Energy Transfer (ET)

Largest EBITDA, Smallest Share of It

Energy Transfer files the biggest adjusted EBITDA of any large midstream partnership, and a good deal less of it reaches the units you would buy than the headline suggests. Consolidated adjusted EBITDA was $15,984 million in FY2025. Distributable cash flow, which is the partnership's own measure of cash available to unitholders and has nothing to do with a discounted cash flow, was $10,615 million on the same consolidated basis but $8,202 million once the parts belonging to other investors come out. That covered the $4,555 million paid to partners about 1.80 times, better cover than Enterprise Products at 1.7× or MPLX at 1.4×.

The ratio is the flattering number. Cash left over after every dollar of capital spending and every distribution paid was negative $2,569 million for the year. Both things are true at once, and the rest of this page is mostly about why.

Business Overview

The asset base spans natural gas gathering, processing and transmission, crude and NGL pipelines, and Gulf Coast export terminals, built up through a long run of acquisitions rather than one franchise system. NGLs are natural gas liquids: ethane, propane, butanes and natural gasoline, stripped out of the gas stream and sold separately. Natural gas assets produced about 40% of FY2025 adjusted EBITDA, and roughly 90% of it came from fee-based margins rather than from owning the commodity.

The consolidated accounts include two separately listed partnerships. Energy Transfer runs the general partner of Sunoco LP and of USA Compression, and consolidates both in full while owning about 15% of Sunoco's common units and 32% of USAC's. Those two segments earned $2,661 million of FY2025 adjusted EBITDA, a sixth of the consolidated total, most of which belongs to somebody else's unitholders.

That is what the $2,413 million gap between consolidated and partner-level distributable cash flow is made of, and the filing shows it line by line: $1,263 million of Sunoco cash flow deducted against $286 million of Sunoco distributions actually received, $386 million of USAC deducted against $97 million received, and $1,153 million belonging to minority partners in joint ventures such as Bakken, Rover and Permian Express. A little under a quarter of the cash the group generates never gets as far as the common units. Any per-unit work starts from $8,202 million, not from the headline.

How the Economics Work

Coverage is distributable cash flow divided by the distributions paid to partners, and Energy Transfer leaves you to do the division: unlike Enterprise, it does not print the ratio. On FY2025 figures it is $8,202 million over $4,555 million, or 1.80 times. On a market capitalisation of roughly $65.6 billion in June 2026, those distributions were a yield of about 6.9%, inside the 5% to 8% range large-cap MLPs typically trade on.

Above 1.4× is where an MLP distribution stops looking tight, so 1.80× reads as comfortable and the yield reads as generous for the cover behind it. Neither reading survives contact with the capital budget. Take the fee-based share with the same caution: 90% is measured against adjusted EBITDA, while Enterprise's 82% is measured against gross operating margin and Kinder Morgan's 96% against a budget. Three different denominators, so the percentages cannot be ranked against each other. The contract quality guide works through what sits behind a fee-based label.

What 1.80× Coverage Leaves Behind

Distributable cash flow is already struck after maintenance capital, interest and preferred distributions, so what remains after the common distribution is the money available for building. At Energy Transfer that was $3,647 million in 2025. Growth capital that year ran to roughly $4.9 billion, and the 2026 budget is $5.6 billion to $5.9 billion. The partnership with the best coverage ratio of these three is the one that has to raise the difference from debt or from asset sales.

Enterprise Products is the contrast, on a worse ratio. It kept about $3.2 billion after its 2025 distribution against a 2026 growth budget of $2.9 billion to $3.4 billion, so it broadly builds out of its own cash flow. Coverage of 1.7× that funds the programme is a stronger position than coverage of 1.80× that does not, which is the whole reason the ratio should never be read on its own.

The ratio has a soft spot of its own. The line between maintenance capital and growth capital is drawn by the company, and no auditor tests where it fell. Move a compressor rebuild from one column to the other and distributable cash flow rises, coverage improves, and the cash left the building either way. Energy Transfer's split does not look stretched: maintenance capital of $1,316 million is 8.2% of adjusted EBITDA, against roughly 6% at Enterprise and 3.6% at MPLX. Nor is the ratio fragile. Double the maintenance charge and coverage still comes out near 1.5×.

The figure that cannot be reclassified is the cash statement. Operating cash flow of $10,149 million, less $6,203 million of capital expenditure, less $4,725 million distributed to partners and $1,790 million to minority holders, leaves negative $2,569 million, and that is before the $2.0 billion Sunoco paid for Parkland. Long-term debt rose from $59.8 billion to $68.3 billion over the year. A partnership can print coverage comfortably above 1.0× for years while doing exactly this, and the two figures only mean anything read together.

Valuation Framework

Relative screens start with EV/EBITDA, and Energy Transfer is the name where enterprise value has to be assembled carefully. Market capitalisation plus debt is not enough: the consolidated debt includes borrowings at Sunoco and USAC, and the minority partners carry $14.6 billion of book equity in the consolidated subsidiaries. Leave them out and you are dividing a whole group's EBITDA by a partial claim on it, which makes the units look cheaper than they are.

For anchors, the Alerian MLP Infrastructure index screens at 8.57× forward EBITDA and the broader Alerian Midstream Energy Select index at 10.94×. The gap is not a reward for being a corporation rather than a partnership: the two indices hold different companies with different assets. Large-cap midstream generally works in an 8× to 10× band, near the 9.0× trailing median of the Wells Fargo universe, and the EV/EBITDA by asset type guide explains what moves a name within it.

Then cross-check against leverage. Management targets 4.0× to 4.5× on the rating agencies' methodology and says it currently sits in the lower half of that, which is a range rather than a number and covers a lot of ground at this size. Set it against Enterprise's 3.0× ± 0.25×: the higher target is a deliberate choice to run the balance sheet harder, and it is the same choice that shows up as the funding gap above.

What to Watch in the Financials

Partner-level cash against the distribution, then against the budget. Recompute coverage each quarter from distributable cash flow attributable to partners, then subtract the distribution and compare what is left with the growth capital being spent. The first number tells you whether the payout is safe this quarter. The second tells you how much of next year's building is coming from the market.

The maintenance capital line over several years. Coverage improving in a year when maintenance capital fell as a share of EBITDA is a footnote worth opening. The distributable cash flow versus free cash flow guide explains why this one line moves the partnership metric and not the GAAP one.

Acquisition cadence and what it consolidates. Deals can lift consolidated EBITDA far faster than partner-level cash, because a purchase made inside Sunoco or a joint venture arrives in full at the top and only partly at the bottom. Sunoco's Parkland acquisition is the live example. Watch what each deal does to the gap between the two distributable cash flow figures, not just to the EBITDA headline.

Natural gas volumes. About 40% of EBITDA sits in gas gathering, processing and transmission, so throughput follows LNG export demand, which the EIA puts at 15.1 Bcf/d across 2025 and forecasts at 17.0 Bcf/d for 2026, and increasingly follows power demand from data centres in Texas and the Southwest. Fee-based contracts do not make volumes irrelevant. They only bind where a minimum volume commitment obliges the shipper to pay for capacity it does not use, and then the protection is only as good as that shipper's credit.

Peer Context

Enterprise Products is the cleaner template on every axis except size: 1.7× coverage it publishes itself, 3.3× leverage with the hybrid adjustment spelled out, 82% fee-based gross operating margin, and a growth programme its retained cash roughly covers. Energy Transfer is larger and requires more work to reach a comparable figure.

MPLX is the opposite trade-off. Its 1.4× coverage sits on the floor rather than above it, but the structure is legible: one sponsor, one set of commercial relationships, and a gathering footprint you can map. Complexity and cover move in opposite directions across the two names.

Williams is the C-corporation alternative, paying dividends reported on a 1099-DIV and using available funds from operations, $5,858 million in FY2025, where the MLPs use distributable cash flow. It competes for the same gas-infrastructure capital while solving a different tax problem for its holders. Compare multiples with structure in mind rather than EBITDA alone.

Key Risks

Consolidation obscures the claim. Every headline figure on this partnership, EBITDA, debt, distributable cash flow, capital spending, is struck across entities whose economics belong partly to other people. Modelling consolidated cash flow as if it were available to the common units is the standard error here, and it flatters the answer by roughly a quarter.

Leverage at the top of the MLP range. A 4.0× to 4.5× target sits above Enterprise's 3.0× ± 0.25×, and the partnership funds a large building programme from outside its own cash flow. Both are manageable while EBITDA grows. Together they leave less room than the coverage ratio implies if it stalls.

Reinvestment risk. The 2026 budget of $5.6 billion to $5.9 billion has to convert into fee-based EBITDA roughly on schedule, on projects including the Hugh Brinson and Desert Southwest pipelines. Delay widens the gap between earnings power and cash available to distribute, and the gap is already being funded with debt.

Structure and tax. A unitholder receives a Schedule K-1 rather than a 1099, usually after the filing season has begun. Much of the distribution is return of capital, which is not taxed on receipt but reduces cost basis and is recaptured as income on sale, so the tax deferred is not tax avoided. Unrelated business taxable income makes an MLP awkward inside a US retirement account, and non-US holders face withholding on distributions and on sale proceeds. The MLP versus C-corp guide sets out the mechanics.