Energy Transfer LP (NYSE: ET) — Cheapest in the Group, Dearest to Itself: A Deleveraged Franchise Drifting Back Toward Empire-Building

Independent fundamental research. The analysis below carries no investment recommendation and no price target; the sole exception is the explicitly-labeled author’s-view block immediately below.

⚡ Claude’s Take

This block is the author’s own independent opinion and general information only — not investment advice. Everything from the Executive Summary onward is deliberately position-free and price-target-free.

Verdict: HOLD at ~$19. Accumulate-on-weakness sub-$17. Not a short. Conviction: medium. Fair-value zone ~$19–23 (≈8.5–9.5x forward EBITDA / ~7.5–8x P/DCF / a ~6.0–6.9% yield). Own it for the ~6.9%, 1.7x-covered distribution plus 3–5% annual growth — a ~10–12% base-case total return that needs no heroics — not for a re-rating to the WMB/KMI multiple, which requires a structural discount to close that has existed for a decade.

This is the cheapest large-cap midstream franchise versus its peers (≈8.1x forward EV/EBITDA vs WMB ~15x, KMI ~13x, MPLX/EPD ~10x) trading at its own richest-ever valuation (AZI composite 96.8th percentile of its ~20-year history; P/B 1.92x, P/E 14x trailing). Both statements are true, and the tension between them is the thesis. The easy money — the $8-to-$21 quadruple from the 2020 distribution-cut trough as ET deleveraged from ~5.5x to ~3.2x (covenant) and the IDRs came out — has been made. What remains is a genuinely wide-moat, irreplaceable, un-permittable pipeline network (Greenwald scale + captivity + intangible rights-of-way, the broadest wellhead-to-water integration in the group) that nonetheless earns only ~7.7% ROIC ≈ its cost of capital. A wide-moat cash-flow franchise; a mediocre return-on-capital compounder. The frame is value/income, not momentum and not a falling knife — beta 0.60, the stock loads on OilPrice (+0.55) and DividendYield (+0.37) with effectively zero Quality loading, and it has been range-bound $16–21 for eighteen months after the re-rate.

The single most important non-consensus fact cuts both ways. The controlling founder, Kelcy Warren, has bought ~$540M of units in the open market over five years with zero sales and holds an ~$5.7B+ / ~8.8% stake — the most credible alignment signal available — while the governance is overtly insider-favoring (GP control, no say-on-pay, no annual meeting, no ROIC or return metric anywhere in executive comp), the buyback authorization sat idle ($880M unused) at a ~7% yield while ET issued equity for deals, and growth capex is re-accelerating $3.1B → $6.3B into a re-opened spend cycle — the empire-building reflex that over-levered ET into 2020 in the first place. You are betting that Warren’s discipline holds and the data-center/AI gas-demand pull (>6 Bcf/d contracted in a year; Oracle, CloudBurst, Entergy, Oklahoma power loads) earns above WACC. Bull trigger: the “~200 data centers / 15 states” funnel converts to FID’d, high-return contracts and the peer discount finally closes. Bear trigger: the capex ramp funds ~WACC projects into a commodity/volume downturn while the McCrea retirement removes the commercial architect — growth without value-compounding, and the discount persists for good reason.

Tag: “The insider is buying his cheapest-in-group toll road at its own dearest price.”

📈 Stock Price Action — Five-Year Event Map

ET round-tripped from a near-death experience to a four-bagger and then went quiet. The units bottomed at ~$4.56 in March 2020 (COVID demand collapse on a balance sheet already stretched by the debt-funded SemGroup deal and the Bakken/Mariner East build), then — after a controversial 50% distribution cut in October 2020 — climbed almost without interruption to a five-year high of ~$21.08 on 30 January 2025, a ~4.6x advance. Since that peak the stock has been range-bound between ~$16 and ~$21 for roughly eighteen months, closing at $19.17 on 26 June 2026 — about 9% below the five-year high, 52-week range ~$16.21–$20.39, beta ~0.60. The repricing was overwhelmingly a deleveraging-and-re-rating story (EV/EBITDA ~6.5x in 2021 → ~9.4x trailing now), not a price-momentum melt-up; the tape today is a quiet, mid-beta, high-yield range, not a one-way street in either direction.

Cycle narrative. (1–2) The 2020 crash was the defining event: a demand shock exposed a balance sheet that prior empire-building had left exposed, and the dividend cut — reviled by income holders — was the price of survival; Warren stepped in personally and started a buying streak that continues today. (3) The 2021–22 recovery paired cyclical tailwinds with the genuinely value-additive elimination of the IDR/GP structure, which lowered ET’s cost of equity. (4–5) 2023–24 was an acquisition-fueled, deleveraging-driven grind higher, culminating in investment-grade upgrades and the start of the midstream “AI-power beneficiary” re-rate that lifted the whole group. (6–7) The January-2025 high marked the end of the easy re-rate; the December-2025 pullback to the 52-week low coincided with the Lake Charles LNG export suspension and a broader energy/rate wobble. (8) 2026 has been a recovery on a record Q4, a Q1 beat, and a ~$750M guidance raise — leaving the stock fairly valued against itself, cheap against its peers, and waiting on whether the data-center gas thesis is real. (Price moves are Fact; attributed drivers are Interpretation. No price target or recommendation is implied here — that judgment sits in Claude’s Take above.)

1. Executive Summary

Energy Transfer LP is one of the largest and most diversified energy-infrastructure franchises in North America: >125,000 miles of pipeline across natural gas, NGLs, crude oil and refined products, organized into five reportable midstream segments plus controlling stakes in two separately-listed MLPs — Sunoco LP (SUN, fuel distribution and, post-NuStar/Parkland, refining) and USA Compression Partners (USAC). FY2025 consolidated Adjusted EBITDA was a record ~$16.0B (+3% YoY), on which ET generated ~$8.2B of distributable cash flow and paid a distribution that now yields ~6.9% and is covered ~1.7x. It is a master limited partnership — a K-1 issuer, not a 1099 dividend-payer — controlled by its general partner, which is majority-owned by founder Kelcy Warren.

The investment tension is unusually clean. On the one hand, ET owns a genuine, wide, durable moat: the broadest wellhead-to-water integration in the large-cap group, anchored by FERC-certificated interstate gas pipe and Permian-to-Gulf-Coast NGL/crude/export assets that, in a world where new long-haul interstate pipe is effectively un-permittable, are irreplaceable. The balance sheet has been transformed — leverage down from ~5.5x+ pre-2020 to a 3.21x credit-agreement ratio (~4.0–4.5x rating-agency basis) at year-end 2025 — and the value-destroying IDR structure was eliminated in 2018–21. On the other hand, that moat secures cash-flow stability but not franchise returns: consolidated ROIC of ~7.7% sits essentially at the cost of capital, mid-pack among midstream peers (KMI ~5.8%, WMB ~7.8%, OKE ~8%, TRGP ~13%). The empire-building instinct that created the 2020 problem is structurally intact — units are up ~27% in five years on a serial-acquisition cadence (Enable, Lotus, Crestwood, WTG, plus SUN’s NuStar/Parkland), growth capex is re-accelerating from $3.1B (FY23) to $6.3B (FY25) with $5.5–5.9B guided for 2026, and ET bought back zero units in 2024–25 despite an $880M authorization and a ~7% yield.

Valuation captures both facts at once. ET trades at ~8.1x forward EV/EBITDA and ~8.0x P/DCF — the cheapest large-cap midstream (vs WMB ~15x, KMI ~13x, EPD/MPLX ~10x) — yet at its own richest-ever multiple on trailing GAAP measures (AZI composite 96.8th percentile of ~20 years; P/E 98.7th, P/B 95.7th, P/S 96.1th). The persistent peer discount is the market’s standing charge for ET’s governance, complexity, leverage history and ~WACC returns; the bull case is that the data-center/AI natural-gas-demand wave (>6 Bcf/d contracted in the last year) finally closes it. The single strongest alignment signal — Warren’s ~$540M of open-market buying over five years and ~$5.7B+ stake — is the chief counterweight to a governance package that includes no say-on-pay and no return metric in executive pay.

What must be true for the units to work from here is modest and largely already happening: hold leverage near 4x, keep coverage above 1.7x, grow the distribution 3–5%, and earn at least the cost of capital on the re-accelerating build. What would break the thesis is the empire-building reflex returning in earnest — debt-funded capex into sub-WACC projects, or a Permian volume/NGL-spread downcycle — while the controlling-insider governance leaves outside unitholders with little recourse.

2. Business Overview

Energy Transfer gathers, processes, transports, stores, fractionates, markets and exports hydrocarbons across essentially every link of the North American midstream value chain. The asset base — >125,000 miles of pipeline across 44 states (ET’s own materials cite ~140,000 miles inclusive of subsidiaries), ~13.5 Bcf/d of gas processing capacity, 1.15 MMBbls/d of NGL fractionation at Mont Belvieu, and NGL/crude export docks at Nederland, Marcus Hook and Lake Charles — is the physical expression of an integration strategy: own the molecule from the wellhead in the Permian through intrastate and interstate transport, into fractionation, and out the export dock, capturing a fee at each step. The company employs ~16,248 people and is run from Dallas.

Reportable segments and FY2025 Segment Adjusted EBITDA (consolidated Adj EBITDA $15,984M; the named segments sum to ~$16,059M before a small “all other” offset):

This is the most diversified large-cap midstream book in the sector — broader than KMI (gas-pipeline-centric), WMB (Transco-centric) or OKE (NGL-centric). The two FERC-regulated interstate-gas and the NGL/crude transport legs — the most contracted, most stable cash flows — are ~56% of segment EBITDA; the more cyclical G&P, intrastate and SUN retail-fuel/refining pieces (~41%) carry volume and commodity sensitivity. ET states its strategy is to grow the fee-based share, and management/IR cite ~90% fee-based EBITDA — a figure supported qualitatively by the 10-K but not quantified verbatim in it, and one that (unlike WMB’s ~95% take-or-pay) includes percent-of-proceeds processing and SUN retail-fuel margin, so it is somewhat less “demand-charge bond-like” than the headline implies.

Revenue vs. economics. ET’s FY2025 reported revenue of $85.5B is a grossed-up commodity-flow number (much of it pass-through marketing and the title-taking crude/NGL/fuel businesses) and is not the right top line to anchor on. Adjusted EBITDA and distributable cash flow are the economic measures. A second structural subtlety matters for per-unit value: ET consolidates SUN and USAC but owns less than 100% of each, so a large slice of consolidated income leaks to public minority unitholders — FY2025 consolidated net income was $5,708M but only ~$4.43B was attributable to ET common (the difference flows to the ~$14.9B of noncontrolling interests). ET also owns all of SUN’s and USAC’s incentive distribution rights at the top 50% tier, so it harvests an outsized share of their distribution growth — a partial offset.

Verdict: a sprawling, genuinely diversified, predominantly fee-based infrastructure toll-collector. Diversification reduces single-basin/single-commodity risk but also dilutes quality — the SUN retail/refining and G&P legs are structurally lower-return and more cyclical than the FERC pipes, and the consolidated-but-not-wholly-owned structure complicates the per-unit read.

3. Industry Dynamics

Structure. US large-cap midstream is an oligopoly. For integrated long-haul infrastructure the relevant peer set is roughly six names — Energy Transfer, Enterprise Products (EPD), Kinder Morgan (KMI), Williams (WMB), ONEOK (OKE) and MPLX, with Targa (TRGP) the purer Permian player. Pipelines are the only economic mode for land transport of gas, NGLs and crude, so along any given corridor the competitor count is “one hand.” That is the empirical signature of real barriers to entry.

Regulation cuts both ways. Interstate gas pipelines earn FERC cost-of-service returns under the Natural Gas Act (and are exposed to §4/§5 rate cases — ET’s Panhandle has been in a multi-year FERC §5 proceeding since 2019, with refunds ordered and the matter now at the D.C. Circuit); interstate NGL/crude/refined lines run on FERC ICA tariffs (indexed roughly to PPI); intrastate Texas lines are lightly state-regulated and earn negotiated rates. FERC regulation is the central paradox of the industry: it stabilizes interstate cash flow and bars duplicative competing builds (a moat), but it caps the allowed return at “just and reasonable,” structurally limiting midstream ROIC to the high-single-digits-to-low-teens rather than franchise-grade 15–25%. This is precisely why even the best-run names in the group cluster near or below their cost of capital on returns while sustaining wide cash-flow moats.

The un-permittable-pipe thesis. The most important structural tailwind is that new long-haul interstate gas pipe is, for practical purposes, no longer buildable in much of the US: Constitution, Atlantic Coast and PennEast were all cancelled, and Mountain Valley took ~six years and serial litigation to finish. That converts ET’s existing rights-of-way and ~20,090 miles of wholly-owned interstate gas pipe (plus ~7,080 JV miles) into scarce, appreciating assets — a genuine intangible/regulatory barrier to entry that did not exist a decade ago. A more “constructive” federal permitting posture in 2026+ is a potential incremental tailwind but does not reverse the scarcity of existing corridors.

Capital cycle (Marathon lens). Post-2020 the sector shifted decisively from growth/empire-building to capital discipline: capex-to-depreciation fell, free cash flow turned positive, distributions/dividends were brought to conservative coverage, and a wave of consolidation (Crestwood, Magellan, NuStar, EnLink, WTG) was absorbed by the majors. That supply-side discipline is the favorable cycle underpinning the group’s re-rating. ET is the least-disciplined major on this axis: it is re-accelerating growth capex to $5.5–5.9B in 2026, has done four-plus acquisitions in three years, and carries the cohort’s worst legacy reputation (the 2020 distribution halving, the abandoned 2016 Williams merger, the DAPL saga). The Marathon red flags — rising capex/depreciation, debt-and-equity-funded M&A — are more present at ET than at WMB, OKE or KMI.

Verdict: structurally good industry — oligopoly, un-permittable barriers, fee-based stability, a favorable supply-side capital cycle, and a secular demand tailwind from LNG and now data-center power. But it is a good toll-road industry, not a high-ROIC franchise industry: FERC caps the upside. ET sits squarely in that good industry, but as its highest-capex, most-levered, most-diversified — and therefore most quality-diluted — operator.

4. Competitive Position

Name the moat. In Greenwald’s taxonomy ET’s advantage is the strongest category — economies of scale + customer captivity + intangibles (irreplaceable FERC-certificated rights-of-way) — expressed as wellhead-to-water integration. The mechanism is concrete: Permian gas and NGL gathering (Midstream) feeds intrastate transport (ET Fuel/Oasis/HPL, with the ability to bypass processing when frac spreads are unfavorable), which feeds Mont Belvieu fractionation (1.15 MMBbls/d), which feeds the Nederland and Marcus Hook export docks. Owning every link lets ET capture the molecule’s fee multiple times and offer producers a single bundled takeaway solution on dedicated acreage. That integration — materially broader than any peer’s — is a real cost-and-captivity advantage, and it is why ET’s cash flows held through 2020 and every commodity cycle since.

Pressure-test: moat or ~WACC toll road? Both. The moat is real where it protects cash-flow stability, but it does not generate franchise-grade returns. ET’s ROIC of ~7.7% sits essentially at its cost of capital, in Greenwald’s “advantages weak/absent at the return level” band (6–8%), not the 15–25% “advantages present” zone. Three forces dissipate the scale advantage at the return line: (a) FERC caps interstate returns; (b) the G&P, intrastate and crude books are contestable and volume/commodity-cyclical; and © ET carries dead capital from premium-priced or large roll-ups (Crestwood all-equity 2023, WTG ~$3.1B 2024, Lotus ~$1.5B 2023, plus SUN’s NuStar/Parkland), whose goodwill and intangibles weigh on returns. The market-share-stability test passes (share is durable on dedicated corridors); the ROIC test does not.

Versus peers. ET’s footprint is arguably the broadest in the group — the only major with material scale across gas transport and G&P and NGL and crude and retail fuel and compression. EPD is the closest integration analog and is higher-quality and better-rated; TRGP is the purer, higher-return (~13% ROIC) Permian wellhead-to-water play; KMI and WMB are gas-pipeline-pure with cleaner records and higher multiples. ET trades the cohort’s widest footprint for the cohort’s worst reputation on leverage and governance. The 2020 distribution cut is the scar tissue that still caps its multiple and anchors the “management quality” knock — and is the single clearest reason ET trades at ~8x forward EBITDA while WMB trades at ~15x.

Verdict: a genuine, wide, durable moat (scale + captivity + un-permittable rights-of-way; the broadest integration in the group) that secures stable, defensive cash flow — but a regulated/competed toll road earning roughly its cost of capital, not a high-return franchise. The quality is real but diluted by the lower-quality SUN-retail/refining and G&P legs and by a poor capital-allocation reputation. If the moat disappeared tomorrow, cash-flow stability would deteriorate sharply — that is what makes it a moat — but returns would barely change, because they are already only ~WACC.

5. Growth History and Forward Opportunities

History — real, but heavily acquired. Consolidated Adjusted EBITDA grew from $13,093M (FY22) to $15,984M (FY25), a ~6.9% three-year CAGR. But the growth is roughly half acquisition, not organic: Lotus Midstream (Permian crude, May-2023, ~$1.5B), Crestwood Equity (Williston/Delaware/Powder River G&P, Nov-2023, ~$7.1B all-equity), WTG Midstream (Permian gas G&P, Jul-2024, ~$3.1B), plus SUN’s NuStar (2024) and Parkland (2025). Decomposing the FY24→25 organic moves is sobering: Midstream (+$254M) and Interstate (+$108M) grew, but Intrastate (−$145M) and Crude (−$235M) fell and NGL was roughly flat; the headline jump was driven by SUN (+$590M, largely structural/deal) and acquired G&P. Underlying organic growth ex-deals is modest mid-single-digit at best. The five-year unit count rose ~27% to ~3.44B, mostly to fund these stock-and-cash deals — so per-unit EBITDA growth materially trails the headline.

Forward — the Lake Charles pivot to data-center gas. The defining recent change is that Lake Charles LNG export — the decade-long flagship growth option — was suspended in December 2025 on capital-discipline grounds, with ET redirecting capital to a contracted backlog it judges higher-return and remaining “open to third parties” who might develop it (which would still give ET the gas-transport economics). In its place, the genuine new growth narrative is natural gas demand from data centers and power:

• >6 Bcf/d of pipeline capacity contracted with demand-pull customers in the last year (data centers, end-users, utilities), off Desert Southwest, Hugh Brinson and other systems.
• Oracle — long-term agreements to deliver ~0.9 Bcf/d to three US data centers; gas began flowing on the first lateral near Abilene, TX, with two more laterals expected mid-2026.
• CloudBurst (flagship Central-Texas AI data center), Nexus Hubbard (~150 MMcf/d, fully reimbursed capex, in service end-2026), Entergy/Intergic Louisiana (20-year, ≥250k MMBtu/d, lateral upsized with an option to 1 Bcf/d), and Oklahoma power loads (~300 MMcf/d contracted, ~400 MMcf/d in advanced negotiation).
• Management cites discussions with power plants “across 15 states” and connection requests from “~200 data centers across 14 states” — a funnel, not booked revenue.

Sanctioned organic projects carrying the near-term build: Hugh Brinson (400-mi/42-inch Permian egress, 1.5 Bcf/d; ~75% of mainline complete, Phase 1 in service Q4-2026); Desert Southwest (upsized to 48-inch / ~2.3 Bcf/d, ~$5.6B, ISD Q4-2029 — management calls it “one of the better rate-of-return projects we’ve ever built”); Permian processing (Mustang Draw I/II, +550 MMcf/d through 2026); NGL export (FlexPort/Nederland, Frac IX Q4-2026, ethane agreements extended to 2041); and FGT/Springerville gas expansions. Permian volumes rose +8% YoY in Q1’26.

Verdict: mixed-to-low quality growth. The headline EBITDA growth is real but ~half acquisition-driven (stock-funded, per-unit value unproven), and the organic growth — concentrated in Permian G&P and data-center gas — is genuine but commodity/volume-levered and is being bought with a re-accelerating, partly debt-funded $5.5–5.9B capex program. The best single forward leg (Lake Charles export) was just shelved; data-center gas is the real but earlier-stage replacement, and most of it is toll-transport (much of it reimbursable capex), not ET capturing AI/power economics. The crux, per Marathon, is whether this asset-growth outlier of the group earns above WACC on the new build — or merely grows.

6. Financial Quality

Earnings power and trajectory. FY2025 was a record: revenue $85.5B, Adjusted EBITDA $15,984M (+3%), operating income $9.31B, net income to ET common $4.43B (diluted EPS $1.29), and distributable cash flow of ~$8.2B. The multi-year EBITDA progression — $13.1B (FY22) → $13.7B (FY23) → $15.5B (FY24) → $16.0B (FY25) — is a steady up-and-to-the-right, though, as noted, materially acquisition-aided. 2026 is guided to $18.2–18.6B (raised ~$750M on the Q1 print), with management “optimistic” it can reach or exceed the high end.

Margins and returns. As a largely pass-through, title-taking enterprise, ET’s GAAP “gross” and operating margins (operating margin ~10.9%, EBITDA margin ~17.5% on grossed-up revenue) are not the right lens. The meaningful return measures are ROIC ~7.7%, ROE ~13–14%, and return on capital ~8.2% (ROIC source). ROIC ≈ WACC is the single most important financial fact in this report: it says the business is stable and cash-generative but is not compounding economic value above its cost of capital. The trend is roughly flat-to-modestly-improving (ROIC 8.6% in 2022 → 7.7% in 2025, partly diluted by recent-deal goodwill), not inflecting.

Cash flow and capital intensity. Operating cash flow was $10.15B in FY2025. The business is highly capital-intensive — total capex (incl. growth) ran $6,303M in FY25, up sharply from $4,164M (FY24) and $3,135M (FY23). Maintenance capex is ~$1.1B; the rest is growth. Distributions paid were $4.73B to common+GP plus ~$1,691M to noncontrolling (SUN/USAC public) holders. After distributions and the re-accelerating growth build, ET is roughly self-funding its equity needs internally — a structural improvement over the pre-2020 capital-markets dependence — but the growing capex line is consuming the retained excess cash that could otherwise retire units or debt.

Balance sheet. Total debt is ~$70.1B against ~$16.0B Adjusted EBITDA — ~4.4x gross, but the credit-agreement Leverage Ratio was 3.21x at 12/31/2025 (against a 5.00x covenant), and the rating-agency net-debt/EBITDA basis sits in the 4.0–4.5x target band, near the low end. S&P and Fitch both rate ET investment-grade (BBB). This is the transformed, de-risked balance sheet that the 2020 distribution cut paid for, and it is genuine — leverage is down from ~5.5x+ pre-2020. ET continuously terms out the ~$70B stack in the capital markets (e.g., $3.0B senior notes Jan-2026 at 4.55–6.30%, $3.0B Mar-2025, $2.0B junior subordinated Aug-2025), so rate exposure is on the refinancing margin, not a near-term wall; the K-1/MLP structure means there is no corporate-tax shield on that interest.

Quality-of-earnings flags (label them). (i) ET’s reported Adjusted EBITDA ($15,984M) sits above the cleaner ROIC-source EBITDA ($14,994M) chiefly because ET adds back ~$726M of proportionate JV (unconsolidated-affiliate) EBITDA — defensible but a reminder that “Adjusted” is doing work. (ii) The March-2025 Greenpeace/Dakota Access jury verdict (~$667M to ET) is a contingent receivable under appeal, not income — do not capitalize it into run-rate. (iii) SUN’s FY24 results included a one-time ~$586M gain on a West-Texas store sale to 7-Eleven. (iv) A ~$2,645M preferred-unit redemption and the all-equity nature of Crestwood mean the cash-flow “cash paid for acquisitions” lines badly understate deal scale — reconcile via units issued. (v) Net income to common ($4.43B) is far below consolidated net income ($5.71B) because of the large NCI leakage to SUN/USAC public holders — a permanent structural feature, not a one-off.

Verdict: high-quality cash flow (stable, fee-based, well-covered, now investment-grade) attached to mediocre returns on capital (~WACC). The economics are durable but do not improve with scale in any return sense — bigger has meant more EBITDA, not higher-return EBITDA. That is the defining financial signature of the business and the central constraint on its valuation.

7. Capital Allocation

Capital allocation is where ET’s history is ugliest and its recent record most improved — and where the verdict is genuinely mixed.

The 2020 distribution cut and deleveraging (the defining event). In October 2020 ET cut its quarterly distribution ~50% (~$1.22 → ~$0.61 annualized) to force deleveraging — a deeply unpopular call with income holders, made by Warren. It worked: leverage fell from ~5.5x+ to a 3.21x credit-agreement ratio, the balance sheet is now solidly investment-grade, coverage runs ~1.7x, and the distribution has since been rebuilt past its pre-cut peak to a ~$1.34 run-rate. This is the strongest single piece of evidence for the bull “discipline” case. The even-handed bear rejoinder stands: ET only had to cut because it over-levered itself into 2020 (debt-funded SemGroup, the Bakken/Mariner East build into a demand collapse). The cut fixed a self-inflicted wound — recovery from mismanagement, not foresight. The durable structural change is the conservative coverage policy (~1.7–1.8x), more than the cut itself.

M&A — serial, but bought cheap. ET is an unrelenting acquirer: SemGroup (2019), Enable (2021, all-equity), Lotus (2023), Crestwood (2023, ~$7.1B all-equity), WTG (2024, ~$3.1B), plus SUN’s NuStar (2024) and Parkland (2025). Units rose ~27% to ~3.44B and goodwill climbed to ~$5.45B. This is a textbook Marathon asset-growth profile — and asset-growth-heavy names tend to mean-revert. The redeeming nuance: the deals were generally struck at low multiples (Crestwood/Enable ~7–8x vs ET’s own ~10–11x), bolted onto contiguous footprints, with realized synergies ($80M+ Crestwood, $300M+ WTG run-rate) — so unlike a premium-paying roll-up, the per-unit math is plausibly DCF-accretive because targets were bought cheaper than ET trades. The honest read: per-unit value was likely created on the better deals, but the empire-building instinct is structural, the dilution is real, and consolidated ROIC at ~WACC shows the acquired growth has not compounded economic value. Growth, not value-creation.

The buyback tell. ET has a $2B repurchase authorization with $880M remaining — and bought back zero units in 2024 and 2025, despite the units trading at ~10–11x EV/EBITDA and a ~7% yield. Meanwhile it issued equity for deals (WTG units, Crestwood all-equity). Issuing cheap equity to grow while declining to retire cheap equity is a capital-allocation inconsistency; the authorization functions as optics, not policy. The revealed preference is unambiguous: distributions + growth capex + M&A ≫ buybacks.

Governance — red flags, one decisive green offset. Green: the IDR/GP structure — the historic value-leak that skimmed ~50% of marginal distributions — was eliminated in the 2018–21 simplification, a genuine, permanent improvement to ET’s cost of equity and per-unit economics. Red: ET is GP-controlled by LE GP, LLC (majority-owned by Warren), with no annual unitholder meeting, no board election, no say-on-pay, and no DEF 14A (Part III sits in the 10-K). The GP has “absolute discretion” to issue units and set distributions, Warren has anti-dilution protections, fiduciary duties are contractually modified, and there is a standing wall of unitholder derivative/securities litigation (Bettiol; the certified ACERS securities class action). Executive comp is discretionary EBITDA-linked cash bonuses + purely time-vested RSUs with no performance-vesting and — critically — no ROIC, return, or leverage metric anywhere in the plan (a “modified Total Unitholder Return” sets grant-date value only, not vesting). Co-CEOs Long and McCrea each earned ~$20.1M in FY25.

The decisive offset — Warren’s buying. The controlling founder is one of the most persistent insider buyers in US large-cap. Over the trailing five years every reported open-market transaction was a purchase — ~47.5M units / ~$540M+, with zero sales — all at prices below today’s ~$19, on top of his famous 2020–21 buying. His aggregate stake was ~302.4M units (~8.8%, ~$5.7B+) as of the Sep-2024 13D/A and has grown since. This is the most credible possible signal that the controlling insider believes the units are undervalued, and it is the single strongest counterweight to the governance red flags. The skeptic’s caveat — it is also control-consolidating, and a GP founder buying his own vehicle is not the same independent signal as an outside director — is fair, but the magnitude and five-year consistency are extraordinary and fully per-unit-aligned.

Verdict: capital allocation has gone from reckless (pre-2020) to disciplined-but-growth-biased (now). ET earned real credibility — the cut executed, leverage halved, coverage conservative, IDRs gone, deals struck below its own multiple. But the empire-building reflex is intact (units +27%, capex $3.1B → $6.3B, zero buybacks at a cheap price), returns sit at ~WACC, and governance is overtly insider-favoring. It grows the empire intelligently more than it compounds per-unit value — redeemed, for now, by Warren’s own checkbook.

8. Changes and Headwinds — Last Two Years

The thesis-relevant developments of the trailing ~18 months, in order of importance:

• Lake Charles LNG export suspended (Dec-2025). The decade-long flagship growth option shelved on capital-discipline grounds. Net de-risking positive — it removes a large speculative capex/execution overhang — but it also removes the single biggest LNG growth leg; ET reframes it as a gas-supply opportunity and remains open to a third-party developer.
• Data-center/AI gas demand became the new growth narrative. >6 Bcf/d contracted in a year (Oracle ~0.9 Bcf/d flowing; CloudBurst; Nexus Hubbard; Entergy/Intergic Louisiana 20-year; Oklahoma power loads), plus a large prospective funnel. Genuine and high-quality (long-term, investment-grade, demand-pull) but mostly toll-transport, much of it reimbursable capex — not transformative per-unit, and partly still MOU-stage.
• Co-CEO Mackie McCrea announced retirement (Jun-1-2026, by year-end 2026); Tom Long becomes sole CEO. McCrea was ET’s chief commercial/deal-making architect — the growth-project face on every recent call. His exit mid-build-out is a governance/execution negative (no commercial successor named); Warren remains Executive Chairman, so strategic control is unchanged.
• 2026 guidance raised to $18.2–18.6B Adjusted EBITDA (+~$750M) on a Q1 beat, with growth capex raised to $5.5–5.9B — confirming both the demand pull and the re-accelerating spend.
• Dakota Access (DAPL): the March-2025 Greenpeace defamation jury verdict (~$667M to ET) is a contingent asset under appeal, not income. The DAPL easement / Army Corps EIS process remains an unresolved operational tail risk (and was notably not discussed on recent calls). Separately, a successful early-2026 DAPL open season extended base shippers beyond the mid-2030s — a positive.
• Ethane export license requirement (China, Jun-2025). BIS imposed a licensing requirement on ethane exports to certain China-related end-users — a live trade-policy headwind for ET, the largest US ethane exporter; it appears to have eased (export records continued into Q1’26) but remains a geopolitical risk to the NGL-export growth leg.
• SUN/USAC roll-ups continued (SUN/Parkland refining; USAC/J-W Power closed Jan-2026), adding EBITDA but also refining cyclicality and minority-interest leakage.

Verdict: on net, the changes modestly strengthen the near-term cash-flow thesis (guidance up, balance sheet de-risked, Lake Charles overhang removed, data-center demand real) while modestly weakening the quality/governance thesis (capex re-accelerating, McCrea departing, returns still ~WACC). The tape sentiment is mildly positive, consistent with the group’s “AI-power beneficiary” re-rate.

9. Risk Analysis

The dominant risks are not solvency — the balance sheet is the strongest it has been in a decade — but return quality and capital discipline: that the re-accelerating, partly debt-funded build earns ~WACC rather than above it, compounded by governance that gives outside unitholders little recourse if it doesn’t. A Permian volume or NGL-spread downcycle would expose the ~40% of EBITDA that is volume/commodity-sensitive. The K-1 structure is not a fundamental risk but is a real, persistent demand-side constraint on the multiple (it excludes many institutional and all tax-exempt buyers).

10. Valuation Discussion (Embedded Expectations)

ET must be valued as an MLP total-return vehicle — yield + distribution growth + any re-rating — not as a bond proxy and not on GAAP P/E (which is distorted by heavy D&A and the NCI leakage). The right lenses are EV/EBITDA, P/DCF and yield, cross-checked against the stock’s own history and the peer group.

Where it trades (at $19.17). Equity value ~$65.9B; EV (including the ~$14.9B SUN/USAC noncontrolling interest) ~$149.7B. That is:

• ~9.4x trailing / ~8.1x forward EV/EBITDA (on FY25 $16.0B / FY26 guided ~$18.4B);
• ~8.0x P/DCF ($8.2B DCF / 3.44B units ≈ $2.38/unit);
• ~6.9% distribution yield, covered ~1.73x.

Two true statements in tension.

1. Cheapest in the group. On the late-2025 peer snapshot, ET’s ~8x EV/EBITDA compares with EPD ~10.3x, MPLX ~10.6x, KMI ~12.9x and WMB ~15.9x; its ~7% yield is among the highest of the large caps. ET has always traded at the group’s largest discount.
2. Dearest to itself. On trailing GAAP measures ET is at its richest-ever valuation — AZI composite 96.8th percentile of ~20 years (P/E 14.1x / 98.7th; P/B 1.92x / 95.7th; P/S 0.76x / 96.1th) — and EV/EBITDA has expanded from ~6.5x (2021) to ~9.4x trailing. Caveat: the “richest-ever” P/E and P/B are partly artifacts — GAAP EPS is depressed by D&A and NCI, and book value per unit has been roughly flat (~$10 tangible) so a re-rated price mechanically lifts P/B. On the forward MLP-appropriate lenses (~8.1x EV/EBITDA, ~8x P/DCF), ET is fairly-to-fully valued against its own ~11.6x historical EV/EBITDA average and reasonable in absolute terms — no longer the 3–4x deep-value wreck of 2020–21, but not expensive either.

What the price embeds. At ~8x forward EBITDA and a ~7% covered yield growing 3–5%, the market is underwriting continuation, not transformation: that ET sustains ~$18B+ EBITDA, holds ~4x leverage and ~1.7x coverage, grows the distribution mid-single-digits, and earns roughly its cost of capital on the new build. It is explicitly not pricing in (a) a re-rate to the premium-peer multiple, (b) Lake Charles LNG (correctly — it’s suspended), or © the data-center funnel converting at high returns. The persistent ~3–7 turn discount to WMB/KMI is the standing charge for governance, complexity, leverage history and ~WACC returns.

Scenario sketch (illustrative, not a target).

• Bear (~$14–16): a Permian/NGL downcycle or a re-leveraging capex misstep takes EBITDA flat-to-down and pushes the multiple back toward ~7x trailing; the yield does the work but the discount widens. ~Where it traded in late 2025.
• Base (~$19–23): EBITDA compounds to ~$18–19B, coverage and leverage hold, distribution grows ~4%; the stock earns its ~7% yield + ~4% growth with the multiple roughly stable at ~8–9x forward. Low-double-digit total return, no re-rating required.
• Bull (~$26–30): the data-center/AI gas demand FIDs at high returns, ROIC inflects above WACC, and the structural discount to peers narrows (say to ~10–11x forward EV/EBITDA) as ET re-rates toward EPD/MPLX. Requires the market to grant ET “premium-peer” credibility it has never had.

The embedded-expectations read: at ~8x forward EBITDA you are paid a well-covered ~7% to wait, with the discount-closing optionality effectively free — but the bear case (re-leveraging into ~WACC capex) is real, and the reason for the discount has not gone away.

11. Variant Perception

Consensus view. ET is the cheap, high-yield, deleveraged large-cap midstream — a ~7% covered yield with mid-single-digit growth and a free option on data-center gas demand, run by a founder who buys his own units. Sell-side is broadly constructive; the AZI tape frames it as an “AI-power beneficiary.” The factor profile confirms the consensus positioning: ET loads on OilPrice (+0.55) and DividendYield (+0.37) with essentially zero Quality loading and only modest Momentum — it is owned as a yield-and-energy vehicle, not as a quality compounder, and the 0.60 beta and range-bound tape say it is neither a crowded momentum long nor an abandoned falling knife.

Strongest bull case. Irreplaceable, un-permittable footprint + the broadest integration in the group + a transformed, IG balance sheet + a ~7% covered, growing distribution + a controlling founder buying ~$540M of units with zero sales + a genuine secular demand pull (LNG, and now data-center/AI power) that could finally close a decade-old peer discount. If the discount closes even halfway, the re-rating alone is worth 30–50%.

Strongest bear case. A ~WACC business run by an empire-builder whose reflex is returning: capex re-accelerating $3.1B → $6.3B, units up 27% in five years, zero buybacks at a cheap price, and a governance structure (GP control, no say-on-pay, no ROIC in comp) that leaves outside unitholders no recourse. The growth is half-acquired and the organic core is commodity/volume-cyclical; the best growth leg (Lake Charles) was just shelved; the commercial architect is leaving; and the peer discount persists because it is deserved. You’re buying scale, not value-compounding.

The 3–5 assumptions that matter most:

1. Does the re-accelerating build earn above WACC? (Bull: data-center FIDs at high returns. Bear: ~WACC, asset-growth mean-reversion.) — the single decisive question.
2. Does the peer discount close, or is it structural? (Bull: IG balance sheet + demand pull re-rate it. Bear: governance/K-1/complexity keep it permanently cheap.)
3. Does capital discipline hold, or does empire-building return? (Warren’s buying vs. the capex ramp and idle buyback.)
4. How much of the data-center funnel is real (FID’d) vs. MOU? (>6 Bcf/d contracted is real; “~200 data centers” is a funnel.)
5. Commodity/volume cycle — does a Permian/NGL downturn expose the ~40% volume-sensitive EBITDA before the new build ramps?

Falsification. The bull case breaks if FY27–28 ROIC fails to inflect above ~8% despite the capex ramp, or leverage drifts back above ~4.5x to fund it, or coverage falls below ~1.5x. The bear case breaks if ET converts the data-center funnel into FID’d, high-return contracts that visibly lift ROIC above WACC and the EV/EBITDA discount to EPD/MPLX narrows — i.e., the market starts grading ET on returns, not just yield.

12. Fact vs. Interpretation

13. Open Questions

1. Is consolidated ROIC ~7.7% depressed by recent-deal goodwill/integration, or is it steady-state ~WACC? If steady-state, the M&A machine grows without creating economic value — the bear’s core claim. (Needs a returns-on-incremental-capital decomposition.)
2. Of the “>6 Bcf/d contracted” and “~200 data centers / 15 states” funnel, how much is FID’d/binding vs. MOU/LOI? Material to the growth-quality verdict.
3. Who succeeds McCrea as the commercial architect? No successor named in the 6/1/26 8-K.
4. Exact fee-based % and the take-or-pay vs. percent-of-proceeds split (ET cites ~90% but the 10-K does not quantify it; POP/retail-fuel margin dilute the “bond-like” claim).
5. DAPL easement / Army Corps EIS status and the appeal/collectibility of the ~$667M Greenpeace verdict — neither was discussed on recent calls; the easement is an unresolved operational tail risk and the verdict is a contingent asset, not income.
6. Will ET ever actually use the $880M buyback authorization, or does the empire-building reflex keep redirecting cash to capex/M&A even at a ~7% yield?

14. What Must Be True

Bull case — what must be true:

• The re-accelerating $5.5–5.9B/yr growth build earns above WACC — i.e., FY27–28 ROIC inflects above ~8% as Hugh Brinson, Desert Southwest and the data-center laterals ramp.
• Leverage holds near ~4x and coverage above ~1.7x while funding that build; the distribution grows 3–5%.
• The data-center/AI gas funnel converts to FID’d, high-return, investment-grade contracts (not just MOUs), and the structural EV/EBITDA discount to EPD/MPLX/KMI/WMB narrows.
• Falsification: if by FY28 ROIC is still stuck at ~WACC despite the capex, or leverage drifts above ~4.5x to fund it, or the peer discount is unchanged — the bull thesis (re-rating) has failed even if the distribution is fine.

Bear case — what must be true:

• The capex ramp funds ~WACC (or worse) projects, and a Permian volume / NGL-spread downcycle exposes the ~40% of EBITDA that is volume/commodity-sensitive before the new build ramps.
• The empire-building reflex returns in earnest — debt-funded M&A, continued zero buybacks at a cheap price, re-leveraging — confirming the Marathon asset-growth mean-reversion.
• Governance (GP control, no say-on-pay, no ROIC comp, McCrea’s exit) leaves outside unitholders unable to force discipline, and the discount persists.
• Falsification: if ET converts the funnel into FID’d high-return contracts that visibly lift ROIC above WACC and the EV/EBITDA discount narrows — the bear thesis (permanently-cheap ~WACC compounder) has failed.

The two cases share a single fulcrum: does the new build earn above the cost of capital, and does the market ever grade ET on returns rather than yield? Everything else — the ~7% covered distribution, the IG balance sheet, Warren’s buying — is largely common ground.

15. Source Appendix

See Appendix B — Source Appendix for the full primary-source list. Primary sources relied upon: ET FY2025 Form 10-K (filed 2026-02-19) and FY2023 10-K (2024-02-16); ET Q4’25 (2026-02-17) and Q1’26 (2026-05-05) earnings-call transcripts (via ROIC.ai); ET 8-Ks (McCrea retirement 2026-06-03; Q3’25 distribution 2025-10-28; ethane-export license 2025-06-04; senior-notes offerings Jan-2026 / Mar-2025 / Aug-2025); Kelcy Warren Form 4 corpus and SC 13D/A (EDGAR CIK 1276191/1276187); AZI valuation-percentile and news feeds; FactorsToday factor model; ROIC.ai fundamentals; and peer disclosures (KMI, WMB, OKE, TRGP, EPD, MPLX) and public midstream-industry sources for framing.

APPENDIX A — Standard Diligence Questionnaire

Energy Transfer LP (NYSE: ET) — supplemental to the research memo. Fact / Interpretation / Assumption labels applied where it matters. ET is a master limited partnership (K-1 issuer); where a question presumes a C-corp, the MLP analog is given.

General

What thoughtful questions have other investors asked about this company? The recurring debates: (1) Is the 2020 distribution cut a sign of discipline or of the over-leverage that required it? (2) Why does ET perpetually trade at the group’s biggest discount — governance, complexity, K-1 friction, or returns — and will it ever close? (3) Does the serial-acquisition / re-accelerating-capex machine create per-unit value or just grow the empire (ROIC ≈ WACC)? (4) How real is the data-center/AI gas-demand pipeline (FID’d vs. MOU)? (5) What happens to governance and succession given Warren’s control and McCrea’s exit? (6) Should ET convert to a C-corp to broaden its buyer base (management calls it an “option,” no near-term plan).

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? Interpretation: a modest cyclical high-ish, but not extreme — FY25 Adjusted EBITDA was a record and FY26 is guided higher, aided by acquisitions and the post-2020 recovery; ~60% of EBITDA is fee-based/contracted and relatively cycle-insulated, while ~40% (G&P, intrastate, crude, SUN retail/refining) is volume/spread-sensitive and would compress in a Permian/NGL downturn.

Driven by external environment or internal actions? Both — internal (deleveraging, IDR elimination, accretive M&A, organic Permian/NGL build) and external (commodity volumes, NGL/frac spreads, US production growth, LNG/data-center demand).

How stable are revenues? Fact: reported revenue ($85.5B) is volatile because much of it is pass-through, title-taking marketing — not the right stability measure. Adjusted EBITDA is far more stable (the ~90% fee-based core held through 2020).

Outlook for products/services; how big will the market be? Structurally growing US natural-gas and NGL demand (LNG exports, data-center/AI power, petrochemical feedstock, Mexico exports), with crude more mature/plateauing. Domestic and increasingly export-oriented (ethane/LPG to 80+ countries).

Business Quality & Competitive Moat

Is the industry getting more or less competitive? Interpretation: less — consolidation (Crestwood, NuStar, Magellan, EnLink, WTG absorbed by majors) plus un-permittable new long-haul pipe entrench the incumbents.

How profitable is the business (ROIC, ROE)? ROIC ~7.7% (≈ cost of capital); ROE ~13–14%; return on capital ~8.2%. Interpretation: stable and cash-generative but not earning above its cost of capital — the central quality limitation.

How profitable is the industry; barriers to entry? Oligopoly with high barriers (capital intensity, FERC certificates, irreplaceable rights-of-way), but FERC caps interstate returns near WACC — a “good toll-road industry,” not a high-ROIC franchise.

Can the business be easily understood? Interpretation: only partially — five segments plus two consolidated-but-minority-owned MLPs (SUN, USAC), heavy intercompany/IDR structure, K-1 tax, and ~250 legal entities make ET one of the more complex names in the group.

Undermined by foreign low-cost labor? No — physical US infrastructure, not labor-arbitrage-exposed.

Do brands matter? Nature of competition? Switching costs? Brands are largely irrelevant (commodity transport); competition is corridor-by-corridor on the physical network and acreage dedications; switching costs are real where ET has dedicated acreage and integrated wellhead-to-export takeaway, but contestable in G&P at contract roll-off.

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet? Interpretation: the irreplaceable, un-permittable rights-of-way / FERC certificates are carried at historical cost and are worth far more than book — the core hidden asset. The IDRs ET holds in SUN/USAC and the ~$667M Greenpeace contingent receivable are also not on the balance sheet as assets.

Off-balance-sheet liabilities? Unconsolidated-JV debt (proportionate); operating-lease and purchase commitments; environmental/litigation contingencies (DAPL easement; derivative/securities suits). No evidence of aggressive off-balance-sheet financing.

How conservative is the accounting? Interpretation: adequate but “Adjusted”-reliant — reported Adjusted EBITDA adds back ~$726M of proportionate-JV EBITDA and various items; net income to common is far below consolidated NI due to NCI leakage. Read DCF and segment EBITDA, normalize one-time items (preferred redemption, store-sale gain, Greenpeace).

How CapEx-hungry? Fact: very — total capex $6.3B in FY25 (maintenance ~$1.1B; the rest growth), re-accelerating to $5.5–5.9B growth in 2026. This is the defining capital-intensity feature and the locus of the empire-building debate.

Capital Allocation & Management

How much FCF; how is it used; philosophy? OCF ~$10.15B; after ~$1.1B maintenance capex, distributable cash flow ~$8.2B. Used for: distributions (~$4.73B common+GP), growth capex (rising), and M&A — with zero buybacks despite an $880M authorization. Philosophy: distribute + grow capacity + acquire ≫ retire equity.

Significant acquisitions recently? Fact: yes, serially — Lotus (2023), Crestwood (2023, all-equity ~$7.1B), WTG (2024, ~$3.1B), plus SUN’s NuStar (2024)/Parkland (2025) and USAC’s J-W Power (2026). Units +27% in 5 years.

Buying back shares? Fact: no — zero units repurchased in 2024–25.

Issuing large amounts of new units to insiders? Equity is issued mainly to fund acquisitions (Crestwood/WTG), not as insider largesse; SBC is tiny (~$148M). Warren’s stake grows via his own open-market purchases, not grants.

Compensation policy? Fact (red flag): discretionary EBITDA-linked cash bonuses + purely time-vested RSUs, no performance-vesting and no ROIC/return/leverage metric; a “modified TUR” sets grant value only. Co-CEOs ~$20.1M each (FY25). No say-on-pay (GP-controlled MLP).

Motivations of management? Interpretation: Warren (Executive Chairman, ~8.8%/~$5.7B+ stake, persistent buyer, no salary in the comp table) is paid through distributions on his units — strongly per-unit-aligned; the comp structure for the co-CEOs, by contrast, rewards size/EBITDA and tenure, not returns. Net: alignment via ownership is strong; alignment via comp design is weak.

Valuation & Market Data

ADR, MLP, or K-1 issuer? Fact: a master limited partnership — issues a Schedule K-1, not a 1099-DIV. Not an ADR. Distributions are largely return-of-capital (tax-deferred, reduce basis, with ordinary-income recapture on sale). UBTI risk makes it generally unsuitable for IRAs/tax-exempt holders above ~$1,000/yr — a structural constraint on the buyer base and a contributor to the persistent valuation discount.

Dividend (distribution) policy? ~$1.34/unit run-rate (~6.9% yield), target growth 3–5%/yr, coverage ~1.7x, leverage target 4.0–4.5x.

How profitable is the business? See ROIC ~7.7% / ROE ~13–14% above — cash-rich, return-modest.

Is net income diverging from cash from operations? Fact: yes, structurally — OCF ($10.15B) far exceeds net income to common ($4.43B) because of heavy D&A and NCI leakage; this is normal for a capital-intensive MLP and is why DCF, not EPS, is the right metric.

Risks & Downside

What would cause the stock to decline? A Permian/NGL volume or spread downcycle; a re-leveraging capex misstep; a distribution disappointment; rising long rates (yield-vehicle de-rating); a DAPL/regulatory shock; or the market simply re-widening the discount on renewed empire-building.

Risk of catastrophic loss? Low — diversified, IG, cash-generative infrastructure. A single pipeline incident or adverse DAPL ruling would dent but not break it.

Chance of a total loss? Very low absent gross mismanagement — the asset base and IG balance sheet make a zero implausible.

Recent News & Events

Has the business environment changed recently? Fact: yes — Lake Charles LNG export suspended (Dec-2025); a new data-center/AI gas-demand growth narrative (>6 Bcf/d contracted; Oracle/CloudBurst/Entergy/Oklahoma); 2026 guidance raised ~$750M; Co-CEO McCrea retiring by end-2026; an ethane-export (China) licensing requirement; and continued SUN/USAC roll-ups.

Significant acquisitions / accounting changes / new markets? Acquisitions ongoing (above); no adverse accounting-policy changes identified; new “markets” are data-center/power gas supply and expanded NGL/ethane export (Nederland to 2041).

APPENDIX B — Source Appendix

Energy Transfer LP (NYSE: ET) — research initiation, as-of 2026-06-27. Primary sources before secondary; every material claim in the memo traces to an entry below or to a research-log entry. All sources are public primary sources (SEC filings, earnings calls, public data).

Primary — SEC filings (EDGAR, CIK 0001276187; EDGAR)

Primary — earnings-call transcripts (via ROIC.ai MCP)

Quantitative data sources

Frameworks applied

• Greenwald & Kahn, Competition Demystified — moat taxonomy (scale + captivity + intangibles); market-share-stability and ROIC tests.
• Marathon / Chancellor, Capital Returns — supply-side capital-cycle and asset-growth-anomaly lenses on the re-accelerating capex and serial M&A.