Summary

  • Apollo-managed funds have agreed to put US$9 billion into a nonvoting Class B interest in a new ONEOK holding company. ONEOK plans to pair that money with US$425 million of cash and commercial paper, buy Brazos Midstream's Permian Midland assets for US$4.425 billion and extinguish approximately US$5 billion of debt.
  • Class B is expected to receive 15% of quarterly cash flow from operations. The applicable capped return is met first; cash above it reduces Apollo's capital account. ONEOK may elect up to 20% under specified conditions, so the account's actual decline matters more than the headline 7.0% cap.
  • No common shares are issued, but common holders have not received costless capital. A new noncontrolling economic claim sits below senior debt and above the residual common interest until distributions or a later ONEOK buyout reduce or remove it.

The extra US$425 million explains the bridge

The announced structure is often compressed into two numbers: US$9 billion from Apollo and US$4.425 billion for Brazos. That leaves the debt programme looking like a residual. ONEOK's presentation supplies the full bridge.

Sources total US$9.425 billion. The Class B investment contributes US$9 billion. Cash held on 30 June contributes US$161 million. New commercial-paper borrowing contributes US$264 million. Uses also total US$9.425 billion: US$4.425 billion for the Brazos assets, US$1.2 billion to repay a term-loan agreement and US$3.8 billion for other debt extinguishment. Transaction costs are excluded from the table.

This is not yet a completed movement of cash. The minority investment is expected to close in the first half of September, subject to customary conditions. The Brazos purchase is expected in the fourth quarter, subject to customary conditions including US antitrust clearance. The useful verbs are therefore “agreed,” “expects” and “intends.” Neither proposed transaction was closed in the source record.

The existing balance sheet gives the scale of the intervention. At 30 June ONEOK reported only US$161 million of cash, against US$1.499 billion of short-term borrowing, US$750 million of current long-term debt maturities and US$30.773 billion of long-term debt excluding current maturities. Total equity was US$23.037 billion. A US$9 billion permanent-equity NCI is consequently a large change in the financing perimeter even though it does not increase the common-share count.

The same filing clarifies the short-term borrowings. US$899 million was commercial paper and US$600 million was drawn on the US$1.2 billion term loan at quarter-end. ONEOK says the remaining term-loan capacity was drawn in July. Repaying the whole facility after the Apollo closing therefore retires a recent bridge as well as part of the reported leverage stack.

A capped return is only the first bucket

The Class B interest is expected to receive 15% of quarterly cash flow from ONEOK, L.L.C., the operating company. The disclosed waterfall applies that cash first to a capped internal rate of return. For the first nine years the cap is 7.0% on the then-current capital account. The target becomes 7.35% in year 10 and rises to a final 7.85% cap in year 15.

Cash above the applicable return does not become an unlimited participation in ONEOK's earnings. It reduces the Class B capital account. ONEOK may, subject to conditions, elect to allocate up to 20% of a quarter's operating cash flow to accelerate that paydown. Conversely, there is no penalty if the quarterly distribution is below the capped return.

These terms resist familiar labels. This is not ordinary debt: there is no stated maturity, missed quarterly distribution does not create the disclosed penalty, and the interest is carried in permanent equity. It is not ordinary common stock: Apollo receives no vote, board seat or uncapped share of growth. And it is not a simple 7% preferred coupon: the disclosed return is an IRR cap on a changing capital-account balance, while the actual cash distribution may be larger because excess cash returns capital.

ONEOK's declining-balance illustration is useful only as a mechanism. It shows the return portion shrinking and the return-of-capital portion growing as the account falls. The bars have no dollar scale and are expressly illustrative. Reading a forecast balance for 2030 or 2035 from their height would manufacture precision the company did not publish.

The observable equation is simpler. Begin with the opening capital account. Apply the contractual return calculation. Observe the quarterly cash distribution. Attribute the return component, then subtract any excess as a capital reduction. The closing balance becomes the next period's base. That sequence, repeated from the filings, will show whether common shareholders are recovering the residual economics quickly or carrying the Class B claim for longer.

“No common equity” is a share-count statement

ONEOK emphasises that it will issue no common shares. That is meaningful: existing owners avoid an immediate increase in the denominator used for earnings per share and votes. It is not the same as saying there is no dilution of economic claims.

Apollo receives a Class B interest in ONEOK Holdings, L.L.C. The holding company owns the Class A interest in the operating company, while ONEOK Inc. controls the Class A interest in HoldCo. The Class B interest is nonvoting, has no board representation or liquidation preference, and carries limited HoldCo consent rights. It is structurally subordinate to ONEOK's senior debt. HoldCo and OpCo boards remain composed of ONEOK executives appointed through ONEOK's board, and OpCo distributions are at its board's discretion.

That allocation separates control from economics. Common holders retain votes and future upside above the Class B cap. Apollo accepts no operating control and structural subordination, but receives a defined channel into operating cash flow until its capital account is reduced or bought out. Senior creditors remain prior to both.

The correct language is therefore precise: no common-share issuance, but a new noncontrolling economic interest. This distinction matters because classifications answer different questions. GAAP permanent equity describes presentation. Rating-agency equity credit describes leverage analysis. Neither answers how much cash will be distributed, how long the claim will remain or whether the acquisition earns more than its total capital burden.

The income statement makes the balance visible

ONEOK expects the investment to be reported as a noncontrolling interest within permanent equity, without hypothetical-liquidation-at-book-value accounting. On the income statement, the company says approximately 7.0% annually—1.75% per quarter—of the remaining investment balance will be subtracted from net income to reach net income attributable to ONEOK.

Payment above that NCI attribution reduces the capital account. The next quarter begins with a smaller base, so the amount attributed away from common earnings should also decline, adjusted for tax, if the mechanism operates as described. This is the central feedback loop: cash in excess of the capped return does not disappear as a higher recurring yield; it lowers the base on which the next attribution is calculated.

That also means a falling Class B balance can create a mechanical improvement in net income attributable to ONEOK even before any new operating synergy. Analysts will need to separate three effects: the earnings produced by the acquired assets, interest savings from retired debt, and the declining NCI attribution. Calling their combined result “accretion” without the bridge would obscure which part came from operations and which part came from financing arithmetic.

ONEOK says the investment has been reviewed with rating agencies and is expected to receive full equity credit. It also targets pro forma 2027 debt-to-EBITDA of approximately 3.25 times. Those are relevant forward assessments, not closing receipts. The first post-transaction balance sheet, debt footnote, cash-flow statement and NCI roll-forward will show what actually entered each category.

US$5 billion of debt reduction is not one tender

The debt programme has several instruments. The cash tender offer covers 20 note series but is capped so the aggregate purchase price does not exceed US$2 billion, subject to ONEOK's right to change the amount and to proration. Acceptance follows disclosed priority levels. The offer is conditioned on the Apollo investment and related reorganisation.

ONEOK separately says it will repay the US$1.2 billion term loan at or shortly after the minority investment closes. It also intends, but is not obligated, to redeem all of its 5.55% senior notes due 2026 and part of its 4.25% notes due 2027, approximately US$250 million in aggregate. The tender announcement explicitly says this expression of intent is not a redemption notice.

The remainder may include other repayments and make-whole calls. Consequently, “US$5 billion debt extinguishment” is a programme objective. The US$2 billion tender is one component, not its synonym. Final accepted principal, cash consideration, premiums, make-whole costs and interest savings will determine the realised deleveraging economics.

The reorganisation is also more than administrative wording. The tender release says a newly formed entity will become ONEOK, L.L.C., while the new top company takes the ONEOK, Inc. name; existing notes are to be assumed by the operating LLC and guaranteed by the new corporation. This puts legal plumbing behind the presentation's HoldCo/OpCo picture. It should be verified in closing documents rather than treated as complete on announcement.

Brazos supplies the operating test

The financing only creates value if the asset and debt outcomes justify its cost. Brazos brings approximately 600,000 dedicated acres under long-term fixed-fee contracts with a weighted-average remaining term above 12 years. ONEOK cited 14 active rigs. After the Cassidy II plant, expected in the third quarter of 2027, it expects about 700 miles of gathering infrastructure and 1.2 Bcf/d of processing capacity. Including plants under construction, the combined ONEOK and Brazos platform is projected at approximately 2.3 Bcf/d.

Management describes the purchase as approximately 7.5 times estimated 2027 EBITDA, including about US$80 million of full-year synergies, and approximately 6.0 times estimated 2028 EBITDA. Those are forecast multiples, not prices divided by reported historical earnings. Cassidy II timing, volumes, synergies, capital spending and integration must all arrive for the denominator to resemble the estimate.

Fixed-fee contracts reduce direct commodity-price exposure but do not eliminate volume, counterparty, construction or basin-development risk. Dedicated acreage and active rigs indicate a commercial perimeter; they do not guarantee every well, every throughput forecast or every schedule. The Class B waterfall will persist regardless of whether a particular synergy appears on time.

The transaction's most defensible claim is therefore structural, not predictive. ONEOK has proposed a way to acquire assets and reduce debt without issuing common shares, while capping a new investor's return and retaining governance. Whether that design transfers enough operating value to common holders will be shown by the Class B account's decline, not by the elegance of the announcement.

Sources