ONEOK Traded a Leverage Target for a Buyback It Hasn't Started
The Raise Was the Cover Story
ONEOK's third quarter was, on the surface, a routine beat-and-raise. Adjusted EBITDA of $2.24 billion, up 6% year over year. Net income of $1.35 billion, or $2.13 per diluted share. And a third upward revision to 2026 guidance, lifting the adjusted EBITDA midpoint to roughly $8.55 billion — $450 million above where the year began. Management walked through the moving parts without flinching: about $50 million of transaction benefits, roughly $75 million of expected 2026 contribution from the newly acquired Brazos assets, offset by $25 million of deal cost, and the remaining $100 million from stronger base operations.
The Brazos integration, closed in early October for the Permian Midland gathering and processing package, is running ahead of underwriting. “Brazos is probably hitting on all cylinders of what we expected to.” — Sheridan C. Swords, Executive Vice President, Commercial · 2026-10-28 Sheridan Swords went further: the current Sundance plants are full, the company is doing offloads, and volumes are above plan for this point in the year.
That is the quarter. It is not the story.
A Philosophy Change Dressed as a Leverage Target
Buried in the CFO's prepared remarks was the most consequential sentence on the call. Walter Hulse reset the capital structure goalpost and explicitly told investors not to model it as guidance.
Our debt-to-EBITDA target going forward is approximately 3.25x or lower and the overall lower matters. This is not guidance It reflects a change in philosophy. ... We now view 3.5x as the upper end of our targeted leverage range.
Seven months ago, the same executive described that 3.5x figure as flexible headroom rather than a hard target: “our 3.5 times target is a self-imposed target. The agencies for our current credit rating would provide us a little bit more flexibility at a higher debt-to-EBITDA than that 3.5 target.” — Walter S. Hulse, Chief Financial Officer · 2026-02-24 The direction has inverted in less than a year. 3.5x has gone from a floor under flexibility to a ceiling that management intends to sit well beneath — Hulse expects to drop below 3x within a few years.
What changed is the balance sheet itself. A $9 billion minority equity investment from Apollo funds closed during the quarter; nearly $5.5 billion of debt was extinguished; S&P and Fitch both upgraded to BBB+ with stable outlooks. A corporate reorganization threw off a $336 million one-time noncash tax benefit, and the Brazos deal pushed the cash-tax runway out a year, to 2032. The company has bought itself the ability to be conservative. The lower leverage doctrine is the receipt.
The prize is capital return optionality, and the current number is small. Repurchases of common stock ran $45 million, down 72% year over year. That is the number to watch if the leverage pledge is real. John Mackay asked the right question — where does run-rate CapEx settle — and got roughly $3.3 billion for 2027, then $2.5–3 billion thereafter. Capital expenditure of $613 million in the latest reported quarter already reflects the wind-down. Falling CapEx plus a lower leverage target equals free cash flow that has to go somewhere.
Data Centers: the Story Got Bigger as the Tape Got Smaller
ONEOK announced two power-generation supply agreements: a 1-gigawatt plant with a 10-year, 250 MMcf/d firm transportation contract due late 2030, and a 500-megawatt expansion taking 110 MMcf/d from May 2028. Swords flagged the pipeline: “we continue to be in discussions with a large number of data centers A lot of them will continue to advance Last quarter we announced 1, this quarter we announced another 1.” — Sheridan C. Swords, Executive Vice President, Commercial · 2026-10-28
This is not new for ONEOK — data centers has been a top-of-stack keyword for the company for well over a year. What is new is the size. In late 2025 these were pitched as bite-sized: “we have been contacted by I mean, well over 30 different projects ... These are not going to be high capital type projects.” — Sheridan Swords, Executive Vice President and Chief Commercial Officer · 2025-10-29
originally, when we go back to that $50 million, that was couple of years ago when we first started seeing opportunities ... that's why the size of the projects have gone up.
Hulse's framing was blunter — the projects once imagined at $50 million are landing at $400–700 million, because hyperscalers want five gigawatts and you cannot pull that off a fully contracted pipe. That is a real change in the gas demand opportunity set, denominated in cubic feet rather than press releases.
The awkward part: the tape has moved the other way. Across the market's price history, the AI data-center complex is up more than 50% over 360 days but down roughly 5% over 90 days, with six advancers against fifty decliners, and data-center-capacity proxies off about 8%. ONEOK is signing contracts into a theme the market has already begun to discount. The announcement will not carry the stock; the contract economics will have to.
What Quietly Left the Slide Deck
Compare the company's own keyword stack to last quarter. The prior period was led by EBITDA growth, PADD 4 product prices, LPG export, the export dock, and "white space" on the existing system. This quarter, none of those appear near the top. The Texas City LPG export terminal — a staple of 2025 and 2026 calls — drew no mention at all. Nor did the Denver/PADD 4 expansion beyond a passing exchange about a next phase that needs customers to sign first.
What replaced them is operational grind: Cassidy 1 and 2, the Delaware NGL expansion into Northern New Mexico, Medford. And hedging, which management went out of its way to emphasise — 75% hedged entering the third quarter, and 2027 layered in more heavily and at higher prices than 2026, precisely because the back end of the curve has risen. That is a company telling you it does not trust its own commodity tailwind enough to leave it unhedged. It also says something about the Northern Delaware expansion: Swords conceded there is real “white space that we have there to be able to leverage and to be able to increase the volume on that system.” — Sheridan C. Swords, Executive Vice President, Commercial · 2026-10-28 They built the pipe bigger than the contracts.
The Arithmetic Underneath
The GAAP picture is less flattering than the EBITDA story, and it is worth sitting with. Total revenue reached $12.0 billion, up 53% year over year, a function of stacked acquisitions. Yet operating margin compressed 4.9 percentage points to 13.2%. Net debt stands at $32.8 billion on an effective net cash basis, better by $1.4 billion across three quarters, with interest coverage of 3.7x. The stock trades at 15.0x trailing net income, near its recent high after a 7.4% ninety-day run.
None of this is alarming. It is the arithmetic of buying growth, and the leverage reset is management's answer to it. But note the tension in guidance: Hulse conceded “we have remained cautious throughout the year just given the whole geopolitical situation. If everything stays like we are today, we are probably on the conservative side.” — Walter S. Hulse, Chief Financial Officer · 2026-10-28 Third raise, still calling itself conservative, with 2027 unguided until February. The setup they did name is specific: Medford's deferred NGL inventory unwinding into earnings, Brazos at a full year, and hedges struck above 2026 levels. On Medford, Swords was explicit — “Medford is going to run full for definitely the first quarter and probably into part of the second quarter when it comes up.” — Sheridan C. Swords, Executive Vice President, Commercial · 2026-10-28
The one loose thread is crude, where volumes looked light. Swords' explanation is the same one he gave a year ago: “That is around our Houston distribution system. Where that is, which is by far the lowest margin, but our gathering volumes are up. Nicely at 7%.” — Sheridan C. Swords, Executive Vice President, Commercial · 2026-10-28 Recurring, coherent, and still requiring a paragraph of defence every quarter.
The verdict: ONEOK spent 2026 buying scale and has now spent the third quarter buying the right to be boring. That is a better trade than it sounds — but it only pays off if the buyback line starts moving.