Frac jobs disclosed to FracFocus · activity through September 14, 2026 · published September 15, 2026
The September permit file landed, and that is the lead.
Five issues ran on one RRC file dated Aug 8, and for the last four of them every Texas permit number in this newsletter was a reprint. The
RRC's Sep 9 cut is in: 4,891 permits issued Sep 1, 2025 – Sep 8, 2026 in the Permian counties,
4,184 of them horizontal (85.5%). August, which printed as a 102-permit stub for five weeks, fills in to
397 — flat against June (408) and July (395). The freeze hid nothing dramatic. The county map book
grows to 4,051 permits, the first growth in four weeks, and every permit arrow in the momentum table is a measurement
again rather than a reprint. The permit cards say what is and is not comparable to last week.
Second: another thin filing week, with a large revision behind it. Against last week's extract,
98 new disclosures landed and 0 disappeared (last week 65/5). This issue's 28-day
tile reads 94 jobs and the settled pace 304.0 per 28 days. But recompute last issue's own 28-day
window on this morning's filings and its published 109-job tile is already 186 jobs — a 77-job late-filing
revision, the largest we have recorded — while its 308.0 settled pace recomputes to 313.0
(+5.0). Read this week's 94 as a floor, not a count.
And read the last columns of every chart with a grain of salt. The week ending September 14 shows
2 jobs on file against a settled pace nearer 76 a week — the newest job end anywhere in
this extract is 2026-09-09, so the week has barely begun to be filed. The same shape appears on the
drilling-versus-completions chart, where the Sep 11 point is 5 frac jobs against 268 rigs. That is
filing lag, not a shutdown.
Also in this issue: a feature on what the completion filings say about the Barnett, and a method change to the
formation table that closes a contradiction the Sep 7 issue printed without noticing.
Frac jobs, last 28 days
94
Aug 18 – Sep 14, 2026 · filings still arriving
New Mexico / Texas split
27 / 67
28.7% of jobs in NM — against last issue's window recomputed at 28.5% (53 of 186), not the 32.1% it printed on the day
Settled pace
≈304 jobs/28d
jobs ended 42–126 days back ÷ 3 — the fully-reported baseline
Median water per job
525,125 bbl
median, n = 94 jobs reporting (the mean is 567,000)
Median proppant per job
21,633,179 lbs
median, n = 71 jobs reporting (the mean is higher, at 22,100,000)
Produced / recycled water
92.5%
volume-weighted share of disclosed base water — up again from last issue's window recomputed at 86.3% on this morning's filings (it printed 82.9% on the day) · by jobs it is 90.6% (48 of 53 reporting a source) · 53 of 94 jobs report source · record-level 85.7% (48 of 56 rows). The tile leads on the volume-weighted figure, as last issue; on a 53-job reporting sample a few large jobs move it, so read the level, not the decimals
EOG, week two — the first step of the schedule happened, exactly as computed
FracFocus disclosures filed by EOG Resources with a 2026 job end, read against the New Mexico
regulator's permit and DUC counts (OCD, queried Sep 15). Last issue printed a computed schedule for when EOG's
August jobs would enter the settled window; this issue reports the first step of it.
What we said last week. That EOG's settled pace would move on the anchor arithmetic, not on new filings:
Aug 4 at the Sep 15 anchor, Aug 5 and 10 at Sep 22, Aug 20, 21, 23 and 24 at Oct 6.
What happened.EOG filed zero new disclosures for a second consecutive week. Its
19 disclosures with 2026 job ends are the identical set; its newest job end is still 2026-08-24, now
22 days back (from 15). Its settled pace moved 1.7 → 2.3/28 d, and that is the
first step happening exactly as computed: the
Aug 4 jobs, 2 of them, crossed the 42-day line at this anchor and now sit in the settled window
(2026-05-13 to 2026-08-04) alongside the Jul 16 batch. Its filed-last-28 count is
5 — the Aug 20, 21, 23 and 24 jobs — and the remaining steps stand: Aug 5 and 10 at the Sep 22 anchor and Aug 20, 21, 23 and 24 at the Oct 6 anchor. EOG is still
flagged current at 22 days, because the flag measures the age of the newest filing and nothing else; it
crosses into slow at the Oct 6 anchor if nothing new is filed before then.
New Mexico has paused too. After three weeks of falls, EOG's unspudded permit bank was all but
flat, 971 → 970 (−1, after −28 last week), and so was its New Mexico DUC stack,
852 → 853 (+1, after +37). The permit-to-drilled-well conversion
that dominated the last two issues paused this week; EOG still leads the basin's DUC table, 119 wells
ahead of Devon (incl. Coterra) on the rolled families the pipeline card prints. Every pace and fleet figure in this issue still understates EOG.
The Barnett, 3,000 feet down — what the completions actually say
There is a Permian Barnett revival story in the trade press: permitting doubled, better
productivity than the Spraberry, less water. This card is not "here is a hot new play". It is what the completion
filings — FracFocus water and proppant, RRC formation tops and field designations, RRC lease production — say about
a play you have been reading about. They agree with the direction and disagree with three of the specifics.
Texas only, because New Mexico has none of it (see the end). Basis: per-job medians over the trailing 24 months
(the stable read; the 12-month sample is too small to characterise), bucketed on the RRC regulatory field name
for Barnett and Woodford and on the deepest RRC bench record for the comparison benches — the same rule the
formation table below adopts this issue. Computed in the Sep 9 analysis on the Sep 8 FracFocus extract, the Sep 5
wellbore file and the Jul 26 PDQ field map; the formation table below re-runs the same rule on this morning's
files and prints Barnett at n = 18 for the trailing twelve months.
Depth is the story, and it is the cleanest number in the file. Median true vertical depth over 24 months:
Barnett 11,682 ft, about 3,100 ft below the Spraberry (8,571) and
2,300 ft below the Wolfcamp (9,338). The Woodford, the other deep bench, sits at
14,694 ft — 6,100 ft below the Spraberry and 4,600 ft below even the Bone Spring
(10,130).
It is small but real.122 Texas completions in 24 months — Barnett 88,
Woodford 34 — out of 7,328 Texas jobs, i.e. 1.7% of Permian completion activity. Growth is real
but it is not a doubling: combined completions went 58 in 2024 to 82 in 2025, +41% raw and
roughly +50% once the field-map coverage of each year is allowed for. And the leaders are not the names in the
coverage: over 24 months they are Continental Resources (35) and Occidental (34).
Bench (TX, 24 mo)
Jobs
Median water (bbl)
Median sand (M lbs)
Median TVD (ft)
Median lateral (ft)
Sand (lbs/ft)
Water (bbl per lateral ft)
Barnett
88
547,488
22.87
11,682
10,301
2,060
51.3
Woodford
34
817,856
25.97
14,694
10,364
2,514
76.4
Spraberry
2,031
518,862
23.85
8,571
10,384
2,016
45.6
Wolfcamp
1,706
539,362
24.64
9,338
10,330
2,259
50.1
Bone Spring
600
498,843
24.44
10,130
10,170
2,468
49.8
Three places our filings contradict the trade narrative.
1. "Lower water" is wrong on the completion side. The Barnett takes 51.3 bbl per
lateral foot against the Spraberry's 45.6; the Woodford takes 76.4. Design is heavier, not lighter.
The industry claim is about produced-water cut over the life of the well, which FracFocus cannot see. The two
get conflated constantly, and a reader who conflates them will size a job wrong.
2. The "Barnett, then Spraberry, then Wolfcamp" productivity ordering does not hold in our
IP90. On the IP90 table in this issue the Barnett makes 62 bopd per 1,000 ft of lateral (73 wells, 72 with a
lateral), the Spraberry 52 and the Wolfcamp 65. The Barnett beats the Spraberry by about 19% and loses to the
Wolfcamp — which the trade ordering puts last. IP90 is a first-three-months number, not a peak-month or EUR number, so
the absolute values are not comparable to the claim; the ordering is, and it fails on the Wolfcamp leg.
3. It is not purely a Midland Basin story. The 24-month county split is Martin 28 / Ector 18 / Midland 11 / Glasscock 5 / Andrews 4 / Howard 1 = 67 on the
Midland Basin side against Crane 21 / Ward 18 / Winkler 9 / Pecos 7 = 55 on the Central Basin Platform and Delaware side. All 34 of the Woodford
wells are in Ward, Winkler and Pecos — 33 of them in the RUPPEL (WOODFORD) field — at 11,400–16,700 ft TVD: a Delaware Basin
Woodford play led by Continental (22 of the 34), with Marathon Oil (5), COG (ConocoPhillips) (5) and Mewbourne Oil (2) on the rest. With Crane's Central Basin Platform
Barnett wells that side is 45% of the combined count. The trailing twelve months in the Sep 9 analysis
(29 wells on its Sep 8 window) is actually Central-Basin-Platform-led (Crane 13, Ward 7, Ector 5).
Two names, checked. Elevation Resources, named in the coverage as developing the play, has
zero disclosures in the trailing twelve months (newest job end Sep 1, 2025, 14 days outside this issue's window). Epsilon Energy
has no FracFocus disclosures as operator at all, which is consistent with a stated non-operated interest.
The permit angle, and why it is a proxy. The RRC permit master carries no field or formation
name, so there is no such thing as a Barnett permit count from this pipeline. Two proxies, both labelled as such. Deep-permit
mix: in the nine Midland Basin core counties the share of permits at 11,000 ft+ total depth went 16.1% → 21.2% → 24.4% → 29.0% → 33.2% (2026 partial), and the
12,000 ft+ count went 446 → 586 from 2024 to 2025. At 11,500 ft+ the two fairway counties roughly doubled 2024 → 2025 —
Ector 28 → 93, Andrews 48 → 95 — while total Permian permits fell 5,472 → 4,499. Total depth on a
horizontal permit is measured depth, not TVD, and it cannot separate the Barnett from any other deep target. Bench-suffix
floor: operators often encode the target bench in the RRC well number, and basin-wide 64 deduped Permian permits
since 2022 end in "BN" — Diamondback E&P 55, Chevron U.S.A. 5, Summit Petroleum 2, Sabalo II 1, Petroplex Energy 1 — all in the fairway at 10,233–14,000 ft. A suffix is a convention, not a filing
requirement, so that is a floor on Barnett permits, not a count; and only 6 of 60 unique BN API numbers show a
completion, so permit-and-do-not-drill is the norm here. Both proxies were measured on the Aug 8 permit master in the
Sep 9 analysis and have not been re-run on the fresh Sep 9 file. The honest framing is that deep permits roughly doubled
while deep completions grew about half that — a real gap worth watching, not a doubling against nothing.
⚠ This card prints no 2026 trend, deliberately. The RRC field map comes from the Jul 26
production dump, and a well only acquires a field designation once it is on the proration schedule; field-join coverage is
62% / 36% / 26% for 2026 Q1 / Q2 / Q3. The apparent collapse to a well or two a quarter is a vintage artifact.
The 2026 counts are floors, not levels.
New Mexico: zero. No New Mexico disclosure in 24 months carries a Barnett or Woodford pool and
the OCD pool set contains no such pool. A Texas-only story, which suits a Texas-only table.
Why the table below used to say four. Until this issue the
formation table found the Barnett only through the RRC wellbore file's formation-top records, and that file stores at
most ten tops per completion filing — the deepest ones, i.e. exactly the Barnett and Woodford, are the rows that get cut.
On the field-name rule the trailing-12-month Barnett count goes 4 → 18 and Woodford
1 → 9. The full method note, and the correction to last issue, is on the formation table.
Special: a decade of sand per foot — the basin found its number
Midland Basin, nine core counties · FracFocus proppant mass joined to RRC W-2
perforated intervals · 20,510 joined wells surviving the sanity filters, 2016 – Sep 2026 ·
median lb of sand per perforated foot by frac year · *2026 is preliminary (n = 336).
This week: ten of the eleven years reproduce to the pound and one moved — 2026 (1,996 → 1,995 on n 313 → 336). The Sep 12
RRC wellbore file took the joined-well count 20,479 → 20,510, and 2025 (n 2,049 → 2,057, median unchanged at 2,040) gained joins without moving.
A one-pound move on a growing preliminary year is the series filling in, not news, and not a method change.
1,474
2016
1,795
2017
1,855
2018
1,852
2019
1,898
2020
1,998
2021
2,002
2022
2,021
2023
2,029
2024
2,040
2025
1,995
2026*
Everyone remembers the loading race. From 2016 to 2018 the median Midland Basin well went from
1,474 to 1,855 lb per perforated foot — a 26% jump in two years, the era when every investor
deck had a "bigger fracs" slide. Then the race just… stopped. Since 2018 the median has crept from
1,855 to 2,040 lb/ft — about 10% in seven years, and the preliminary 2026 column sits at
1,995 lb/ft, statistically indistinguishable from 2025 and, if anything, a shade below it. Call it
~2,000 lb/ft: the design equilibrium the basin settled on once more sand stopped paying for itself.
Two joins have to land for a column to shift — a perforated interval from the RRC and a proppant
mass from FracFocus — and the Sep 12 file added 31 joined wells overall, of which
23 landed in 2026 and 8 in 2025. The 2026 column is becoming a real number rather
than an early one, and it is doing so without leaving the equilibrium.
The sand bill still doubled — but sideways. Median perforated interval went
from 7,642 ft (2016) to 10,499 ft (2025), and median sand per well from 11.8M to
24.6M lb. The growth was laterals, not loading. The 2026 wells joined so far run a median
10,456 ft at the same loading, which is the whole thesis in one line.
Design still isn't monolithic: among 2025–26 completions, Pioneer's
median well takes 30.7M lb (n = 750) against Diamondback's 20.8M (n = 841) — same basin, same benches,
a ten-million-pound difference in what shows up on the pad. That line is computed sand-only, with no perf join.
Method: proppant = disclosed mass of
crystalline silica (CAS 14808-60-7) summed per disclosure; perforated interval = RRC W-2/G-1 perf
span joined by API (2,000–20,000 ft sanity); medians throughout; filings over 100M lb excluded as data errors; a perf span
from a recompletion can differ from the original lateral — a small minority of joins. The joined-well
count is taken after the sand and lateral filters. 2026 reflects wells both frac'd and filed with the
RRC — it will keep drifting as the year fills in.
Weekly frac job count — trailing 13 weeks
By job end date. Faded columns are recent weeks still filling in — operators have up to ~30 days
to file, so the newest weeks always start low and grow. The newest column, the week ending Sep 14, reads 2, and
that is filing lag rather than a shutdown. The newest job end anywhere in this week's extract is
2026-09-09. Watch how the previous column fills in: the week ending Sep 7 printed 0 as last
issue's newest column and reads 18 this morning. Fade rule, stated plainly: a column is faded when any day
inside it is less than 42 days back, i.e. when it is not yet fully inside the settled window. That fades 5 of the 13 columns,
the same rule and the same count as last issue.
82
6/22
94
6/29
54
7/6
93
7/13
67
7/20
72
7/27
66
8/3
45
8/10
94
8/17
45
8/24
29
8/31
18
9/7
2
9/14
TexasNew Mexicostill filing
Permian rig count — 268 flat wk/wk
Active drilling rigs in the Permian, weekly · latest 2026-09-11 · the basin peaked at
318 rigs in Apr 2024 and has now printed 268 two weeks running, after 267 the two weeks before;
that is up 30 from the February low of 238 and the joint-highest since late June 2025, when the count was
270 (2025-06-27). Not a new high, and the wording is tie-inclusive: a repeated 268 does not move the
anchor, which only moves at 271 · Source: Baker Hughes
Drilling vs completions pace — trailing 30 weeks
Rigs drill the wells; frac crews complete them. Orange is Permian rigs; blue is frac jobs
ended each week from FracFocus. Read the right-hand end of the blue line as unfiled, not idle. The
newest point, 2026-09-11, is 5 frac jobs against 268 rigs — not a basin that stopped
completing wells, but a week almost nobody has filed for yet (newest job end in the extract: 2026-09-09). The
five newest points are all unsettled — the same 42-day rule that fades 5 of the 13 bars on the
weekly chart above. Judge the blue line from the middle of the series. The gap between the two
lines is the DUC build showing up in real time, and New Mexico's regulator now counts 4,657 drilled-and-waiting
wells, up 27 this week after last issue's record +117 (and see the pipeline card for why part of even
this week's small build is a dating artifact). The national frac-spread series remains unsourced
this issue. · Sources: Baker Hughes; FracFocus
Permian rigsFrac jobs/week (FracFocus)
Most active operators
Bar & big number = settled pace (wells/28d from jobs ended 42–126 days back);
gray number = jobs filed in the last 28 days. Diamondback holds the top slot at 36.7 and extends its lead:
that is +5.0 against the 31.7 that last issue's own window recomputes to on this
week's filings (+5.0 against the 31.7 it printed on the day). The other risers in the top ten are
XTO (ExxonMobil) at 25.3 (+1.3) and Mewbourne Oil at 11.3 (+0.7);
the largest faller is Devon (incl. Coterra) at 31.7 (−6.0), and four more are
down, two flat.
Three of the top ten filed nothing at all in the last 28 days — Mewbourne Oil, Apache and Chevron — and Pioneer (2) and COG (3) barely
more. That is filing lag rather than idleness; every one of them still holds its settled place.
Method change this issue: starting this issue Cimarex and Coterra filings are counted under Devon (incl. Coterra),
reflecting Devon's completed acquisition of Coterra (closed May 7, 2026); prior-issue comparators in this issue are
recomputed on the same basis. Last issue printed Devon 22.7 and Cimarex 15.0 separately — 37.7 together, on the unmerged
basis — so the 31.7 that Devon (incl. Coterra) prints this issue, second on the list, is −6.0 like-for-like against the
37.7 that window recomputes to on the merged basis, not a jump from the 22.7 printed under the Devon
name alone. Everyone else is shown as filed on the disclosure, with no ownership roll-ups, so Pioneer and XTO (both
ExxonMobil) still appear separately. Operators marked · / ·· file
moderately / slowly, so recent filings alone understate them. EOG is still absent: it filed nothing again this week,
its settled pace moved to 2.3/28d on the anchor arithmetic alone, and it still does not clear the 5.5 bar. See the
card above.
Diamondback
36.7
13
Devon (incl. Coterra)
31.7
4
XTO (ExxonMobil)
25.3
22
Occidental
23.7
12
Permian Resources
21.0
7
COG (ConocoPhillips)
17.7
3
Pioneer (ExxonMobil)
14.0
2
Mewbourne Oil
11.3
0
Apache ·
11.0
0
Chevron ·
9.7
0
Blackbeard Operating ·
8.7
0
Civitas (SM Energy) ·
8.7
0
Matador
8.3
6
Spur Energy Partners
5.7
0
Avant Operating II ·
5.7
0
Activity by county
Jobs ended in last 28 days (sums to 94). Martin leads again at 17
(it printed 24 last issue), Upton is second at 16 and Lea third at 14 (from 23).
Against last issue's printed counts the biggest fall is Lea (23 → 14) and the biggest
rises are Glasscock (6 → 10) and Midland (2 → 6) — but last issue's counts were a 109-job tile that has since
recomputed to 186, so those moves are mostly which operators filed this week, not where the crews went.
New Mexico takes 28.7% of the basin's filed jobs against
28.5% when last issue's own 28-day window is recomputed on this morning's filings
(53 New Mexico jobs of 186). Last issue printed 32.1% on the day; that gap is filing lag, not a
swing in activity, so the recompute is the comparator to use.
Midland is 6, up from the 2 it printed for two issues running; the county that led this table through most of
the summer is filling back in, as the filing-lag reading of those two prints said it would.
Martin, TX
17
Upton, TX
16
Lea, NM
14
Eddy, NM
12
Glasscock, TX
10
Midland, TX
6
Reeves, TX
6
Crane, TX
3
Andrews, TX
2
Ector, TX
2
Howard, TX
2
Irion, TX
2
Chaves, NM
1
Kent, TX
1
Completion pipeline — New Mexico (work coming to a pad near you)
From NM OCD well records (queried Sep 15), horizontal wells in Lea / Eddy / Chaves / Roosevelt:
8,522 approved permits not yet spudded (the drilling backlog — includes stale, unexercised permits of
all vintages) and 4,657 wells spudded 2024–2026 still carried as "New" — drilled or drilling with no
completion on record, the OCD's closest thing to a DUC list.
Both series calmed down this week. The permit bank rose 33 to 8,522, after last issue's
−49 — the first decline in the series. The DUC proxy built 27 to 4,657, against last issue's
record +117; by spud year that is 2024 +18, 2025 +3, 2026 +6.
Read the 2024 number as a dating artifact, not a build, and read it inside Devon's row. All of it is one filer, the
Coterra Energy Operating name that this issue counts under Devon (incl. Coterra): its Lea County 2024-spud bucket went
86 → 104 while its Lea unspudded stack fell, which is the OCD backfilling spud dates onto old
"New" permits — a 2024 spud cannot be new drilling. On that filer name alone the OCD reads unspudded 680 →
641 (−39, third to fifth on the raw filer list) and DUCs 256 → 274 (+18); strip that out and the basin's DUC build is +9.
EOG's conversion trend paused — unspudded 971 → 970 (−1 after −28), DUCs 852 → 853 (+1 after +37) — and
it filed no completions at all (see the card above). Devon (incl. Coterra), the rolled row the table prints, reads unspudded
2,123 → 2,114 (−9) and DUCs 715 → 734 (+19) on last issue's
families restated on the same basis; that keeps it first on the permit column and second on the DUC column, behind
EOG Resources at 853. Inside it, the Devon Energy Production filer resumed permitting hard: 1,338 →
1,368 (+30, after +12) with its own DUC stack flat (414 → 415, +1); the Coterra filer is the −39 / +18 artifact above; the WPX Energy Permian and Cimarex of Colorado filers did not move. Every well in that second column needs a frac crew and a
drillout. Operators are grouped into display families as in previous issues, with one change: Devon (incl. Coterra) now
includes the Coterra Energy Operating and Cimarex filers as well as Devon Energy Production and WPX Energy Permian
(the Devon roll-up disclosed on the operators card, applied to last issue's figures too);
Permian Resources includes Earthstone and Admiral, XTO includes both XTO entities, Occidental the three Oxy
entities, and COG includes ConocoPhillips Company; Marathon Oil Permian is shown on its own line. Otherwise as
filed with the OCD; parentheses note current owners.
Operator
Permits (unspudded)
Spudded '24–'26, awaiting completion
Top county
Devon (incl. Coterra)
2,114
734
Lea
EOG Resources
970
853
Lea
Permian Resources
750
472
Eddy
Mewbourne Oil
580
583
Eddy
Matador
666
351
Lea
XTO (ExxonMobil)
306
603
Eddy
Occidental
487
228
Eddy
COG (ConocoPhillips)
550
29
Lea
Chevron
266
276
Eddy
Marathon Oil (ConocoPhillips)
239
26
Lea
Spur Energy Partners
154
115
Eddy
Riley Permian
196
20
Eddy
Civitas (SM Energy)
112
62
Lea
Completion pipeline — Texas permit backlog
The Texas permit numbers in this issue are fresh, for the first time since the Aug 10 issue. daf804 is a
monthly file; the RRC's Sep 9 cut (permits issued through Sep 8) replaced the Aug 8 file that the Aug 10,
Aug 17, Aug 24, Aug 31 and Sep 7 issues all ran on. The window advances one calendar month with it, to
Sep 1, 2025 – Sep 8, 2026: 4,891 permits in the Permian counties, 4,184 of them
horizontal (85.5%).
What the freeze was hiding: not much. August 2026, which printed as a 102-permit stub for five issues, is
complete at 397 — flat against June's 408 and July's 395. September's 126 is the new partial column
(through Sep 8, five in-file working days across the Labor Day weekend), so the last bar is a stub, not a cliff.
These totals are not like-for-like with last issue's 4,782: the window moved (August 2025's 308 permits aged out)
and the file filled in, so do not read 4,891 against 4,782 as a change in permitting. Over the months both files cover,
four permits were withdrawn from the master (which is why July prints 395, one below last issue's 396) and June is
unchanged at 408. The county map book, which runs on a trailing-365-day window, grows for the first time in four
weeks: 3,961 → 3,891 → 3,831 → 3,775 → 4,051 permits — the three-week shrink was the static file, as the last three issues said, and it has reversed
on the first fresh cut. Permits lead the frac by six to twelve months.
368
9/25
475
10/25
236
11/25
305
12/25
491
1/26
327
2/26
422
3/26
473
4/26
468
5/26
408
6/26
395
7/26
397
8/26
126
9/26
Permit holders — trailing 12 months
Permits issued Sep 1, 2025 – Sep 8, 2026 by operator, named from the P-5 number
(100% coverage, 4,891 of 4,891). Names are as filed, no ownership roll-ups, with two
editorial exceptions: Anadarko-numbered permits are counted under Occidental, and from this issue the former Cimarex and
Coterra P-5 numbers — P-5 103285 (Coterra Energy Operating Co.) and P-5 153438 (Cimarex Energy Co.) — are counted under Devon (incl. Coterra), the Devon roll-up disclosed on the
operators card (117 permits, eleventh on the list; Devon's own P-5 numbers carry 0 permits
in this window, so the row is the old Cimarex row under its owner's name).
Fresh file, and a method change to disclose. The name on each P-5 number is the majority vote of the
FracFocus filer strings on that number's API list, and that vote is now gated: a label is kept only when the
winning name holds at least half of the votes and at least five of them. The gate was added after two
mislabels were found — an Enron-era P-5 with no permits since 1999 was printing as Summit Petroleum on
three stray votes across its 385 API numbers, and Canan Mowrey Operating was printing as Three Summits on
three of its five votes. Every P-5 the gate cannot name falls back to the RRC P-5 Organization registry
(Aug 21, 2026 vintage), which takes coverage from 4,459 of 4,891 on the vote alone to all 4,891.
The registry names 92 P-5 numbers this window — 128 horizontal permits (those reach the county maps)
and 304 vertical — and they are mostly saltwater-disposal and water-midstream operators (the largest:
Waterbridge Stateline 41, NGL Water Solutions Permian 27, Deep Blue Central 19, Deep Blue 18 and WES Water 18; 124 of the permits carry a water or disposal name outright), companies that never file a frac disclosure, which is exactly why
the vote could never name them. The largest horizontal newcomers are DE V 18, Crescent Energy 15 and Sharp 8.
A voted label is never replaced by the registry. The top ten permit holders did not move: Pioneer (ExxonMobil) 612, Occidental 518, Diamondback 494, COG (ConocoPhillips) 181, Permian Resources 169; lower down the top forty, Waterbridge Stateline (41) and NGL Water Solutions Permian (27) enter and Jetta (21) and Petroplex Energy (21) drop out.
Operator
Permits
Pioneer (ExxonMobil)
612
Occidental
518
Diamondback
494
COG (ConocoPhillips)
181
Permian Resources
169
Ovintiv
157
Double Eagle (DE Central)
157
Apache
142
Continental Resources
125
Blackbeard Operating
125
Devon (incl. Coterra)
117
Kinder Morgan
91
Permit counties — trailing 12 months
Same window, same fresh file. Midland still leads at 660;
Reeves (fourth to second) moves up the order against the frozen table, and the two Delaware counties, Reeves and Loving, hold 959 permits between them against 1,097 for Midland and
Martin. Ranks against last issue are ranks against a different window on a different file, so read them as
where the fresh file puts each county, not as movement.
County
Permits
Midland
660
Reeves
514
Upton
454
Loving
445
Martin
437
Reagan
289
Glasscock
268
Andrews
245
Ector
194
Crane
180
Gaines
179
Howard
158
Smaller operators to watch
Everyone outside the top-10 settled list with 3+ completions since Apr 5 —
the private and mid-cap programs that buy sand, water, and services every month without making headlines.
The window is the anchor date minus 163 days, the same offset every recent issue has used.
Large independents (Mewbourne, Civitas, Matador, Ovintiv, Continental) are excluded regardless of count, and so
is EOG, which would otherwise walk in as a "smaller operator" on a settled pace of 2.3.
One rule tightened this issue: an operator in either this issue's or last issue's top ten is excluded; with
last issue's list restated on the Devon roll-up basis the two lists hold the same ten names, so nobody is excluded on that
ground alone this week.
Green dates = a job ended inside the last ~30 days,
i.e. a crew was just there. Service co. from supplier records where identifiable.
Operator
Jobs (5 mo)
Top counties
Service co. (ID'd)
Median water (bbl)
Newest job
Blackbeard Operating
29
Crane TX
—
119,000
2026-07-23
FireBird Energy II
28
Upton TX, Ector TX
—
501,000
2026-09-03
Fasken Oil and Ranch
22
Andrews TX, Midland TX
—
505,000
2026-08-28
VTX Energy
21
Reeves TX
NexTier (Patterson-UTI)
637,000
2026-08-26
Spur Energy Partners
20
Eddy NM, Lea NM
Halliburton
281,000
2026-08-04
HighPeak Energy
20
Howard TX, Borden TX
Liberty, ProFrac
480,000
2026-07-03
Surge Energy
18
Howard TX, Borden TX
NexTier (Patterson-UTI)
416,000
2026-08-11
Avant Operating II
17
Lea NM
Liberty
341,000
2026-07-17
Triple Crown Resources
16
Irion TX, Upton TX
—
519,000
2026-08-25
BPX
15
Reeves TX, Loving TX
Halliburton, NexTier (Patterson-UTI)
501,000
2026-05-30
TRP Operating
15
Loving TX
—
542,000
2026-07-09
BTA Oil Producers
13
Crane TX, Lea NM
FTS International
716,000
2026-08-12
Basin Oil and Gas
12
Winkler TX, Ector TX
—
4,000
2026-08-17
Blue Arrow
12
Reeves TX, Pecos TX
—
631,000
2026-07-13
Double Eagle (DE Central)
12
Glasscock TX, Reagan TX
Halliburton, NexTier (Patterson-UTI)
559,000
2026-05-07
Ring Energy
11
Crane TX, Andrews TX
—
193,000
2026-08-22
Read the Basin Oil and Gas row with suspicion.
Its median disclosed water is 4,064 bbl across 12 jobs; every other row in this table runs between
about 119,000 and 716,000 bbl, and the rows immediately either side of it are 716,000 and 631,000. That is two
orders of magnitude below anything that could be a modern Permian frac, and on the order
of a fraction of a barrel per lateral foot. These disclosures are almost certainly not describing full stimulation
jobs (a recompletion, a partial filing, or a units error are all likelier than a 4,000-bbl frac). The row is left in
rather than quietly dropped, but it should not be read as a completions programme. The same operator supplies the
top of the fleets-on-location table below, which carries the same caveat.
Completion intensity by operator
Median job design over each operator's trailing 12 months · operators with
≥8 jobs and a disclosure inside the last 6 months, sorted by median sand · sand in millions of lbs, water in thousands of bbl ·
the lateral and lb/ft columns stay blank again this issue. The join was re-checked against the fresh
Sep 12 RRC dbf900 and it is still too thin and too biased to print: median perforated-interval coverage across the
18 rows below is 35.6% (42.9% job-weighted), 4 rows sit under 25%, and among the basin's
biggest programmes XTO joins on 1.2% of its jobs (4 of 321). The wells that do join are the ones whose
operators file promptly, and they skew high: the values that would have printed run 2,077–2,986 lb/ft. That range is
not like-for-like with the 1,995 lb/ft in the sand-per-foot special above — 1,995 is the 2026-to-date,
preliminary, nine-county Midland-Basin median from that card, while these are trailing-12-month, basin-wide,
per-operator medians. Printing them would say more about paperwork than about design. They are retained in the
build data if anyone wants to argue the point. The basin-wide perf join is shown properly in the sand-per-foot
special above, where the history is deep enough to be honest.
One calibration warning about the top of this table. The rows are sorted by median sand, so the least
calibrated rows come first: Petroplex Energy leads on 20 jobs, and without the suppressed per-foot column there is
nothing on the page to say whether a big median sand number is a big frac or simply a long lateral — or, at the
top of the table, a small sample. Treat the first few rows as small-n and uncalibrated until the lb/ft column
can be printed.
Operator
Jobs
Median sand (M lbs)
Median water (k bbl)
Median lateral (ft)
Sand (lbs/ft)
Petroplex Energy
20
38.9
744
—
—
SM Energy
16
36.0
871
—
—
Discovery Natural Resources
15
35.1
1206
—
—
Black Swan
11
34.1
786
—
—
RP Operating
10
33.7
1042
—
—
PBEX Operations
14
31.3
696
—
—
Continental Resources
40
30.5
763
—
—
HighPeak Energy
43
29.8
502
—
—
Pioneer (ExxonMobil)
420
29.6
646
—
—
BTA Oil Producers
39
29.1
618
—
—
Summit Petroleum
31
28.7
547
—
—
Sabalo II Operating
11
28.1
514
—
—
XTO (ExxonMobil)
321
27.1
541
—
—
Blue Arrow
22
27.1
633
—
—
Tap Rock
12
26.3
590
—
—
Double Eagle (DE Central)
67
26.2
587
—
—
VTX Energy
31
25.5
638
—
—
Occidental
413
25.2
468
—
—
Water & sand by well depth (TVD)
Trailing 12 months, both states, by disclosed true vertical depth. The shallow band is a
different business — a third of the sand and just over half the water of every deeper band. Below that line the
design is remarkably flat: from 7,000 ft to past 12,000 ft the median job barely moves, which is the
lateral-length story again rather than a depth story. Four of the five bands move by 17 jobs or fewer as
the 12-month window rolls; the exception is 7,000 - 9,000, up 35 (1,610 → 1,645).
The New Mexico share of the deep bands is not moving in one direction, so this issue does not claim it is:
9,000 - 10,500 34.09% → 34.67%; 10,500 - 12,000 49.64% → 49.94%; over 12,000 ft 62.19% → 61.56% — two of the three up by 0.58 and 0.30 of a percentage point, over 12,000 ft down 0.62. That is noise at this
sample size, not a trend. The Lea/Eddy concentration the county table shows is real; this table is not where you
can watch it move.
TVD band
Jobs
Median water (bbl)
Median proppant (lbs)
of which NM
under 7,000 ft
364
284,000
7,666,000
83
7,000 - 9,000
1,645
517,000
23,484,000
332
9,000 - 10,500
1,650
531,000
25,173,000
572
10,500 - 12,000
817
510,000
24,519,000
408
over 12,000 ft
307
536,000
24,597,000
189
Water & sand by target formation — Texas
Texas jobs ended in the trailing 12 months (3,199), formation from the RRC wellbore join —
1,948 named (60.9%), against 1,933 (60.7%) last issue.
⚠ Method change this issue, and it moves a row. This table has always found a well's target by taking the deepest
formation-top record the RRC wellbore file holds for it. That file stores at most ten formation tops per completion
filing, and the record-04 remarks on the same wells enumerate rows well past ten — so the deepest rows exist on the paper
form and are simply not stored. The deepest target is the last row, so the truncation loses precisely the Barnett and
Woodford; on top of that, about half of the Texas jobs deeper than 11,000 ft have no formation record at all yet. So one
rule was added, ahead of the existing gates: if the well's RRC regulatory field name contains BARNETT or WOODFORD, it is
bucketed there (field names from the Jul 26 production dump). Every other field still falls through to the deepest-record
rule, deliberately — the two channels agree on only 62% of jobs, because SPRABERRY (TREND AREA) is one enormous RRC field
spanning Spraberry, Wolfcamp, Dean and Jo Mill, which is the whole reason the deepest-record rule exists. Effect on this
week's table: Spraberry 796 → 795; Clear Fork 103 → 95; Delaware Mtn 34 → 32; San Andres 24 → 23; Barnett 4 → 18; Woodford 1 → 9; Grayburg 7 → 6; named 1,939 → 1,948, 26 jobs reassigned.
Barnett clears the under-10 rule on its own at 18 and prints below for the first time; its previous printed count
(4, dropped under the rule) was wrong low and is not comparable. Woodford gets its own bucket rather than being folded into
Barnett (different depth, design and geography), sits at 9, below the rule, and prints nothing; Queen holds at exactly
10 and still prints. The 2026 field map is a floor — a Barnett well completed after the Jul 26 dump can only be found by
the old rule — so do not read the row as a trend.
Correction. The Sep 7 issue printed Barnett 73 in its IP90 table and, two sections later, "Barnett (4)" among the
rows dropped from this one — the same underlying data on two different methods (IP90 has always bucketed on the RRC field name),
both in print and neither disclosed as different. That should not have shipped. With this change both sections bucket the
Barnett and Woodford on the same channel; the counts still differ by construction — IP90's 73 is a production cohort of
wells frac'd Jul 2024 – Feb 2026 with 90 days of reported output, this table's 18 is Texas jobs ended in the last twelve
months — and that difference is window, not method.
The blank-formation-name gate introduced last issue is now applied at source in the wellbore parse and moved nothing this week.
The depth > 0 gate still applies (126 of the 2,174 joinable jobs are excluded by it).
Rows with fewer than 10 named jobs are dropped, which this cycle removes Woodford (9), Strawn (7), Grayburg (6), Penn (4).
Strawn (7) and Penn (4) still cannot be reproduced at the 38 and 20 the Aug 17 and Aug 24 issues printed; read those as
artifacts of those builds. The 12-month window is the honest unit for this join: the RRC completion filing trails the frac by
months. New Mexico is excluded (the OCD pool join was not refreshed this cycle, and it carries no Barnett or Woodford pool
anyway). lb/ft is the median of per-well sand ÷ perf span with a 2,000–20,000 ft gate, and
prints "—" where fewer than 10 wells in the row carry a perf record. Because the RRC filing trails the frac,
the named rows still skew toward operators that file promptly.
Formation
Jobs
Median water (bbl)
Median proppant (lbs)
Sand (lbs/ft)
Spraberry
795
548,000
24,375,000
1,999
Wolfcamp
617
561,000
24,314,000
2,258
Bone Spring
214
496,000
24,943,000
2,495
Dean
118
585,000
26,966,000
2,182
Clear Fork
95
462,000
17,593,000
1,693
Delaware Mtn
32
664,000
24,487,000
2,370
San Andres
23
471,000
22,705,000
2,262
Barnett
18
601,000
26,056,000
2,509
Queen
10
640,000
25,644,000
—
Current conditions — fleets likely on location (inferred)
The rule, in full, because it matters more than the number: a pad is an operator plus a
surface location rounded to two decimal places of latitude and longitude (about a 1 km grid); a pad counts as
"likely on location" when its latest job end is within 1.5× that pad's own median gap between job end
dates. There is no minimum-days floor, and the count is not rounded to any floor either.
On that rule 76 pads pass this issue, against the 75 last issue published. But recompute last issue's window on this week's fuller data and the same rule returns 77, so the honest
comparison is a fall of one pad to 76 — not a rule change, and the
75 → 77 part of the picture is late filing.
Now the honest part, and read it before you read the table. 38 of those 76 pads have a median
inter-job gap of more than a year, so on a strict reading of the rule they "pass" on a job-end date that is
itself months old. The freshest of them this week is a Matador pad in Lea County whose own median gap is
628 days. The table below is therefore drawn from the safer list — the 38 pads whose median gap is
under a year — while the 76 headline above refers to the full no-floor list. Those are two different lists and
the difference is 38 pads; read the "pad gap" column before you read the status either way.
Several large operators batch-date a whole pad's wells to one end date, which zeroes their measured gap and
drops them out of the green list entirely — so this view skews toward smaller, steadier programs. Disclosures lag
the frac by up to 30 days; treat this as a floor and a lead list, not a map. The "recently released" rows are the
freshest pads whose own cadence says the crew has moved on — the freshest this week is XTO (ExxonMobil)'s 3-well pad in
Martin County, last job end Sep 6.
Status
Operator
Service co.
County
Wells so far
Pad gap (d)
Last job end
🟢 Likely on location
Basin Oil and Gas
—
Winkler, TX
7
156
Aug 17
🟢 Likely on location
Basin Oil and Gas
—
Winkler, TX
5
189
Aug 13
🟢 Likely on location
Occidental
Halliburton
Martin, TX
6
97
Aug 11
🟢 Likely on location
Ring Energy
—
Ector, TX
4
30
Aug 11
🟢 Likely on location
Surge Energy
NexTier (Patterson-UTI)
Howard, TX
6
155
Aug 11
🟢 Likely on location
Basin Oil and Gas
—
Winkler, TX
8
89
Aug 10
🟢 Likely on location
Surge Energy
NexTier (Patterson-UTI)
Howard, TX
8
92
Jul 31
🟢 Likely on location
Burnett Oil
—
Eddy, NM
15
77
Jul 25
🟢 Likely on location
Blackbeard Operating
—
Crane, TX
6
105
Jul 23
🟢 Likely on location
Burnett Oil
—
Eddy, NM
6
204
Jul 22
⚪ Recently released
XTO (ExxonMobil)
Liberty
Martin, TX
3
0
Sep 6
⚪ Recently released
SEM
—
Irion, TX
2
0
Sep 3
⚪ Recently released
Diamondback
Halliburton
Glasscock, TX
5
0
Sep 3
⚪ Recently released
Occidental
NexTier (Patterson-UTI)
Glasscock, TX
28
0
Sep 2
⚪ Recently released
Occidental
—
Glasscock, TX
3
0
Sep 2
And read the top of the green list with the same
suspicion as the smaller-operators table above. Three of the ten green rows (positions 1, 2 and 6) are
Basin Oil and Gas pads in Winkler County, and Basin's disclosures carry a median water volume of about
4,000 bbl per job — two orders of magnitude below a Permian frac. Whatever those filings describe, it is probably
not a frac crew on location. The rows are shown rather than dropped, because the rule put them there and hiding
them would hide the weakness; but the head of this list is the least trustworthy part of it, and the rule keys on
filing cadence, not on job size.
Frac service companies — who's pumping for whom
Pressure pumper identified on 60.9% of the last 12 months' jobs
(2,913 of 4,783) —
from FracFocus supplier records directly, plus same-pad propagation (a labeled sibling well within the pad
names the crew, matched on a 3-decimal-place pad grid over the full disclosure history). Coverage is
flat like-for-like against last issue's 61.1% (−0.2 pt): the name list did not change this week, so the
movement is jobs rolling through the 12-month window, not method. ProFrac and FTS International are shown
separately, as published — the raw matcher folds FTSI and the old U.S. Well Services spelling into ProFrac,
and splitting them back out gives ProFrac 266 and FTS International 85.
KLX is deliberately excluded: it is coil and wireline work, not a frac fleet. The rows below sum to
3,018 against 2,913 identified jobs, because 105 jobs name more than one pumper and are counted
under each — the same disclosure every recent issue has carried.
"Not identified" jobs are genuinely unlabeled in public data — largely self-sourcing operators and ProPetro's
current work — so shares are a floor, not a census.
Service co.
Jobs (12 mo)
Halliburton
1,303
NexTier (Patterson-UTI)
705
Liberty
581
ProFrac
266
FTS International
85
BJ Energy Solutions
43
SLB
35
Not identified
1,870
Implied frac fleets & drillout demand
Transparent math, two assumptions: a frac fleet turns ~5.6 wells per 28 days
(the basin-average throughput calibrated in the July issue: 398 wells ↔ 71 fleet-equivalents), and a
coil/workover unit drills out ~6 wells per month. Fleet-equivalents = settled wells/28d ÷ 5.6;
drillout units = wells/28d ÷ 6. Method note: too many operators batch-date pad filings for per-pad
cadence to be reliable, so a single basin-average throughput is used — simul-frac heavy operators'
true crew counts run lower than shown, and slower programs' higher. Basin total on this basis:
304.0 wells/28d ≈ 54 fleet-equivalents and ≈ 51 drillout units kept busy, against
55 / 51 last issue. The pace eased 4.0 against the 308.0 we printed then
(down 1.3%), and 9.0 against the 313.0 that the same window recomputes to on this
week's filings — and this week, as last, that easing is mostly a thin filing week rather than a thinner basin. Every figure
here understates EOG, which is not on the settled list at all.
Operator
Wells / 28d (settled)
Implied fleet-equiv.
Implied drillout units
Diamondback
36.7
6.6
6.1
Devon (incl. Coterra)
31.7
5.7
5.3
XTO (ExxonMobil)
25.3
4.5
4.2
Occidental
23.7
4.2
3.9
Permian Resources
21.0
3.8
3.5
COG (ConocoPhillips)
17.7
3.2
2.9
Pioneer (ExxonMobil)
14.0
2.5
2.3
Mewbourne Oil
11.3
2.0
1.9
Apache
11.0
2.0
1.8
Chevron
9.7
1.7
1.6
Blackbeard Operating
8.7
1.6
1.4
Civitas (SM Energy)
8.7
1.6
1.4
Matador
8.3
1.5
1.4
Spur Energy Partners
5.7
1.0
1.0
Avant Operating II
5.7
1.0
1.0
What the wells actually made — IP90 after the frac
Nothing in this section moved, and this issue can say exactly
why. The whole pipeline was rebuilt from scratch again rather than carried forward, and it reproduced every
published number exactly: all 3,357 Texas wells and all 2,843 New Mexico wells came back with identical oil and
gas volumes — zero missing, zero extra, zero mismatched. This week even the cohort is identical: the fresh FracFocus
extract added no well to and removed no well from the 9,882-well cohort, and with both production files frozen a
no-change result is the only possible outcome. It is verified, not assumed.
Both production vintages are stale, and one is now past our own refresh
rule for a third consecutive week. The Texas side runs on the RRC Production Data Query dump of Jul 26 — 51 days,
7.3 weeks old, past the five-week rule this newsletter set itself, for the third issue running. An Aug 27 vintage
exists on the RRC's file-transfer server and could not be retrieved from the build host (a multi-gigabyte file with no
resumable download and a per-call time limit on our side); it needs a different machine, and the rule stands as written.
The New Mexico side runs on OCD C-115 well-level volumes scraped on Aug 5 — now 41 days stale, because the OCD
production route sits behind a Cloudflare human-verification gate; re-probed once this morning and still redirected.
Treat every figure below as a July-vintage read.
⚠ Correction to the Sep 7 issue. That issue printed Barnett 73 in this
section's formation table and, in the water-and-sand formation table, dropped "Barnett (4)" as too small to print — the same
filings under two methods, neither disclosed as different. This table has always bucketed Texas wells on the RRC field name;
the other table used the deepest formation-top record, which the RRC file truncates. As of this issue the other table uses
the field name for Barnett and Woodford too, and prints Barnett at 18. The two counts now differ only by window —
73 here is the production cohort (wells frac'd Jul 2024 – Feb 2026 with 90 days of reported output), 18 there is Texas
jobs ended in the last twelve months — and that is stated on both.
Wells frac'd July 2024 – Feb 2026, 6,200 with a clean 90-day production window
(2,843 New Mexico + 3,357 Texas). New Mexico: true well-level oil volumes from OCD C-115 filings. Texas:
oil + condensate from RRC lease-level filings, restricted to single-well leases — leases carrying exactly one
wellbore on the proration schedule, so lease volumes are well volumes with no pad allocation; that keeps 3,357 of
6,703 TX cohort wells (50%; the rest sit on multi-well leases). Where a well touches more than one single-well
lease, the lease with the largest windowed oil volume is used. All three months of a well's window must carry
a filed row; wells with production predating the frac (refracs, legacy leases) are excluded.
IP90 = oil over the first three reported production months ÷ 90 days. Median across all 6,200
wells: 692 bopd (NM 762, TX 638); median gas 1,250 mcfd.
Formation
Wells
Median IP90 (bopd/well)
bopd per 1,000 ft
Wolfcamp
2390
690
65
Bone Spring
2053
768
63
Spraberry
1209
639
52
Dean
175
726
55
Clear Fork
137
561
50
Barnett
73
602
62
Delaware Mtn
35
576
—
Yeso
35
225
—
San Andres
25
170
21
Penn
17
298
—
Strawn
12
536
51
Queen
11
438
42
Other / unreported
28
374
26
Operator
Wells
Median IP90 (bopd/well)
ExxonMobil (Pioneer/XTO)
1191
700
Devon (incl. Coterra)
622
895
EOG Resources
542
857
Mewbourne Oil
509
625
Occidental
413
997
Permian Resources
364
672
Diamondback
339
745
Chevron
301
611
ConocoPhillips (COG)
258
759
Matador Production
236
622
Per-1,000-ft columns are Texas-weighted: laterals come
from the RRC perf join (3,255 TX wells) and from OCD completion reports for only 62 NM wells, so a dash
means the lateral sample was under 10. These laterals and formations are deliberately held at the Aug 15 wellbore file,
not the fresh Sep 12 one. Filling only the blank laterals from the newer file was tested this week and rejected: it
would move three printed cells (Dean per-foot 55 to 54, Occidental 97 to 95, Diamondback 62 to 63), and the formation half
cannot be refreshed cleanly at all, because 125 of the Spraberry (Trend Area) wells flip bucket on the new file. Refreshing one
half and not the other would mix bases, so both are held until that rule is pinned down.
Operator rows here use ownership roll-ups (unlike the rest of the issue)
because the production files are keyed to current operatorship. Note that EOG sits third on this table at 542
wells and 857 bopd — its production filings never went dark even while its FracFocus paperwork did, and its
paperwork has now been quiet for two weeks.
Who's ramping, who's easing off
Three signals per operator: settled pace (wells/28d, jobs ended 42–126 days back)
against its own prior three-month window (126–210 days back); TX permits issued in the
trailing 8 weeks (Jul 15 – Sep 8) against the prior 8 weeks (May 20 – Jul 14) (intent — leads the frac 6–12 months; arrows
at ±20%); and jobs filed in the last 28 days as a share of settled pace (a filing-freshness read,
not a slowdown detector — slow filers run low here every issue). The universe is every operator on the
settled-pace list above, i.e. 5.5 wells/28d or better.
The permit column is a measurement again. It runs on the fresh Sep 9 permit file, and its two 8-week windows are now
derived from the file's own cutoff (Sep 8) rather than fixed dates. That also means the arrows are not comparable to the ones
printed in the last five issues, which all ran on identical windows ending Aug 7 — the whole window moved. Read them as
this week's intent. On the new windows the permit count is up more than 20% for Diamondback (26 → 72) and Devon (incl. Coterra) (14 → 18), and down more than 20% for eight operators, the largest drops being Pioneer (ExxonMobil) (116 → 81), Occidental (95 → 61) and Chevron (22 → 3).
Read the first column against a filing week of 98 new disclosures; hover a row for the figures.
The "Reads as" rule is unchanged (a filer is current only when its filing-currency flag
reads current AND its filed share is at least 40% of settled pace), so this column is row-comparable to
the last two issues but not to the Aug 24 issue and earlier, which used a filed-share threshold alone.
Operator
Settled pace /28d
TX permits
Filed last 28d
Reads as
Diamondback
· 36.7vs 40.7 prior 3 mo
▲ 728 wk to Sep 8 · vs 26 prior 8 wk
13 (35%)
holding pace; a slow filer, so recent weeks understate it
Devon (incl. Coterra)
▼ 31.7vs 40.0 prior 3 mo
▲ 188 wk to Sep 8 · vs 14 prior 8 wk
4 (13%)
off on paper, and filing slowly — likely lag as much as a real slowdown
XTO (ExxonMobil)
· 25.3vs 27.0 prior 3 mo
—no TX permits in either window
22 (87%)
holding pace, and filings are current
Occidental
▼ 23.7vs 36.7 prior 3 mo
▼ 618 wk to Sep 8 · vs 95 prior 8 wk
12 (51%)
genuinely easing — filings are current, so the drop is real
Permian Resources
· 21.0vs 25.0 prior 3 mo
▼ 158 wk to Sep 8 · vs 27 prior 8 wk
7 (33%)
holding pace; a slow filer, so recent weeks understate it
COG (ConocoPhillips)
· 17.7vs 18.3 prior 3 mo
· 298 wk to Sep 8 · vs 31 prior 8 wk
3 (17%)
holding pace; a slow filer, so recent weeks understate it
Pioneer (ExxonMobil)
▼ 14.0vs 31.7 prior 3 mo
▼ 818 wk to Sep 8 · vs 116 prior 8 wk
2 (14%)
off on paper, and filing slowly — likely lag as much as a real slowdown
Mewbourne Oil
▼ 11.3vs 20.0 prior 3 mo
▼ 108 wk to Sep 8 · vs 13 prior 8 wk
0 (0%)
off on paper, and filing slowly — likely lag as much as a real slowdown
Apache
▼ 11.0vs 18.3 prior 3 mo
▼ 148 wk to Sep 8 · vs 31 prior 8 wk
0 (0%)
off on paper, and filing slowly — likely lag as much as a real slowdown
Chevron
▼ 9.7vs 13.3 prior 3 mo
▼ 38 wk to Sep 8 · vs 22 prior 8 wk
0 (0%)
off on paper, and filing slowly — likely lag as much as a real slowdown
Blackbeard Operating
▲ 8.7vs 2.0 prior 3 mo
▼ 128 wk to Sep 8 · vs 23 prior 8 wk
0 (0%)
pace up on settled filings; permit intake not following yet
Civitas (SM Energy)
▼ 8.7vs 12.3 prior 3 mo
▼ 08 wk to Sep 8 · vs 5 prior 8 wk
0 (0%)
off on paper, and filing slowly — likely lag as much as a real slowdown
Matador
▼ 8.3vs 11.3 prior 3 mo
—no TX permits in either window
6 (72%)
genuinely easing — filings are current, so the drop is real
Spur Energy Partners
▲ 5.7vs 3.0 prior 3 mo
—no TX permits in either window
0 (0%)
pace up on settled filings; permit intake not following yet
Avant Operating II
▲ 5.7vs 0.0 prior 3 mo
—no TX permits in either window
0 (0%)
pace up on settled filings; permit intake not following yet
How current is each operator's paperwork?
Every operator with 8+ Permian disclosures in the trailing 18 months, sorted
worst-first by how old its newest job-end date is. FracFocus has no enforced filing deadline, so this
is a paperwork-currency table, not an activity table — a dark filer can be busy and a current filer can be
idle. Late batches arrive constantly and revise settled pace upward afterward, which is why the lead
recomputes last issue's window every week.
Red = newest job over 120 days old (dark);
amber = 42–120 days (slow);
green = inside 42 days (current).
Split this issue: 22 dark / 22 slow / 24 current across 68 rows, against
last issue's 22 / 21 / 25 across 68 once its Devon and Cimarex rows are folded into one
Devon (incl. Coterra) row, the basis this issue uses (it printed 22 / 22 / 25 across 69 with the two
apart). Most of that drift is the calendar
(4 band changes on the calendar against 2 on new filings).
11 of the 68 operators have a newer job on file than they did last week. Band changes on the calendar alone
(same newest job, seven more days old): PBEX Operations (slow → dark), Apache (current → slow), Civitas (SM Energy) (current → slow) and Continental Resources (current → slow). Band changes on new filings: Ovintiv (dark → current, newest job now 2026-08-12) and Occidental (slow → current, newest job now 2026-09-02).
Operator
Newest job on file
Filed last 28d
Settled pace /28d
Disclosures, 18 mo
OxyRock
2025-08-28 (383d)
0
0.0
86
Rockport Energy Solutions
2025-10-10 (340d)
0
0.0
9
Fourpoint Energy
2025-11-19 (300d)
0
0.0
13
Iron Orchard
2025-12-09 (280d)
0
0.0
8
Upcurve Energy
2025-12-26 (263d)
0
0.0
17
Vital Energy
2026-01-16 (242d)
0
0.0
20
IKE
2026-01-16 (242d)
0
0.0
21
Unitex
2026-02-05 (222d)
0
0.0
8
Hannathon
2026-02-16 (211d)
0
0.0
18
Paloma Permian Assetco
2026-02-25 (202d)
0
0.0
22
Hibernia Resources IV
2026-03-02 (197d)
0
0.0
26
Petro Hunt L L C
2026-03-25 (174d)
0
0.0
56
Marathon
2026-04-04 (164d)
0
0.0
11
Dallas
2026-04-08 (160d)
0
0.0
9
SM Energy
2026-04-12 (156d)
0
0.0
57
Manti Tarka Permian
2026-04-22 (146d)
0
0.0
14
Petroplex Energy
2026-04-26 (142d)
0
0.0
20
Black Swan
2026-04-27 (141d)
0
0.0
22
Kaiser-Francis
2026-05-01 (137d)
0
0.0
15
Double Eagle (DE Central)
2026-05-07 (131d)
0
0.0
116
Discovery Natural Resources
2026-05-09 (129d)
0
0.0
33
PBEX Operations
2026-05-12 (126d)
0
0.0
14
Summit Petroleum
2026-05-25 (113d)
0
1.0
56
Tap Rock
2026-05-30 (108d)
0
0.7
22
BPX
2026-05-30 (108d)
0
1.3
85
Sabalo II Operating
2026-06-15 (92d)
0
2.3
23
3R
2026-06-21 (86d)
0
1.7
24
Crescent Energy
2026-06-22 (85d)
0
1.0
13
Jetta
2026-06-25 (82d)
0
2.3
8
Permian Deep Rock
2026-06-27 (80d)
0
1.7
25
HighPeak Energy
2026-07-03 (74d)
0
5.0
59
TRP Operating
2026-07-09 (68d)
0
3.0
28
RP Operating
2026-07-10 (67d)
0
1.3
10
Chevron
2026-07-13 (64d)
0
9.7
290
Blue Arrow
2026-07-13 (64d)
0
2.7
28
Avant Operating II
2026-07-17 (60d)
0
5.7
17
Strata
2026-07-18 (59d)
0
2.0
22
Greenlake Energy
2026-07-19 (58d)
0
0.3
30
Blackbeard Operating
2026-07-23 (54d)
0
8.7
143
Hilcorp Energy
2026-07-23 (54d)
0
2.3
9
Burnett Oil
2026-07-25 (52d)
0
0.7
13
Apache
2026-07-31 (46d)
0
11.0
220
Civitas (SM Energy)
2026-07-31 (46d)
0
8.7
159
Continental Resources
2026-08-01 (45d)
0
1.0
88
Spur Energy Partners
2026-08-04 (42d)
0
5.7
54
Surge Energy
2026-08-11 (35d)
0
4.0
74
Circle-s Energy
2026-08-11 (35d)
0
0.0
9
BTA Oil Producers
2026-08-12 (34d)
0
2.3
55
Ovintiv
2026-08-12 (34d)
0
0.0
94
Mewbourne Oil
2026-08-16 (30d)
0
11.3
380
Basin Oil and Gas
2026-08-17 (29d)
0
1.7
52
COG (ConocoPhillips)
2026-08-18 (28d)
3
17.7
393
Mack
2026-08-18 (28d)
1
1.3
9
Ring Energy
2026-08-22 (24d)
3
1.7
23
Bayswater
2026-08-22 (24d)
2
0.0
34
EOG
2026-08-24 (22d)
5
2.3
315
Pioneer (ExxonMobil)
2026-08-25 (21d)
2
14.0
712
Triple Crown Resources
2026-08-25 (21d)
3
2.7
66
VTX Energy
2026-08-26 (20d)
3
2.7
42
Fasken Oil and Ranch
2026-08-28 (18d)
2
5.3
83
Devon (incl. Coterra)
2026-08-31 (15d)
4
31.7
627
Occidental
2026-09-02 (13d)
12
23.7
596
Diamondback
2026-09-03 (12d)
13
36.7
774
Permian Resources
2026-09-03 (12d)
7
21.0
393
SEM
2026-09-03 (12d)
2
1.3
23
FireBird Energy II
2026-09-03 (12d)
3
4.0
100
XTO (ExxonMobil)
2026-09-06 (9d)
22
25.3
494
Matador
2026-09-09 (6d)
6
8.3
206
The EOG line, two weeks on. EOG filed zero
disclosures for a second consecutive week; its newest job end is still 2026-08-24, now 22 days old;
its settled pace moved 1.7 → 2.3 purely because the Aug 4 jobs crossed the 42-day line at this
anchor, exactly as last issue's schedule said they would; and its filed-last-28 count is 5. It is still flagged
current, because that flag measures the age of the newest filing and nothing else — it turns amber at the Oct 6
anchor if nothing new arrives. Five of the names are still worse than 250 days —
OxyRock at 383, Rockport Energy Solutions at 340, Fourpoint Energy at 300, Iron Orchard at 280 and Upcurve Energy at 263 — and they are the ones to watch next.
Biggest jobs of the period
Jobs ended in the last 90 days (751 of them), ranked by total base water volume — a wider
window than the 28 days this table once used, because at this filing vintage a 28-day window
holds too few settled monsters to rank honestly. Pad siblings dominate the top of any such list: a four- or
six-well pad filed on one end date will fill several rows with near-identical numbers, which is a batch
completion, not several record jobs. XTO (ExxonMobil)'s Butters pad in Midland County holds 4 of the
twelve rows, followed by Pioneer (ExxonMobil)'s Marsh wells in Midland (4 rows) and FireBird Energy II's Beverly wells in Upton (2 rows). The newest row is XTO (ExxonMobil)'s WINDHAM 17A 201H in Upton, job end Aug 24,
22 days back; the top of the book is older, because a 90-day window rewards the batch pads filed in early summer and
little that filed in the last two thin weeks was large enough to displace them. Several rows disclose no proppant mass,
hence the dashes. Operators are shown as filed, under this issue's no-roll-ups convention, with the one exception the
operators card discloses (Devon (incl. Coterra)).
Operator
Well
County
Water (bbl)
Proppant (lbs)
Job end
XTO (ExxonMobil)
BUTTERS 5A 201H
Midland, TX
1,497,000
74,539,000
Jul 2
XTO (ExxonMobil)
BUTTERS 5C 203H
Midland, TX
1,482,000
74,648,000
Jul 2
XTO (ExxonMobil)
BUTTERS 5D 204H
Midland, TX
1,470,000
74,609,000
Jul 2
XTO (ExxonMobil)
BUTTERS 5B 202H
Midland, TX
1,467,000
74,658,000
Jul 2
XTO (ExxonMobil)
WINDHAM 17A 201H
Upton, TX
1,330,000
—
Aug 24
FireBird Energy II
Beverly 4726 C 501H
Upton, TX
1,152,000
47,228,000
Aug 1
Pioneer (ExxonMobil)
MARSH 36F 106H
Midland, TX
1,148,000
—
Jul 19
RP Operating
Cheddar E 202-176 5H
Reagan, TX
1,146,000
38,281,000
Jul 10
FireBird Energy II
Beverly 4726 E 402H
Upton, TX
1,143,000
47,930,000
Aug 24
Pioneer (ExxonMobil)
Marsh 36 P 116H
Midland, TX
1,127,000
40,166,000
Jul 9
Pioneer (ExxonMobil)
Marsh 36 N 114H
Midland, TX
1,123,000
40,554,000
Jul 9
Pioneer (ExxonMobil)
Marsh 36 O 115H
Midland, TX
1,119,000
40,373,000
Jul 9
Where the permits are — the county map book
Every Texas horizontal drilling permit issued in the trailing 365 days that carries a
surface location — 4,051 permits across 25 counties, one page each, biggest county first. Each permit is
drawn as a line from its surface location toward the permitted bottomhole, so the map shows lateral
direction and length, not just a dot; 4,036 of them, 99.63%, resolve to a line (15 fall back to a gray dot where the RRC
GIS has no wellbore geometry yet — a targeted re-query again found no drilled wellbore for any of them, so a dot is the
correct rendering). Median drawn lateral this issue: 11,248 ft
(5th percentile 1,979 ft, 95th 20,576 ft, longest 23,737 ft). Lines are coloured by
the county's top-eight permit holders, everyone else gray; names are as filed, no ownership roll-ups except the Devon
roll-up disclosed on the operators card (the former Cimarex and Coterra permit numbers draw and label as Devon (incl. Coterra),
one page this week: Culberson (95)). Every plotted permit
carries an operator name: the gated FracFocus vote described on the permit-holders card names most of them and the RRC P-5
registry names the rest — 124 of the 4,051 points, mostly small operators the vote had never seen. Where a
registry-named operator is big enough to rank in a county's top eight it takes a legend colour like any other label,
six pages this week: Glasscock (DE V 18), Dawson (Laguna 8, Lithos 3), Ward (Crescent Energy 9), Winkler (3R 7), Gaines (BCE-Mach 7) and Pecos (MARS Energy 3, Swordfish Energy 2). Tan lines underneath are
Interstate, US and state highways from TxDOT — locational context only, no county boundaries are drawn; all 25 pages carry highway geometry and 111 route shields are labelled.
Permits lead the frac by six to twelve months, so this is a map of where crews will be, not where they are.
This book is drawn on the fresh Sep 9 permit file, so the trailing-365-day window
(Sep 15, 2025 onward, through Sep 8, 2026) is advancing against a moving file again. The point count
grows for the first time in four weeks: 3,961 → 3,891 → 3,831 → 3,775 → 4,051. The three-week shrink was the static file aging out permits with none
arriving, as the last three issues said, and it reversed on the first fresh cut. Permit surface coordinates: the new APIs were
queried against the RRC GIS surface-location layer this morning and the rest carried from the Sep 1 cache; the
bottomhole lines were queried fresh on Sep 15 for all 25 counties.
Highways: the TxDOT geometry stays on its Sep 1 vintage (service verified healthy this morning); three frames grew with the new permit set and were refetched on the unchanged query — Irion (13 → 14 features), Loving (2 → 3) and Ward (10 → 11) — taking the stored geometry to 374 features. Last issue's
correction to the Kent, Sterling and Terry pages stands and all three carry highways again this week.