea. 9 of 20 are independently visible in the ownership register, which is the check on whether a curated deal happened as described.
The practical consequence is that a position on this shelf cannot be marked to transactions. There is no comparable set deep enough to carry a valuation, and the deals that do price are priced on the buyer's own view of a specific development, not on a market clearing rate for barrels in the ground. An owner asking what an interest is worth has to answer it bottom-up — volume, concept, host, capital, the fiscal regime — which is what the rest of this book does for every discovery on the shelf, on one consistent basis.
Read the record for direction instead of price: who is accumulating around which hub, which licences are being tidied ahead of a decision date, and which counterparties keep appearing. That is a reliable signal, and it is available here because the register publishes the moves even when nobody publishes the money. The map below is that record on the geography; the deadline table in chapter 7 is what forces the next round of it.
The implied-multiple column is mostly empty, and that is the finding.
A dollars-per-barrel
figure needs a price and a volume disclosed on the same basis. On this shelf that combination is
rare: parties disclose a price without volumes, or volumes without a price, or a number that
prices only part of the package — a balancing payment in a swap, the cash leg of a share-based
merger, a post-tax figure. Of the transactions above, only one supports a defensible multiple.
Where an enterprise value was disclosed alongside an equity cheque the multiple is taken on the
enterprise value and labelled as such; dividing the cash payment by the reserves acquired would
understate the deal by whatever debt came with it.
This is the practical case for valuing bottom-up. A comparable-transactions approach here is
starved of comparables, so chapters 2 and 4 compute what each asset is worth from its own
development economics instead of inferring it from what somebody else paid. Use the multiples as
a sanity check on the engine, never the other way round.
Where the equity moved
Derived from the dated licensee history in the public register: entries, exits and
working-interest changes, placed on the acreage they happened to. This is the part of deal flow
that needs no announcement to become visible, and it is where a portfolio reshuffle shows up
before anyone writes about it.
More on this
Reading a move as a row in a list loses the thing that
usually decides whether it matters — what it sits next to. An extra nineteen points in a licence
sixteen kilometres from a host with ullage is a development position; the same nineteen points in
open water is an option on a future well. The map answers both at once, and the cap table on each
callout says who is now at the table when that decision gets taken.
Every registered change in the window
Effective
Company
Asset
Interest
Counterparties
2026-06-30
Equinor Energy AS
36.0% → 75.0%
—
2026-06-30
DNO Norge AS
25.0% → 40.0%
—
2026-06-30
Equinor Energy AS
36.0% → 75.0%
—
2026-06-30
Wellesley Petroleum AS
5.0% → 30.0%
Equinor Energy AS
2026-06-30
Equinor Energy AS
60.0% → 35.0%
Wellesley Petroleum AS
2026-06-30
OKEA ASA
20.0% → 50.0%
—
2026-06-30
OKEA ASA
20.0% → 50.0%
—
2026-06-30
Wellesley Petroleum AS
entry at 20.0%
—
Derived from Sodir licensee-history tables (NLOD 2.0). Re-dating noise is coalesced, so events mark genuine working-interest / operator changes only. Buyer/seller attribution is inferred by matching opposite-sign same-date moves on the asset by magnitude — exact-size pairs are tagged 'exact', unbalanced remainders are split pro-rata and tagged 'partial' (Sodir does not link buyer and seller); transaction prices are not published. Showing 26 of 128 changes in the last 12 months.
Showing 8 of 26 ownership changes. The monthly edition carries the rest of this table. About this sample
Who holds the shelf
The chart below splits every company's position into what it has already booked as reserves in
producing and sanctioned fields, and what it holds as its net share of undeveloped discoveries we
have screened.
More on this
The split is the point. A company weighted towards booked reserves is harvesting a
position; one weighted towards undeveloped resources is carrying a development problem, a
farm-down opportunity, or both — and that is where a deal conversation starts.
Booked reserves in fields (Sodir)Net share of screened undeveloped discoveries (AFP estimate)
Legacy company names are folded into their current entity, so a position reads as one company through its mergers. The first segment is the regulator's net figure; the second is licence ownership times the volume we screened, and moves when a licence trades.
Table view — company positions
Company
Booked reserves MMboe
Undeveloped MMboe
Total MMboe
% undeveloped
Equinor Energy AS
4,398
671
5,069
13%
Petoro AS
3,826
214
4,040
5%
Aker BP ASA
1,518
414
1,933
21%
Vår Energi ASA
1,123
189
1,312
14%
TotalEnergies EP Norge AS
717
3
719
0%
Harbour Energy Norge AS
412
72
484
15%
ConocoPhillips Skandinavia AS
361
31
392
8%
A/S Norske Shell
379
0
379
0%
ORLEN Upstream Norway AS
306
26
332
8%
DNO Norge AS
183
83
266
31%
INPEX Idemitsu Norge AS
102
128
230
56%
OMV Norge AS
121
4
125
3%
OKEA ASA
81
19
100
19%
Repsol Norge AS
33
0
34
1%
Lime Petroleum AS
14
8
22
35%
7 · Drilling inventory, results and licence deadlines
2 wells are on the bit and 9 more are permitted — 11 forward wells across 11 licences, one well each — the forward inventory as the regulator has it, not as anyone has compiled it (a further 2 open permits are contingent sidetracks whose main bore was already drilled dry, and are excluded). Separately, 16 screened discoveries sit in licences expiring within two years, and a deadline forces a decision whether or not the project is ready.
What is on the bit, and what is permitted next
A granted drilling permit is the last public step before a well is spudded, so the permit
register is a forward inventory rather than a compilation of intentions.
More on this
Wells appear here on the regulator's record, with a permit number, an operator and a rig —
before anything is drilled. Nothing on this page is an estimate of what a well will find; it is a
statement of what has been authorised, by whom, and on what acreage. A permit can lapse, and some
do, which is why the permitted wells are shown without an invented spud date.
2
on the bit today
9
permitted, awaiting spud
2
rigs committed
4
operators with a forward well
81
drilled in 24 months
34/65
wildcat discovery rate
on the bitnew discoveryhydrocarbons, not a new discoveryshows onlydry
Each dot is one wellbore; a month's column height is its spud count. Permitted wells carry no public spud date and are counted in the rail rather than placed on the axis. Source: Sodir wellbore and drilling-permit registers, snapshot 2026-07-14.
Browse the same wells as cards, with rig, licence and cap table
On the bit 2
Spudded, still drilling. Result lands within weeks.
2 prior E&A wells, 1 of 2 wildcats found something — last 2022 shows
Permitted, awaiting spud 9
Sodir has granted the drilling permit. Flagged rows are contingent sidetracks whose main bore has already been drilled. The 2 flagged below are lapsed contingencies, listed but not counted in the 9.
The same records as a table. Filter it to any company to see only the wells that company carries an
interest in.
Licence
Wellbore
Prospect / pre-drill
State
Operator
Working interest (operator underlined)
Rig
Last E&A well in this licence
Wildcat hits
Licence to
PL979
15/6-17
Svarteknippa 10–60 MMboe
on the bit
Aker BP ASA
Aker BP ASA 60%*; Equinor Energy AS 40%
SCARABEO 8
—
—
2029
PL782S
25/7-13
Linga 10–50 MMboe
on the bit
Equinor Energy AS
Equinor Energy AS 60%*; Aker BP ASA 40%
COSLInnovator
25/7-10 2022 · shows
1/2
2045
PL1174S
16/4-14
Pedalo North
permitted
Equinor Energy AS
Equinor Energy AS 80%*; Petoro AS 20%
COSLInnovator
—
—
2030
PL169
25/11-30 A
Lit⚠ contingent
permitted
Equinor Energy AS
Equinor Energy AS 57%*; Petoro AS 30%; Vår Energi ASA 13%
DEEPSEA ATLANTIC
25/11-30 S 2025 · dry
5/12
2030
PL1203
25/8-24 S
Jotun Deep 60 MMboe · 20–50% (partner band, medium)
permitted
Vår Energi ASA
Equinor Energy AS 30%; Vår Energi ASA 30%*; DNO Norge AS 20%; Petoro AS 20%
COSLPioneer
—
—
2030
PL190
30/8-7 S
Grimsnes
permitted
Equinor Energy AS
Equinor Energy AS 50%*; Petoro AS 40%; TotalEnergies EP Norge AS 10%
DEEPSEA BERGEN
30/8-5 2017 · dry
4/5
2032
PL050HS
34/10-56 S
Rav Beta
permitted
Equinor Energy AS
Equinor Energy AS 70%*; Petoro AS 30%
COSLInnovator
—
—
2031
PL554E
34/8-20 A
Narvi Brent Sor⚠ contingent
permitted
Equinor Energy AS
Equinor Energy AS 40%*; Aker BP ASA 30%; Vår Energi ASA 30%
COSLInnovator
34/8-20 S 2025 · dry
0/1
2030
Showing 8 of 94 wellbores. The monthly edition carries the rest of this table. About this sample
How these wells were selected from the register
Where the acreage history comes from. The last-well and wildcat-hit columns are computed over the entire exploration wellbore register rather than maintained by hand, so they are complete for every licence in the table.
Hits are counted off the discovery-wellbore flag, not the discovery name. The register names a discovery against appraisal wells too, because an appraisal is drilled on a find that already exists. Counting names rather than the flag can therefore overstate success. On this window it happens not to bite: across the 65 wildcats drilled in the last 24 months the two constructions land within one of each other (33 by name against 34 by flag), because no appraisal campaign fell inside it. The flag is used anyway — the gap is a property of the window, not of the method. Hits are therefore prior wildcats that were the discovery wellbore, over prior wildcats drilled, with appraisals excluded from both sides. All classifications are the regulator's, not an operator resource statement.
Licence clocks on screened discoveries
The most reliable predictor of an interest coming to market is not economics but a deadline.
Every screened discovery whose licence carries an expiry date is here, sized and valued — the whole
register rather than the nearest few, because the shape of the pipeline behind the urgent front is
what tells you whether this is a wave or a trickle. The nearest items also appear in the
near-term calendar in chapter 1, alongside host cessation dates and
first-oil years.
More on this
Where a licence carrying a screened discovery approaches expiry, a decision is forced whether or
not the project is ready — drill, commit to a development, trade the interest, or hand the acreage
back. For a buyer, that calendar is the cheapest source of negotiating leverage available.
Discovery
Licence
Expires
Years left
MMboe
Best NPV MNOK
35/12-2 (Grosbeak)
PL925
2026
0
87.3
2,638
34/4-15 S (Dugong)
PL882
2026
0
32.4
754
31/1-2 S (Røver Nord)
PL923
2026
0
30.1
349
35/6-3 S (Ofelia)
PL929
2026
0
12.0
-240
35/10-9 (Heisenberg)
PL827 S
2026
0
1.3
-744
30/2-5 S (Atlantis)
PL878
2027
1
8.0
-596
6407/8-4 S (Galtvort)
PL1223
2027
1
0.1
-823
3/7-8 S (Trym Sør)
PL147
2027
1
0.3
-1,760
All 89 screened discoveries carrying a licence expiry date: 16 inside two years, 15 in three to five, 58 beyond. Licence validity from the Sodir licence register. A short clock forces a decision — drill, commit to a development, trade the interest, or hand the acreage back.
Showing 8 of 89 licences. The monthly edition carries the rest of this table. About this sample
13 of 13 permitted and drilling wells have a prospect sourced to a primary operator or regulator document, of which 7 carry a publicly stated volume and 3 a stated chance of success. Every forward well in this snapshot is covered; the gap this section exists to report is currently closed, and it reopens the day a permit is granted for a well nobody has read a filing on. A further 1 curated row (7220/5-EC-2H) describes a well that holds no permit and is not on the bit in this snapshot, so it is carried but not counted as coverage.
Showing 8 of 14 prospects. The monthly edition carries the rest of this table. About this sample
Every figure above is quoted from the linked operator or regulator document and is reproduced, not modelled — the basis column states exactly what the source said it was. Nothing here is scaled, risked or blended by us, and no subscriber research contributes a number: a row citing one is rejected by the loader whatever confidence it claims.
Two things this table refuses to do. Chance of success is shown as the band the operator published, never as its midpoint — on this shelf only one company discloses it at all, and only as bands. And a resource figure booked after a well was drilled is not a pre-drill estimate: where the only public number is a post-discovery booking, the volume cell stays empty rather than borrow it.
What a wildcat is actually worth drilling into
The prospect volumes above are the operators' own, and only three of the wells carry
a chance of success at all — one company's band, repeated. That is not a dataset. What the register
does carry is the outcome of every wildcat drilled on this shelf since 1990, and that turns out to
answer a more useful question.
Why this holds
An exploration engineer is normally handed a chance of geological success
and left to treat "will we find hydrocarbons" as the question. On the Norwegian shelf it stopped
being the question some time ago, and this book's own contents prove it: the register lists dozens
of undeveloped discoveries, every single one of which was a technical success. Finding it was never
the hard part.
So the rate below is a chance of commercial success, and it is the product of two things
the book already computes in different chapters. The first is the register's verdict on the wells:
of the wildcats drilled in this province, what share were the wellbore that made a new find. The
second is the minimum economic field size of chapter 8: given what this province typically yields, what
share of finds are large enough to pay for a route to market. Multiply them and you have the
probability that drilling here ends in something buildable.
It is computed twice, against two thresholds, holding the prospect constant. Once assuming the
find lands within tie-back reach of a host that still has room, and once assuming it does not and a
standalone development is the only route. The distance between those two answers is the reason this
chapter sits in the same book as the host atlas: on a mature shelf the dominant term in an
exploration decision is not the prospect, it is whether anything nearby can process what you
find.
Found hydrocarbonsCommercial, tied backCommercial, standalonebars are per 100 wildcats, so they are directly comparable
43%of 1,026 wildcats since 1990 made a new discovery — technical success on this shelf is close to a coin flip
43%chance of a commercial outcome in the Barents Sea when the find lands within tie-back reach of a host with room
0.8%chance in the North Sea when it does not, and a standalone development is the only route
17×median ratio between those two answers across the shelf — the largest single lever on an exploration decision, and it is not in the prospect
Read this as a base rate, not a prospect risking. The first stage is the register's own verdict on 1,026 wildcats drilled since 1990 — a well counts as a discovery only where Sodir flags it as the wellbore that made the find, which is stricter than "encountered hydrocarbons" and roughly seven points lower. The second stage asks whether a find of typical size for that province clears the minimum economic field size from chapter 8, once against the tie-back threshold and once against a new-build. Both stages are realised outcomes; neither is a forecast, and a specific prospect should be argued against these numbers rather than replaced by them.
The find-size distribution is the 92 undeveloped discoveries the register currently carries with a booked volume. That is a survivorship-free sample of what this shelf yields and cannot develop, which is exactly the population an exploration decision should be measured against — but it excludes finds already developed, so it understates the upper tail.
Exploration base rates by province — every input
Province
Wildcats
New discoveries
Dry
Finds sized
P90 MMboe
Median
P10
Tie-back needs
Standalone needs
CoS tied back
CoS standalone
North Sea
640
266 (42%)
247
51
4.5
25.1
65.0
13
154
28%
0.8%
Norwegian Sea
271
124 (46%)
105
25
6.0
29.1
59.4
13
182
31%
1.8%
Barents Sea
115
56 (49%)
38
16
12.4
24.2
110.1
13
182
43%
3.0%
What this rate is not.
It is a base rate for a province, and a base rate is
something a prospect-specific number should be argued against, not replaced by. It says nothing
about your seismic, your analogue or your seal — a well-understood prospect in a proven segment
deserves better than the province average, and a stratigraphic play deserves worse. Two further
limits worth stating: the find-size distribution is drawn from discoveries the shelf has
not developed, which is the right population to measure an exploration decision against
but understates the upper tail because the successes have left the sample; and the tie-back
threshold assumes a host with free capacity, which chapter 5 shows is exactly the thing in
shortest supply.
8 · The seat view — Equinor Energy AS
The same engine output as every other chapter, restated at a company's working
interests. Nothing here is a different analysis; it is the shelf analysis with equity applied,
which is what turns a shelf view into a work programme.
Every company the shelf's ownership register knows has a seat, and every seat ships in every
edition — the list is derived from who holds what, not assembled from who was asked. There is no cut
prepared for one subscriber and withheld from the rest. Take any company's chair, or read the shelf
as an independent with no seat at all: the netting, the ranking, the host franchise and the
minimum-economic-field-size chart all re-compute for whichever seat you pick. A seat changes the
framing, never the contents.
More on this
Switching between two
companies on the same acreage is the fastest way to see why a counterparty values an asset
differently — the same barrels, a different equity share and a different set of hosts already in
hand. Sign-in decides which seat the book opens on, and the address bar carries it, so a particular
view is a link you can send someone.
One seat in this sample. The chapter above says every company has a seat and every seat ships in every edition. That is true of the edition and not of this cut, which carries Equinor Energy AS alone. A subscriber's copy offers all 22 and re-nets every figure on the shelf to whichever chair you sit in. About this sample
Counterparts — who this seat should be talking to
The seat above is what a company holds. This is who else is standing near it. Three joins, each
printed rather than weighted: a discovery within tie-back reach of a host somebody else owns, two
undeveloped discoveries close enough to be one area development, and licences the two already hold
in common. The threshold column is the host's own minimum economic field size at that
step-out, read off its curve — so a row does not say two assets are near each other, it says whether
the barrels pay for the pipe.
Hosts are attributed by ownership, not operatorship. Only seven companies on this shelf
operate a single host; a seat that co-owns a platform without running it still owns the ullage, and
keying on the operator column would leave four seats with no infrastructure position at all when
they in fact hold some of the shelf's largest.
More on this
Counterparts are ranked by
their strongest single clearing join, not by how many joins they have. Ranking on the count ranks by
how much of the shelf a company owns, which puts the same two or three names at the top of every
seat. Equinor still leads most seats — that is what operating half the shelf's hosts looks like, and
it is said once here rather than implied on every card.
The counterpart scan is not in this sample. It is a cross-seat product — it joins one company's acreage against every other seat's hosts and licences — so with one seat here there is nothing for it to join against. The panel below reports no counterparts for that reason, not because the edition finds none. About this sample
9 · How to improve value — the levers, ranked
Everything above is diagnosis. This is the part a business developer is paid for:
given that the shelf looks like this, what actually moves a number. The levers below are ordered by
how much value they typically release per unit of effort, based on what the screening shows moves
between concepts.
1
Negotiate the better host, do not accept the default
Summed across the 12 candidates with a scored alternative, the gap between the best concept and the runner-up is worth this much. It is commercial work — access and tariff — not technical work, and it is usually larger than the value of another appraisal well.
The median break-even of research-anchored cases sits this far below the generic-cost cases. Most of that gap is estimation conservatism, not real cost — so pinning a public cost basis on an asset you hold is the cheapest re-rating available.
67 screened discoveries want a host slot the capacity-constrained portfolio cannot give them, across 27 contested hosts. Being first in that queue is worth more than improving the project, because the project cannot proceed at all without the slot.
Cluster the near-marginal, do not develop them alone
21 discoveries break even between $40 and $70/bbl — individually marginal, collectively able to share a route, a template and a host slot. This is where area projects come from, and the only lever that turns un-developable volume into inventory.
16 screened discoveries sit in licences expiring within two years. Their holders face a forced decision regardless of readiness, which is the cheapest negotiating position a buyer ever gets handed.
What does not work, and why it keeps getting tried.
Two moves recur in area studies and
both usually fail. Funding a host expansion to unlock satellites transfers the expansion capital
onto the marginal barrels, which are the least able to carry it — above a computable threshold
cost, waiting for natural ullage beats paying for capacity. And appraising a marginal discovery
further rarely rescues it: the uncertainty that matters is commercial, not volumetric, so another
well tightens the wrong band. Both are attractive because they are actions a technical
organisation knows how to take. The lever that works is usually a negotiation.
10 · Economic basis, assumptions and provenance
Why this holds
The claim this book makes is reproducibility, so the basis is stated in full rather than
summarised. The figures below are read live from the same configuration the screening consumed —
they cannot drift from the numbers the valuations were computed with.
What "the screened set" means here. Every count in this book is taken over the
93 oil and gas cases it analyses, of which
76 have a solved break-even. A further
3 CO₂ storage cases exist in the same screening revision and are
deliberately excluded throughout: their economics are a tariff per tonne, not a break-even oil
price, so ranking them against hydrocarbons would be meaningless. Where a chart drops cases it
says so in its own caption.
Prices and discounting
The base case. Every break-even in this book is the oil price at which net present value reaches zero holding everything else here fixed.
Input
Value
Note
Oil price
$60/bbl
screening base case, real terms
Gas price
$9.0/MMBtu
valued against oil on an energy-equivalent basis
Price sensitivity band
$45 – $75/bbl
the sweep behind every price-sensitivity statement
Discount rate
8% real
post-tax, applied to all net present values
Field life
20 years
screening horizon cap
Decline rate
14%/yr
post-plateau, where no profile is researched
Operating cost
$14/bbl
screening default; overridden by a research anchor where one exists
Fiscal regime
Norway's marginal rate is high and its uplift is generous, which is why a pre-tax and a post-tax break-even can differ enough to change a decision.
Input
Value
Note
Ordinary corporate tax
22%
Norwegian petroleum tax regime
Special petroleum tax
56%
combined marginal rate 78%
Investment uplift
12.4% × 4 years
applied to qualifying capital expenditure
Exchange rate
10.5 NOK/USD
screening deck
Carbon cost
included
CO₂ tax and allowance cost carried in every case; host power source affects the figure
Timing and conservatism
Where the screening deliberately leans pessimistic, and by how much — the conservatism is a stated policy, not a hidden margin.
Input
Value
Note
Tie-back lead time
3 years
from decision to first oil
New-build lead time
6 years
standalone development
New-build screening price
$40/bbl
deliberately below the tie-back price — a new-build must clear a harder test to be believed
New-build hurdle rate
12%
higher than the portfolio rate, reflecting execution risk
New-build capital uplift
×1.35
conservatism applied where no researched cost exists
Uncertainty band
P90 ×0.60 / P10 ×1.50
factor sweep producing the P90–P10 range on every case
What moves a break-even most
Measured on the 76 cases in this revision that carry a break-even, a capital estimate and a volume — rank correlation against break-even price:
Driver
Correlation
capital intensity (CAPEX per barrel)
+0.98
recoverable volume
-0.88
Capital intensity is the stronger of the two here (+0.98), which is what you would expect on a tie-back shelf: revenue is largely fixed by the reservoir while cost is negotiated. Host tariff and price are held at the deck values above and so do not vary across cases — their influence cannot be read off this revision and is not claimed here.
The practical consequence: if you disagree with one number in this book, disagree with the capital estimate. The per-case input tables in chapter 4 show exactly which capital figure was used and where it came from.
What the engine does, in order
Read the booked recoverable volume and fluid split for the discovery from the public
resource register.
Apply any curated research anchor that passes a confidence gate — a published project cost,
a measured gas-oil ratio, a stated concept.
Enumerate candidate concepts: a tie-back to each host inside fluid-dependent reach, plus a
standalone development.
For each concept, place templates and wells, then solve the subsea layout and flowline
network.
Build a production profile and run it through the Norwegian fiscal regime, including the
special petroleum tax and uplift, host tariffs and carbon cost.
Find the break-even oil price by bisection, and sweep a P90/P50/P10 factor band.
Keep the best concept by net present value; then re-solve the assignment across all cases
under host capacity constraints.
What it does not do
It does not know a host's real commercial terms, which are private.
More on this
It does not know whether a
licence group will agree a concept, or whether a partner will veto one. It does not carry an
operator's internal cost estimate unless that estimate is public. Where a project's actual capital
cost has been published the engine's figure can be compared against it directly; where it has not,
the number is a parametric estimate and is labelled as one everywhere it appears. None of this is
a substitute for a licence holder's own evaluation — it is a way of getting to a defensible first
answer across a whole shelf on one consistent basis.
Valuation basis and input provenance
Valuation basis across the 93 analysed cases
Basis
Cases
Meaning
Research-anchored
18 of 93
a curated public-source anchor pins at least one economic input
— of those, with a solved break-even
16 of 76
the population the supply curve and its medians are drawn from
Engine-valued
91 of 93
full MILP layout and cash-flow solve
Proxy-valued
2 of 93
volume-booked discounted cash flow where the layout solve could not close
Unresolved contradictions
271 of 93
sources disagree on a discovery input and neither has been adopted
— host-level, on tie-in windows
27 of 31
a host's curated cessation year disagrees with a projection from its own production decline; both candidates retained. UNFILTERED — the projection is an oil-exhaustion date, not a facility cessation, so this counts unreconciled dates and not errors (see the host table)
Where the inputs came from
Origin
Inputs
Snapshot
96
Research Overlay
18
Override
0
Default
86
Counted across every economic input the screen consumed. "Default" means a parametric cost-database assumption — honest, generic, and the first thing a licence holder should replace with its own estimate.
Reproducing this edition
Screening revision
hub-20260726T200807Z-12366484 (revision 12)
Solved
2026-07-26T20:12:59+00:00
Engine commit
9c02dc6
Sodir snapshot
2026-07-14T08:59:35+00:00
Cost database
cd99f047d99b
Research overlay
fd05bf23a367
Where our own sources disagree
Every number in this book is traceable to the source it came from. On a handful of inputs there
is more than one defensible source and they do not agree, so those are listed here rather than
resolved quietly.
Why this holds
A traceability claim is only worth something if it survives the cases where
tracing leads somewhere awkward. Two operators state a tie-back distance differently because they
are routing to different hosts; a licence expiry is one year in the register and another in a
partner release; an ownership split changes hands between a swap being announced and being
re-papered. These are not data-quality defects to be cleaned up — they are the shelf being genuinely
ambiguous, and a reader who has to defend a number needs to know which ones are contested.
Several of the rows below record something more uncomfortable: that the value this application
previously stored is the weaker of the two. Those are kept in deliberately. The screening uses the
higher-confidence value and says so, but a register of disagreements that quietly omitted the ones
where we were wrong would be advertising rather than provenance.
None of these are resolved by us preferring an answer. They are resolved when someone publishes a
better source, and until then the disagreement is part of the basis of preparation.
Var Energi KU Gjoa Subsea Projects, 16 Mar 2026 - Cerisa share of the programme, stated exactly as 3 milliarder. https://varenergi.no/wp-content/uploads/2026/03/Konsekvensutredning-Gjoa-Subsea-Projects.pdf
Var Energi discovery release Dec 2022 / GGE PDO 2025-26 (tie-back to Goliat FPSO)
82.6 km
low
2024
App stored value (direct Melkoya routing)
35/2-1 (Peon)
Tie-in host
GJØA host
high
2026
Equinor-Var Energi asset swap, 30 Jun 2026 (concept stated as a ~60 km tie-back to Gjoa, gas processed at Karsto)
GULLFAKS host
low
2026
INFERENCE from Sodir planned-survey EQ26020 (Equinor Energy, vessel Fugro Zephyr, 01.09-31.10.2026), read as a corridor between Peon and Gullfaks - https://factpages.sodir.no/no/survey/TableView/Planned
35/12-2 (Grosbeak)
Tie-in host
TROLL B host
high
2026
Equinor Ringvei Vest concept 18 Jun 2026 (seabed separation then ~33 km trunk to Troll B)
GJØA host
low
2026
Pre-2026 tie-back study (Sodir factpage cites a Gjoa tie-back); host choice reported as still open into June 2026
7 input(s) on which two defensible sources disagree. The screening uses the higher-confidence value — the top row of each pair — and several of these record that the value this application previously stored is the weaker one. None of them are resolved by us preferring an answer; they are resolved when someone publishes a better source, and until then the disagreement is part of the basis of preparation rather than something to tidy away.
Data downloads
Exports are part of a subscription. The edition ships the full 93-case screened inventory and the 20-transaction record as CSV. About this sample
Why this holds
Data licence and attribution. This report contains data made
available by the Norwegian Offshore Directorate (Sodir) under the Norwegian Licence for Open
Government Data (NLOD) 2.0. Sodir does not endorse this analysis and bears no responsibility for
it. Resource volumes, ownership records, licence status, facilities, production history and
wellbore data are reproduced or derived from the public FactPages; the snapshot date is stamped
above.
Independence of sources. Transaction records are curated from primary public
announcements only — company releases and stock-exchange notices, each linked. No figure in this
report is taken from a third-party subscription research product. Development economics,
break-even prices, layouts, portfolio assignments and cash flows are computed from the inputs
disclosed above by AFP, the development-planning engine behind NCSpeil.
Nature of the analysis. This is screening-grade outside-in
analysis for business-development use. It is not investment advice, not a reserves certification,
and not a substitute for a licence holder's own technical and commercial evaluation.