Built by Grok 4.6 High

KMI | Earnings Review

Reported October 21, 2026. Prepared October 26, 2026. The quarter is still growing, and the growth rate is rolling over. Adjusted EBITDA was $2,180 million, +9.5% versus $1,991 million. Adjusted EPS was $0.35, +20.7% versus $0.29. Revenue was $4,531 million, +9.3% versus $4,146 million. EBITDA +9.5% and adjusted EPS +20.7% are faster than last Q3 (+5.9% and +16.0%). Revenue +9.3% is slower than last Q3's +12.1%. The last three prints are not: EBITDA +17.7% to +11.5% to +9.5%, adjusted EPS +41.2% to +32.1% to +20.7%, revenue +13.8% to +10.8% to +9.3%. Natural gas adjusted EBDA, the core, slowed to +6.1% from +17.2% in Q1. Fee-service revenue stayed in a +5.6% to +7.6% band. Commodity sales, not the pipes, did most of the revenue deceleration.

Double beat, shrinking surprise. Versus FMP, adjusted EPS beat by +$0.029 / +9.2% ($0.35 versus $0.3206) and revenue beat by +$157 million / +3.6%. It is the fourth straight beat on both lines. Last-twelve-quarter adjusted-EPS beat rate is still 4/12 (33%). Inside the streak the EPS surprise went +21.2% to +15.4% to +9.2%. Beat magnitude is deteriorating. FMP EBITDA is not Kinder Morgan adjusted EBITDA and is not scored. GAAP diluted EPS is $0.34, +21.4% versus $0.28. The transcript's "$0.24" does not reconcile to net income of $764 million or to the +21% the same sentence claims. Use the filing.

Third raise, and it is a floor that gives Q4 back. The printed FY2026 budget is still adjusted EBITDA $8.6 billion and adjusted EPS $1.36. The call moved the verbal to more than 6% and more than 13% above that budget: floors above $9.12 billion and $1.54. That is about +$86 million and +$0.01 versus the Q2 floors, the smallest step of the three raises. Nine-month EBITDA of $6,918 million already banks the year. The residual Q4 floor is above $2,198 million, about −3% versus Q4 2025 EBITDA of $2,271 million, and adjusted EPS above $0.34, about −14% versus $0.39. About $225 million of 2026 outperformance may not recur and about $175 million is commodity versus a $60 WTI budget. Together that is $400 million of a "well more than $515 million" overage. FY2027 was refused. FMP FY2027 EPS is $1.553, flat versus FY2026 $1.560. Street and management agree that 2026 does not annualize. They do not agree on — because the street is not asked to model — the backlog.

What the street is not capitalizing. Project backlog is $14.4 billion, up from $9.6 billion, +54.8% versus $9.3 billion a year ago. The release puts a 5.7x multiple on a $13.7 billion take-or-pay slice, about $2.4 billion of first-full-year project EBITDA, in service from 2027 through 2032 (Trident only 30% of contracted EBITDA in 2027; Permian Link and Southeast Connector 2030-32). The call applies 5.7x to the whole $14.4 billion and says $2.5 billion. That is a definition break, not a new run-rate. Quarter-end net debt / adjusted EBITDA was 3.5x. Year-end is expected at 3.6x versus a printed budget of 3.8x.

Hold, because the rate is falling and the unmodeled EBITDA is not this year's. Guidance accuracy still cuts for management: each 2026 floor has been cleared and then raised. The live position is pipes, not power. The trajectory that dominates is deceleration in the fee engine's growth rate, a shrinking beat, and a Q4 floor that does not require growth. The contrarian piece — demand-pull utilities and an unnamed A-rated hyperscaler still signing 20-year take-or-pay at higher steel and compression costs — shows up in 2028-32, which management already said will be uneven. Internal SharePoint, OneNote, Outlook, Excel, and Teams were unavailable. Visible Alpha and Bloomberg were not connected.

Kinder Morgan is a fee-based midstream business. It does not report gross margin. Gross margin here is revenue minus income-statement costs of sales, so it falls when pass-through commodity sales outgrow fees. Adjusted EBITDA margin is the profitability lens. Adjusted segment EBDA is the operating driver. Growth is same-quarter year over year. Eight reported quarters are in the tables; the Q1 2026 peak and the two step-downs after it are the slope.

Q3 gross profit is $4,531 million minus costs of sales $1,566 million, or 65.4%, versus 66.4% a year ago (−92 bps). Adjusted EBITDA margin is 48.1% versus 48.0% (+9 bps). Margin expansion has faded from +173 bps in Q1 to +33 bps in Q2 to +9 bps now. All four adjusted segments still grew. The mix flipped: gas cooled, and products, terminals, and CO2 carried more of the growth, with commodity price in the last three.

Services have not left a two-point band since 2024Q3. Commodity year-over-year cooled from the low-20s to +13.1%. About 88% of the shrink in the year-over-year revenue dollar gain from Q1 2026 to Q3 2026 is commodity sales.

Adjusted segment EBDA is the bridge behind "all four segments grew." Do not read GAAP segment EBDA this quarter: GAAP products is $172 million against adjusted $313 million, and GAAP terminals is $385 million against adjusted $314 million. The products "collapse" is certain items. Terminals adjusted EBDA is missing in Daloopa for Q1 and Q2 2025, so those cells are blank.

Natural gas transport volumes were 49,273 BBtu/d versus 47,461, +3.8%, matching the call's +3% (Q2 was +7%). Reported gathering was 4,722 BBtu/d versus 4,380, +7.8%, which does not foot to the call's +14% gathering and KinderHawk +40%. Use the call for Haynesville. Do not treat the Daloopa gathering line as that statistic.

Five completed years. FY2026 is open. Annual adjusted EBITDA growth into 2025 was a slow acceleration (+0.6%, +5.0%, +5.7%). 2026 year-to-date EBITDA is +13.0%, and the last three quarters are giving that rate back. 2022 is the lesson: revenue $19,200 million was +15.6% and adjusted EBITDA $7,516 million was −5.4%, because the revenue was commodity pass-through. Services grew about +6.5% in both 2024 and 2025. The 2025 revenue re-acceleration was commodity.

Double beat versus the street, fourth in a row, pattern Mixed. Rule: beat if surprise is above +1.0%, miss if below −1.0%, in-line otherwise. Street EPS is adjusted EPS. FMP epsActual of 0.35 matches $0.35, not GAAP $0.34. Revenue actuals are Daloopa, not the FMP revenue-actual field.

The quarter was also ahead of the internal budget: the CFO said EPS 15% above budget and adjusted EBITDA 6% above budget. Implied budgets are about $0.304 and $2,057 million (calculated, not tagged). The street EPS bar of $0.3206 was already about 5% above that implied EPS budget, so roughly a third of the "15% above budget" was in consensus before the print.

Heatmap, last eight quarters. This quarter is underlined. B beat, M miss, IL in-line. Surprise is Daloopa actual versus FMP.

Last twelve quarters (2023Q4 through 2026Q3): adjusted-EPS beat rate 33% (4 beats, 5 misses, 3 in-line). Revenue beat rate 50%. Last four quarters are 4/4 on both. The regime flipped from miss to beat at 2025Q4. From the Q1 2026 EPS blowout the surprise has compressed two quarters in a row. Revenue beat size on the six-quarter run is +5.6%, +4.2%, +4.4%, +4.3%, +6.0%, +3.6%. This quarter is the low end, not a break. Five straight revenue misses in 2023Q4-2024Q4 were mostly the street's commodity-sales model, not a fee-cash breakdown. Adjusted EPS is the metric that should drive the call: a recent beater with a shrinking beat, not a twelve-quarter compounder of surprises.

Quality of the beat: natural gas was volume, rate, and projects (transport +3%, and the call's KinderHawk +40%). Products, terminals, and CO2 explicitly include commodity price. EBITDA margin was flat. Diluted shares 2,227 million versus 2,224 million are not a buyback beat. GAAP and adjusted EPS are $0.01 apart, so this is not an adjustments story. The full-year budget overage they are willing to size is heavily non-recurring and commodity. That is why the shrinking quarterly surprise is the more honest read than "fourth beat in a row."

KMI does not guide revenue, gross margin, operating margin, or the quarter. The live guide is an annual budget set at the Q4 print, left unchanged in the formal table, and then described as a percentage above that budget. The new information on October 21 is verbal, and it is the third raise of 2026.

Waterfall of the FY2026 adjusted EBITDA floor, in millions. Each company step is a floor. The Q1 step excluded Monument. Later steps include it. The FMP bar is yellow because it is not company adjusted EBITDA.

Q4 is the residual, not a company target. Nine-month adjusted EBITDA is $2,539 million + $2,199 million + $2,180 million = $6,918 million. A floor above $9,116 million leaves Q4 above $2,198 million. Nine-month adjusted EPS of $1.20 leaves Q4 above $0.34. Q3 ran ahead of its old seasonal share, so the full-year raise of about $86 million was absorbed and the Q4 bar implicit in the floor went down. A Q4 that merely matches last year's EBITDA would put the year near $9.19 billion, about 6.8% above the $8.6 billion budget, already through the "more than 6%" language. The new floor does not require Q4 growth.

The full-year floor still shows faster growth than FY2025 delivered, only because the first nine months already banked it. The incremental quarter inside that floor is negative year over year. That is deceleration on purpose. It matches the call: a large piece of 2026 is weather, a contract buyout, a FERC true-up, a Waha-Houston spread, and commodity, and they will not put a colder-than-normal winter into the 2027 budget. Trident contributes only 30% of its contracted EBITDA in 2027. Kim Dang said 2026 will be a difficult year to replicate and that growth will not be even. The EPS floor sitting slightly under the EPS consensus, after three raises, is conservative on the metric the street actually uses.

Tone versus the prior two calls: the business rhetoric got louder and the near-term financial guide did not. July was the cautious call. October answered the backlog skeptics with signed projects and a hotter close from Rich Kinder, and refused 2027 in the first sentence. That is the same under-promise pattern, now aimed at next year rather than the back half.

Eight quarters, Q4 2024 through Q3 2026. Acceleration is the change in the year-over-year rate versus the prior quarter, in basis points. It is not sequential growth in dollars. Adjusted EPS is the series management quotes. GAAP is underneath because Q4 2025's +50% GAAP rate is not the trend.

1. Q1 2025 — revenue inflection, EPS air pocket. Revenue year-over-year jumps from −1.26% to +10.39% (+1,165 bps), the largest revenue acceleration in the window, as commodity sales go from −8% to +21%. Adjusted EPS year-over-year falls to 0% only because the dollar was unchanged at $0.34 against a $0.34 base. The rate collapsed. The level did not.
2. Q1 2026 — joint peak. Revenue +13.84% and adjusted EPS +41.18% ($0.48 versus $0.34). EPS acceleration that quarter was +1,930 bps. High of both series.
3. Q2 2026 — the turn. First down quarter for both rates. Revenue −308 bps to +10.76%. Adjusted EPS −903 bps to +32.14%. Still clearly growing. The direction of the rate changed.
4. Q3 2026 — deceleration continued, and on EPS it widened. Revenue −148 bps to +9.29%, the first sub-10% print since the Q1 2025 inflection. Adjusted EPS −1,145 bps to +20.69%. The second step down in the EPS rate is larger than the first. GAAP EPS only decelerated about 45 bps, to +21.4%, because it is coming off a lower Q2 rate. Reading GAAP alone makes Q3 look like a stabilization. Adjusted EPS says the rate is still falling. Q4 2025 GAAP at +50% is a false peak and is not the trend.

Nothing in the print says growth has rolled over. Both series are still up year over year. The acceleration phase ended in Q1 2026, and Q3 extended the slowdown. The call explains the level, not the basis points: transport +3%, gathering +14% on the call, refined-product volumes −3%, crude and condensate −11% with price and butane blending making up the products earnings line. The backlog does not explain why this quarter's rates stepped down. It is a later-decade driver.

Consensus already owns the 2026 earnings beat and agrees that 2027 does not repeat it. FMP FY2026 EPS of $1.56 is about 15% above the $1.36 budget, through the "more than 13%" verbal. FMP FY2027 EPS of $1.553 is flat. The argument the street is not capitalizing is duration. Next scheduled print is January 20, 2027 (2026Q4). Q4 itself is a "do not give the beat back" quarter: FMP Q4 EPS is $0.384, slightly below Q4 2025 adjusted EPS of $0.39.

The demand slides (Shell LNG toward 750 million tons by 2050, NERC peak power +225 to 245 GW by 2035, Wood Mackenzie about +46 Bcf/d of North American gas demand by 2035) are why the backlog signed. They are not why Q4 revenue should beat $4.55 billion. The Q4 tell is narrower: does Tennessee Gas LNG feedgas keep transport from decelerating further, from +3.8% toward flat? A re-trade of the build multiple above 6x, or a slip of Southeast Connector past 2032, is the fundamental disappointment on the bridge. More shippers toward 3.5 Bcf/d, versus the sanctioned up to 2.6 Bcf/d, is the upside still outside the $3.1 billion.

Balance sheet is not the 2026 constraint. Quarter-end leverage 3.5x is the low end of the range they describe. About $3 billion a year of cash available for expansion capex, against a run-rate "around $4 billion-ish," means the $14.4 billion is fundable only if the spend is spread and leverage is allowed to sit in the range. That is a 2028-30 capex-hump question, not a 2026 funding problem. Q3 DCF is not yet tagged; the last quarter on that series is 2026Q2 at $0.67 per share.

The street did not litigate the quarter. Fourteen analysts got on, and almost every question was about the project book. 19 well answered, 9 deflected. Those counts score each part of a dual-part question on its own. The table below merges some of those parts, so the rows will not add to 28. The deflections cluster on 2027 quantification, the residual shadow-backlog dollar, customer-specific timing and returns, and projects they do not operate.

Adjusted EPS $0.35 and adjusted EBITDA $2,180 million were accepted. The open question is the air pocket: management said 2026 one-timers will be hard to repeat and that growth will be uneven, then refused to preview 2027.

What was actually said, where it changes a model:

• 2027, the cleanest non-answer. Jeremy Tonet asked for help ex the one-timers, with Trident late. David Michels: "We are going to avoid giving any previewing the 2027 guide." The useful residue is a shape, not a number. 2026 is flattered by about $225 million of items plus about $175 million of commodity versus budget. 2027 does not get a full year of Trident. "No concern" about three-to-five-year growth is a different statement from "2027 grows."
• Shadow backlog. Julien asked whether the old "greater than $10 billion" opportunity set survived the conversion. Kim Dang: "substantial remaining," and the new pipes create the next bottlenecks. No dollar. Rich Kinder's "more horses just outside the corral" is the same non-number.
• 219 South. Sital Mody corrected the street. About $1.5 billion is the brownfield piece. Scaling to 2-3 Bcf/d "would be higher." Interest is over 2 Bcf/d. Sanction in pieces over a few quarters. Nothing is sanctioned until a customer signs.
• Returns and cost inflation. John Mackay asked whether customers had cooled and whether higher steel and compression forced a lower hurdle. Kim: tone has not changed, return thresholds are unchanged, and there has not been a lot of pushback. If the customer will not share overruns, the rate is higher. She would not give the Western Gateway sharing clause or the Permian Link hurdle.
• Permian Link timing. Jean Ann Salisbury asked whether the hyperscaler plant is online before the pipe. Sital would not separate the two. The anchor "is driving the need for speed." Other customers are in discussion. The pipe can be upsized with compression.
• Capture and the 200 GW slide. Kim's "about 45%" of southern power and "about 40%" of U.S. gas are illustrations. She said there is no science behind them. The 8 Bcf/d translation of a 200 GW interconnection queue assumes every request is built and all of it is gas. She said it will be some combination. Southeast Connector is 2.6 Bcf/d, and 3.5 is plausible, not signed.

Four pairs cannot both stand. Two change what a model should copy off the call: GAAP EPS, and what 5.7x is allowed to mean. One same-call sentence puts most of the pre-existing backlog online by the end of 2025, which the rest of the call and the release deny. One older pair is a false claim of continuity on owning power. The Q2 winter-storm "one-time versus structural" tension is real as emphasis and is not a hard break once the Q3 dollars are read: about $225 million is tagged as may-not-recur, and tightness is still expected to persist.

Raised guides, a larger backlog, and the step from "at least 5%" to "more than 6%" are updates, not contradictions. The Daloopa guidance series is still the $8.6 billion budget. Model the raised floor, not a restated budget. Products "all four segments up" is the adjusted column; GAAP products was more than offset by certain items. Both can be true.

The useful read-through is industrial, not consumer. Customers on the new gas projects are demand-pull. They are still signing at higher steel and compression costs, and KMI did not cut return hurdles. Versus the July 22 call, OpenAI, Southern Company, and Georgia Power were not named again. The new named equity partners are on refined products, not gas. Rates, CPI, and consumer sentiment were not discussed.

Project-cost inflation is being passed through. Kim Dang said cost per mile is up because steel and compression have increased, return thresholds did not change, and there has not been a lot of customer pushback. That is better for unnamed pipe, steel, and compression suppliers. It is neutral to slightly better for builders who can hold hurdles and put the cost into 20-year take-or-pay rates. It is worse for shippers only on the tariff, and management says they are not pushing back.

No consumer read. Refined-product volumes were down 3% versus Q3 2025, which the release assigns to a higher commodity-price environment. Crude and condensate volumes were down 11%. More than half of that is Double H out of crude service for an NGL conversion; the rest is fewer Hiland well connects in the Bakken. That is a Bakken activity comment, not a U.S. consumer comment. The offset is export: Houston Ship Channel docks moved 427,000 barrels a day, +39% versus Q3 2025 and a record, which management tied to a worldwide refined-product shortage. The Jones Act waiver added uncertainty; the fleet is 100% leased for the rest of 2026, 97% through 2027, and 80% through 2028 if likely options are exercised.

Crude stays on a fixed deck. Anthony Ashley: the curve has come up and is still backwardated; EOR is screened to lower-price cases; there is concern about where crude ends up if the war ends. They will not jump because the current year goes to $85. Budgeted WTI was $60, and commodity contributed about $175 million versus budget. That $175 million, and the $225 million of other items, are what FMP's flat 2027 is taking out.

Permitting is not a schedule upgrade on KMI's own telling. The bill in Congress has not passed. Kim supports it and said it would help certain projects, not theirs. MSX and South System 4 were permitted inside a year. The October 1 D.C. Circuit petition is aimed at those two certificates, which are the EBITDA that shows up before Southeast Connector and Permian Link. A wide Waha-Houston spread was good for KMI this year and was put in the may-not-recur bucket; it is typically bad for producers netted back at Waha. RNG sellers face a higher bar: operational lessons learned, no comment on live processes, and any larger deal would need a higher return.