American Electric Power Company AEP

Reference table

1. What the business is, and the load that changed it

American Electric Power is an electric utility holding company headquartered in Columbus, Ohio, with 17,581 employees, chaired and run by William J. (Bill) Fehrman. It reports four segments: Vertically Integrated Utilities, Transmission & Distribution Utilities, AEP Transmission Holdco, and Generation & Marketing.

Second-quarter 2026 GAAP earnings by segment, from the 2026-07-30 earnings release ($M):

Read that table honestly: the second quarter was flat to down at every segment. The GAAP decline from $1,226 million to $713 million looks catastrophic and is not — the prior-year quarter carried a $480 million benefit from the FERC NOLC Order for years 2021-2024, a one-off. Adjusting for it, operating earnings fell 3.1%. The company attributes the shortfall to "the 2025 transmission minority interest sale and the timing of tax-related items."

That minority-interest sale is worth naming precisely, because it explains a balance-sheet line the vendor payload shows without explanation. The FY2025 cash-flow statement carries "Proceeds from the Midwest Transmission Holdings Noncontrolling Interest Transaction, Net of Transaction Costs: $2,783 million." AEP sold a stake in its transmission business. That is why noncontrolling interests jumped from $42.3 million at 2024-12-31 to $1,080 million at 2025-12-31 and $1,210 million at 2026-06-30, and it is why AEP Transmission Holdco's contribution to earnings is flat year on year despite a growing rate base — a slice of it now belongs to somebody else.

The load data, which is the actual story

From the earnings release, kilowatt-hours delivered, three months ended June 30 (millions):

And for the six months, the divergence is sharper still: T&D commercial +24.8%, industrial +8.5%, residential −5.0%; VIU commercial +15.4%, residential −2.9%.

That is the shape of a data-centre build appearing in a delivery network. Residential demand is flat to falling — the normal state of a mature US utility, and the reason utilities were dull for two decades. Commercial demand is growing at mid-teens to mid-twenties percentages. The knowledge base's single claim on AEP, from August 2025, said "24GW new load by 2030." The company's own July 2026 release says 69 GW. The claim was directionally right and its magnitude has nearly tripled.

Management's forward framing, half-weighted per house rule: the release states AEP "expands new load additions to 69 gigawatts (GW) through 2030," "secures approximately 13 GW of gas-fired turbine capacity; additional 10 GW under evaluation," reaffirms "an annual operating earnings growth rate of 7% to 9% through 2030, with an expected operating earnings compound annual growth rate (CAGR) of greater than 9%," and describes "line of sight to incremental capital investments of more than $10 billion, including the fuel cell project in Wyoming, the Piketon transmission opportunity in Ohio, and incremental generation in AEP's footprint."

2. The free-cash-flow arithmetic, which is the reason for the verdict

This is a recomputation, not a quibble, and it inverts the most attractive number in the payload.

Filed cash flows, from the 10-K (FY2025/2024/2023) and the 10-Q (H1 2026/2025), $M:

Trailing twelve months to 2026-06-30, computed from the filings: operating cash flow $7,694M ($6,944M − $2,671M + $3,421M); construction expenditures $10,039M ($8,453M − $4,020M + $5,606M); free cash flow MINUS $2,345M, a free-cash-flow yield of −3.35% on a $69.93 billion market capitalisation. Including generation-facility acquisitions of $3,409M, the all-in figure is −$5,754M.

The vendor payload reports FY2025 free cash flow of +$6,814M and freeCashFlowYieldTTM of 12.807%.

Here is exactly how it got there, and the mechanism is worth stating because it is diagnosable from the payload alone. The vendor's capitalExpenditure for FY2025 is −$130,000,000. The 10-K's investing section reads, in order: "Construction Expenditures (8,453) · Acquisitions of Nuclear Fuel (130) · Acquisitions of Generation Facilities (3,453)." The vendor took the nuclear-fuel line. Not an approximation of capex, not a definitional variant — the wrong row of the same table, one line below the right one. It is a 98.5% understatement.

The payload then compounds it by adding rather than subtracting. freeCashFlowOperatingCashFlowRatioTTM reads 1.1764. A ratio of free cash flow to operating cash flow cannot exceed 1.0 unless capex is being added. Confirming: operatingCashFlowPerShareTTM is $13.994, capexPerShareTTM is $2.468, and freeCashFlowPerShareTTM is $16.462 — which is 13.994 plus 2.468, to three decimals. This is the same defect signature the programme documented on DUK, and it is present here in both halves: the wrong capex number, added instead of subtracted.

Consequently rejected outright: freeCashFlowYieldTTM (12.807%), freeCashFlowPerShareTTM ($16.462), priceToFreeCashFlowRatioTTM (7.808x), evToFreeCashFlowTTM (13.738x), freeCashFlowToEquityTTM ($6,676.9M), capexToOperatingCashFlowTTM (0.176), capexToDepreciationTTM (0.371), capexToRevenueTTM (0.0596), capitalExpenditureCoverageRatioTTM (5.670x) and dividendPaidAndCapexCoverageRatioTTM (2.247x). The rating block's discountedCashFlowScore of 1 out of 5 is the only vendor field that appears to have noticed something was wrong, and it reaches the right conclusion for reasons the payload does not disclose.

What this means economically, stated without drama. Negative free cash flow is normal and correct for a regulated utility in a build cycle — the capital goes into rate base and earns an approved return for decades. It is not a distress signal. What it does mean is that AEP cannot fund its own dividend, let alone its own growth, and must return to the debt and equity markets every year for five years. That is the risk the price carries, and a 12.81% free-cash-flow yield in a screening tool describes the opposite of it.

3. The funding: debt, equity, and 23.5 million shares the payload has never seen

Capitalisation from the 10-Q ($M):

The 10-Q's own language: "AEP's ratio of debt-to-total capital increased from 60.3% to 61.4% as of December 31, 2025 and June 30, 2026, respectively, primarily due to an increase in long-term debt to support AEP's capital investment plan in addition to working capital needs."

The equity side is the corporate action the payload missed. From the 8-K filed 2026-05-14:

> "On May 12, 2026, the Company entered into an Underwriting Agreement... relating to the registered public offering and sale by the Forward Purchasers of 20,472,442 shares of the Company's common stock, par value $6.50 per share. On May 13, 2026, the underwriters exercised in full their option to purchase an additional 3,070,866 shares."

That is 23,543,308 shares — 4.3% of the 544,362,629 outstanding — sold forward through Bank of America, Goldman Sachs and Morgan Stanley, not yet settled and therefore not yet in the share count, the market capitalisation, or any per-share figure in the vendor payload. The 8-K also warns explicitly that a forward purchaser may accelerate settlement "irrespective of the Company's interests, including the Company's need for capital," which "would result in dilution to the Company's earnings per share."

A second corporate action sits in the balance sheet and is easy to miss: authorised common shares increased from 600,000,000 to 900,000,000 between 2025-12-31 and 2026-06-30. A company does not raise its authorised share count by 50% without intending to use some of it.

On top of that, $405 million of common stock was issued in the first half of 2026 (against $132 million in the first half of 2025) and $5,045 million of long-term debt (against $3,163 million). The funding machine is running at roughly double last year's rate, and the equity component is where the per-share arithmetic leaks.

4. Income statement and the quality of the trailing numbers

Two cautions on this table. First, FY2025 GAAP EPS of $6.66 is not the number to value against — it includes the $480 million FERC NOLC benefit and gains associated with the transmission minority-interest sale. The company's own operating measure is materially lower, and the FY2025 consensus epsAvg of $5.913 is the correct trailing anchor. Second, the FY2025 diluted share count of 520.2 million is identical to FY2023's 520,206,258 to the share — the vendor has carried a stale value forward. The 10-Q's actual count at 2026-06-30 is 544,362,629, 4.6% higher.

Recent quarters, from inc_q:

The Q2 2026 miss is the largest surprise in either direction in the file, and it is the wrong direction. Note also that inc_q reports Q2 2026 operating income of $2,584M against Q1's $1,360M while net income fell — an internal inconsistency in the vendor's quarterly operating-income line that we do not rely on. The GAAP-to-operating bridge in the earnings release is the authority and is used throughout.

The bridge itself is clean and small. Q2 2026 GAAP $1.31 → operating $1.36 through four items: mark-to-market hedging −$0.02, Unified Tracker Mechanism partial disallowance +$0.04, wholesale customer contract agreements +$0.04, tax effect −$0.01. A 3.8% adjustment. Compare the prior-year quarter, where the bridge ran $2.29 GAAP to $1.43 operating — a 37.6% adjustment driven by the $480 million FERC item. The current adjustment is honest; the prior-year one is why the comparison looks so bad.

Three of the four year-to-date adjustments are regulatory disallowances, and that is a pattern worth naming: the Pirkey Plant partial disallowance in the 2025 Texas base-rate case ($31M), the AEP Texas Unified Tracker Mechanism partial disallowance ($22M), and SWEPCo wholesale customer contract agreements ($23M), partially offset by a WVPSC order benefit (−$35M). $76 million of probable disallowances booked in six months is not large against $1.6 billion of first-half operating earnings, but it is the mechanism by which a regulator declines to pay for something, and this company is about to ask a regulator to pay for $78 billion.

5. Valuation — priced in or room?

At $128.45 (market cap $69.93B, 544,362,629 shares):

Estimate coverage is adequate on the near years and thin beyond. FY2027 rests on 13 analysts with a range of $6.776 to $6.986 — a 3.1% spread, the tightest of any name in this batch and a reflection of how mechanical utility earnings are. FY2029 rests on 2 analysts and is excluded from every conclusion in this dive. FY2030's 5 analysts are used only for the long-horizon growth rate.

est.ebitdaAvg and ebitAvg are REJECTED for a fixed-ratio fabrication signature. In every forward year from FY2024 through FY2030, ebitdaAvg is exactly 39.61% of revenueAvg and ebitAvg is exactly 22.65% — to four significant figures, across seven consecutive years, for a company whose actual EBITDA margin has ranged from 37.2% to 41.7% over the same span. The FY2023 row breaks the pattern (ebitAvg at 16.86% of revenue), which locates where the fabrication begins. All forward valuation in this dive uses epsAvg, which cross-checks well: the FY2026 consensus of $6.363 sits 0.6% below the midpoint of the company's own guidance range of $6.25-$6.55, raised five days ago and evidently not yet fully absorbed.

Enterprise value — checked, and the NCI omission is real but immaterial. The vendor reports enterpriseValueTTM of $123,067M against a market capitalisation of $69,944M, implying net debt of $53,123M. The filing's total debt at 2026-06-30 is $52,836M against cash and equivalents of $375M and other temporary investments of $228M, giving net debt of $52,233M. The vendor's figure is $890M higher, consistent with including roughly $662M of lease obligations that the company excludes from its own capitalisation table. Separately, the $1,210M of noncontrolling interests is NOT added — the standard omission this programme has now found on the majority of names with minority interests. The two errors run in opposite directions and largely cancel: a correctly built EV is approximately $123,616M, 0.4% above the vendor's. We record the vendor's EV as substantially correct — an unusual finding and reported as such — and note that netDebtToEBITDATTM of 6.176x reconciles exactly to trailing EBITDA of $8,601M summed from inc_q, and is consistent with the filing's 61.4% debt-to-capital. The derived leverage ratios do NOT contradict the raw block on this name.

Peer context. The vendor peer set is appropriate for once: Dominion, Duke, Consolidated Edison, Entergy, Exelon, National Grid, PSEG, Sempra, Vistra and Xcel. We draw no peer multiple comparison because this dive holds no verified current EPS estimates for those names, but the set itself is correctly constructed, unlike several in this programme.

5a. What today's price assumes (the inversion)

At $128.45 — 20.2x FY2026 consensus, 18.7x FY2027, 2.10x book — the price embeds:

• EPS reaches $6.363 in 2026 and $6.856 in 2027. (Consensus; 11 and 13 analysts.) The first half delivered $3.01 of operating EPS, so the full year needs $3.24-$3.54 in the second half to land inside the guided $6.25-$6.55. The third quarter is seasonally the strongest — $1.80 last year — so the arithmetic works, but it works with the second-quarter miss already inside it.
• The regulator approves substantially all of a $78 billion capital plan at the current allowed returns. (Our derivation; the plan figure is the company's.) This is the most fragile assumption in the price. Three probable partial disallowances were booked in the first half of 2026 alone. A capital plan is a request, not a receivable.
• The equity funding gets done without a de-rating. (Our number.) Free cash flow is roughly −$2.3 billion trailing, the dividend costs approximately $2.06 billion a year, and 23,543,308 shares are already sold forward. The price assumes the market absorbs several years of this at or above 18x forward earnings.
• The market keeps paying roughly 18-20x forward earnings. (Our number.) At 17x FY2027E the stock is $117; at 22x it is $151. Utilities have historically re-rated with rate expectations more than with growth, and nothing in this file gives us a rate view.
• 69 gigawatts of pipeline converts into energised, billed load. (Company disclosure, half-weighted.) Pipeline is not contract and contract is not energised. The commercial kilowatt-hour growth of 14.9% and 17.4% is the part that has actually happened.

5b. The return bridge (why the multiple moves)

Expected return over the next twelve months decomposes as: EPS growth (+7.8%, FY2026E $6.363 to FY2027E $6.856) + multiple drift (roughly HELD, 20.2x on the forward year to about 19.5x on the then-forward year) + dividend yield (+2.94%) − share issuance (roughly −1.5% to −2.5% annually at the current pace) ≈ +7% to +9%.

Our base assumes the multiple holds. It does not assume expansion and it does not need it. The entire return is earnings growth plus dividend, less dilution. That is a defensible, low-variance return and it is approximately what a regulated utility should deliver — which is precisely why it does not clear a Buy bar. At $137 the base is 6.7% above spot; adding the dividend gives roughly 9.7% over twelve months, before the dilution drag.

If the multiple compressed to 17x FY2027E the price would be $117 (−9.2%). If it expanded to 22x, $151 (+17.6%). The bull case at $162 is 23.2x the FY2027 consensus high — a level that requires the load story to be re-rated as growth rather than as regulated utility, which has happened to other names and is not our base.

5c. Variant perception (where we differ, what would surprise)

• We differ from the vendor block far more than from the street. Our base of $137 is 2.5% below the consensus $140.55, and the analyst low of $129 is above spot — 22 buy, 14 hold, no sell. On the level, we are essentially at consensus and we say so. Where we differ materially is on the cash-flow characterisation: any screen built on the payload's 12.81% free-cash-flow yield is looking at a company that does not exist, and the correction is worth more than the two-and-a-half-dollar valuation gap.
• We think the funding cost is under-discussed relative to the load story. The bullish case circulating on this name — including the knowledge base's single claim — is about gigawatts. The gigawatts are real and larger than the claim said. The question the price has to answer is what 4.3% of forward-sold shares, plus $4 billion of net new debt in six months, plus a 61.4% debt-to-capital ratio, does to per-share earnings over five years. Watchable number: shares outstanding at each quarter-end, and debt-to-total-capital in the liquidity section of each 10-Q.
• Positive surprise that would force a re-rate: third-quarter commercial kilowatt-hour growth holding above 15% with a rate-case outcome granting full recovery on new generation, which would convert the pipeline from a capital-intensity story into an earnings-acceleration story; or the FY2027 consensus moving above $7.00 on the strength of the raised guide.
• Negative surprise that would break the thesis: a material disallowance in the 2025 Texas base-rate case beyond the $31 million already booked; debt-to-total-capital above 63%, or a credit-rating action; a second consecutive quarterly miss that forces the $6.25-$6.55 guide down; or acceleration of the forward sale agreements at a lower price, which the 8-K explicitly warns may happen "irrespective of the Company's need for capital."

Synthos fair values

All three anchors are multiples of the FY2027 consensus EPS distribution (mean $6.856, low $6.776, high $6.986, 13 analysts), each cross-checked against FY2026 and FY2028.

• Bear ~$108 — 15.9x the FY2027 consensus LOW of $6.776. Cross-check: 17.0x FY2026E; 1.5% above the 52-week low of $106.44. The scenario: rate cases go badly, the capital plan is trimmed, debt costs rise, the equity issuance is done at a discount and the market applies a utility-average multiple to a below-guidance earnings path. −15.9%.
• Base ~$137 — 20.0x the FY2027 consensus MEAN of $6.856. Cross-check: 21.5x FY2026E; 18.0x FY2028E; 2.24x book. Sensitivity, stated openly: 18x gives $123 — below spot — and 22x gives $151. The answer lives in a 4-point multiple band on one estimate row, and that band is wider than the entire analyst EPS range of 3.1%. The scenario: guidance is met, the load pipeline continues converting, funding gets done without incident and the multiple holds. +6.7%.
• Bull ~$162 — 23.2x the FY2027 consensus HIGH of $6.986. Cross-check: 21.3x FY2028E; 17.7x FY2030E; 16.8% above the 52-week high of $138.69. The scenario: 69 GW converts faster than modelled, incremental capital beyond the $78 billion plan is approved, the growth rate is re-rated toward the top of the 7-9% band or above, and AEP trades as an infrastructure growth asset. +26.1%.

Base is 6.7% above spot; asymmetry roughly 1.64:1 (15.9% down, 26.1% up), plus a 2.94% dividend. That is not enough asymmetry to earn a Buy tier on a name whose funding requires continuous market access, and the verdict is Hold.

6. Knowledge base — 243 raw hits, one survivor, and it undershot the company

Raw hits: 243. After a case-sensitive entity re-run: 1. Name-level claims on AEP: 1, and it is BULLISH. Discarded: 242.

The primary sweep ran the entity terms AEP, American Electric Power and AEP Transmission, plus deliberately broad free text on utility and utilities, across all 52,021 distilled claims. It returned 243 hits across 45 channels — and the sweep is discarded in its entirety, said plainly.

Why the raw sweep is worthless here. The free-text terms caught the entire two-year macro conversation about electricity demand and AI power consumption: jordi_visser (25), forward_guidance (16), raoul_pal (15), real_vision (13), darius_dale (12), lyn_alden (11), compound_and_friends (10), arthur_hayes (10) and thirty-seven other channels. None of it names this company. "AEP" as a three-letter token is also a live homograph risk of exactly the class this programme documented on CEG, EMR, USB and MMM. A case-sensitive entity re-run collapses 243 to 1, and that is the correct discipline.

The single surviving claim, verbatim:

> 2025-08-03 · BULLISH · conviction 75 · horizon: thesis · entities: AEP · channel: jordi_visser · no named speaker · role: independent

> "Utilities won't trade like boring utilities; AEP has pricing power, signed long-term contracts, 24GW new load by 2030 and a raised $70B capital plan — no recession from AI."

What to do with it, stated without inflation. The claim is eleven months old and its thesis has been confirmed while its magnitudes have been overtaken by the company's own disclosure in the reader's favour:

Three of the four elements have been confirmed and enlarged. The fourth — "pricing power" — is the one this dive disputes, and the disagreement matters. A regulated utility does not have pricing power in the ordinary sense; it has a right to petition for cost recovery, and in the first six months of 2026 that petition was partially refused three times. The claim's framing is the one place where an otherwise well-aimed call overstates the business model.

Attribution. The claim carries speaker: null and is channel-attributed to jordi_visser with speaker_role: independent — sourcing at the weaker end of what the 4-lane policy admits, and the speaker_role field is treated per house rule as unreliable and judged on content instead, which here is clearly independent commentary rather than management voice.

Conclusion. Breadth 1, claim count 1, net conviction positive-very-low. The Synthos knowledge base holds one claim on American Electric Power. It is bullish, it has been right about the demand and wrong about the pricing power, and one claim from one channel with no named speaker is not a basis for conviction in either direction. The lane is reported at its true size.

7. Data integrity — what we rejected and why

Seven findings. AEP's payload contains the single largest capital-expenditure error this programme has documented on a utility, one missing corporate action, one empty block and one broken segment table — against which the enterprise value, the share-count reconciliation and the leverage ratios are all substantially correct, which is worth recording explicitly.

1. capitalExpenditure for FY2025 is the WRONG LINE ITEM — a 98.5% understatement — and free cash flow is computed by ADDING it. REJECTED. The vendor reports FY2025 capitalExpenditure of −$130,000,000. The 10-K's investing section lists Construction Expenditures of $(8,453)M, then Acquisitions of Nuclear Fuel of $(130)M, then Acquisitions of Generation Facilities of $(3,453)M. The vendor took the nuclear-fuel line. It then reported FY2025 free cash flow of +$6,814M (= $6,944M operating cash flow + $130M) and a freeCashFlowYieldTTM of 12.807%. The addition is confirmed independently by freeCashFlowOperatingCashFlowRatioTTM of 1.1764 — arithmetically impossible — and by freeCashFlowPerShareTTM ($16.462) equalling operatingCashFlowPerShareTTM ($13.994) plus capexPerShareTTM ($2.468). The correct FY2025 figure is MINUS $1,509M; the correct trailing-twelve-month figure is MINUS $2,345M, a yield of −3.35%. Ten derived fields are rejected as a consequence and listed in Section 2. Note that FY2022, FY2023 and FY2024 capex in the payload (−$6,772M, −$7,507M, −$7,771M) are all approximately correct against the filings — the defect appears only in FY2025 and in the TTM block, which is the definition-change signature.

2. The 2026-05-12 forward equity sale — 23,543,308 shares, 4.3% of the count — is absent from every vendor field. From the 8-K filed 2026-05-14: 20,472,442 shares sold forward on 12 May, plus a fully exercised option for 3,070,866 more on 13 May, through BofA, Goldman Sachs and Morgan Stanley. Unsettled, so not in the share count, market capitalisation, book value per share or any per-share estimate. The same balance sheet shows authorised shares raised from 600,000,000 to 900,000,000. This is the "corporate action absent from the payload" class, and it is the eleventh consecutive batch in which it has fired.

3. seg_geo is EMPTY — zero rows. No geographic revenue split of any kind. AEP is a US-only regulated utility so the economic loss is small, but the field is void and is reported as void rather than described.

4. seg_prod loses the largest segment for the two most recent years — REJECTED for FY2024 and FY2025. The FY2023 and earlier rows carry the genuine four-segment structure (Vertically Integrated Utilities $11,459.9M, T&D Companies $5,713.3M, Generation & Marketing $2,036.7M, AEP Transmission Holdco $1,765.7M). The FY2024 and FY2025 rows drop Vertically Integrated Utilities and AEP Transmission Holdco entirely, leaving FY2025 as Generation And Marketing $2,697M + Transmission And Distribution Companies $6,097M + "Product and Service, Other" $526M = $9,320M against total revenue of $21,783M — 57% of the business is missing. All segment figures in this dive come from the earnings release.

5. est.ebitdaAvg and est.ebitAvg carry a fixed-ratio fabrication signature — REJECTED. Across FY2024 through FY2030, ebitdaAvg is exactly 39.61% of revenueAvg in every single year and ebitAvg is exactly 22.65%. Seven consecutive years at a constant margin to four significant figures is not an estimate. The FY2023 row (16.86%) sits outside the pattern and marks its start. All forward valuation uses epsAvg only, which validates well against the company's own guidance (FY2026 consensus $6.363 against a guided midpoint of $6.40).

6. weightedAverageShsOutDil for FY2025 is stale — it repeats the FY2023 value to the share. FY2023 reads 520,206,258 and FY2025 reads 520,206,000. The actual count at 2026-06-30 is 544,362,629. Any per-share figure built on the FY2025 income statement is 4.6% flattered. The quote block's implied count is correct — see below.

7. totalDebt includes lease obligations the company excludes — a 2.9% overstatement, minor. The vendor's FY2025 totalDebt of $50,241M against the 10-K capitalisation table's $48,830M, a $1,411M gap of which $690M is the explicitly reported capitalLeaseObligations. This is the operating-lease inflation class, present but small here, and it is what makes the vendor's enterprise value net out approximately correct despite the NCI omission.

Verified CLEAN — recorded because a clean check is a finding:

• Share count. The 10-Q states 545,549,444 issued less 1,186,815 treasury = 544,362,629 outstanding at 2026-06-30, against 544,395,442 implied by market capitalisation ÷ price — a 0.006% match.
• Enterprise value. The vendor's $123,067M against a rebuilt $123,616M (market cap + filed total debt − cash − temporary investments + noncontrolling interests) — a 0.4% difference. The $1,210M of noncontrolling interests IS omitted, per the standard defect, but the $662M of lease obligations wrongly included offsets most of it. Net effect immaterial; the multiple is usable.
• Leverage ratios do NOT contradict the raw block. netDebtToEBITDATTM of 6.176x reconciles exactly to trailing EBITDA of $8,601M summed from the four quarters in inc_q, and is consistent with the filing's stated 61.4% debt-to-total-capital. This is the check that failed on another utility in this programme; here it passes.
• tech versus quote. tech.hi52/lo52 ($138.69/$106.44) against quote.yearHigh/yearLow ($140.58/$105.70) — a 1.4% and 0.7% discrepancy. tech is used throughout for internal consistency. tech.last of $128.54 against quote.price of $128.45 is a 0.07% difference; the quote price is used as the strike.
• tech.max_dd_from_peak of −7.318% coincides exactly with pct_from_hi. Per the data contract this is correct, not a defect — the six-year peak falls inside the last twelve months. It is the current distance from the multi-year high and is described as such, never as a maximum drawdown.
• dividendPerShareTTM of $3.78 is consistent with profile.lastDividend of 3.78 and with FY2025 common dividends paid of $2,008M against roughly 531 million average shares ($3.78). Clean.

Vendor composite rating — C+ / 2 overall, with discountedCashFlowScore of 1 out of 5. That sub-score is the only field in the payload that behaves as though the cash-flow data were bad. Its conclusion happens to be right; its inputs are the same rejected free-cash-flow figures, so it earns no weight here.

Non-equity tripwire — checked and passed. AEP is common stock, $6.50 par value, NASDAQ-listed. Price of $128.45 is not par-like; beta is 0.504; volume was 2.83M shares (~$364M of turnover); the 52-week band of $106.44 to $138.69 is a 30% range; the dividend is a regular quarterly common dividend. This is common equity.

8. Technicals

• Price $128.45. −7.3% from the 52-week high of $138.69; +20.8% above the 52-week low of $106.44. Position within the annual range: 68th percentile.
• Split moving averages, with the shorter one below the price. 2.4% BELOW a 50-day average of $131.60; 1.7% ABOVE a 200-day average of $126.36. The 50-day sits above the 200-day, so the longer-term structure is intact while the short-term one has rolled over.
• RSI 38.3 — weak, approaching but not at oversold. MACD −1.17 — negative.
• Current distance from the six-year peak: −7.3%, identical to the 252-day figure, which per the data contract means the six-year high was set inside the last twelve months.
• Relative performance: 3-month −4.5% against SPY +7.6% and QQQ +7.7% — a 12-point deficit; 6-month +8.6% against SPY +11.1%; 12-month +13.2% against SPY +24.3% and QQQ +30.8%. AEP has underperformed the index over every window in the file, and the deficit is widest over the most recent quarter — which contains the second-quarter miss.
• Sentiment: 22 buy, 14 hold, 0 sell across 36 analysts; consensus target $140.55 (+9.4%), median $142, high $150 (+16.8%), low $129 — 0.4% ABOVE spot. No covering analyst has a target below the current price. As on other names in this programme, we treat unanimity about the absence of downside as a caution rather than a comfort.

Today's move and what it does to the entry

AEP closed 2026-08-04 at $128.45, up $0.13 or 0.10%, from $128.32. It opened at $128.20 and traded $126.27 to $128.83 on 2.83M shares. The nearest company events are the 2026-07-30 earnings release and 10-Q, five days earlier, and Form 4 filings dated 2026-08-04 covering 2026-08-01 transactions.

The honest read: the entry is into weakness, not strength, and the weakness is deserved. Below the 50-day, RSI 38, down 4.5% in three months against a market up 7.6%, five days after an 8.1% earnings miss. That is a better place to buy a good business than the alternative — but the base case is +6.7%, and buying a fairly-valued utility in a downtrend on a 1.64:1 payoff is not a decision, it is an activity.

9. Insiders — eight filings, no open-market purchase, no open-market sale

The reading is that this file contains no signal at all, and saying so is more useful than manufacturing one. Two of the eight are F-InKind — shares withheld by the company to cover tax on vesting restricted stock units, a mechanical, non-discretionary event. Four are A-Award grants of phantom stock units to directors at zero price, the standard quarterly director-compensation entry. Two are Form 3 initial statements of beneficial ownership by directors elected on 2026-07-20 — Charles J. Meyers and David S. Marriott, whose appointments are documented in the 8-K filed 2026-07-21 and who hold 225 shares and zero shares respectively.

There is not one open-market purchase and not one open-market sale in the file. Chief executive William Fehrman holds 133,539 shares after the withholding, worth approximately $17.2 million at the closing price. On a name where the entire question is whether management can execute a $78 billion capital plan without destroying per-share value, the complete absence of discretionary insider activity is uninformative, and it is reported as uninformative.

10. Verdict, kill-criteria and flip conditions

Hold.

The case for the business, stated at full strength. American Electric Power is experiencing the first genuine demand growth its industry has seen in a generation, and the evidence is in delivered kilowatt-hours rather than in projections: commercial volumes up 14.9% and 17.4% by segment in the quarter, and 15.4% and 24.8% for the six months, while residential volumes fell. The company raised full-year guidance five days ago to $6.25-$6.55, expanded its disclosed load pipeline to 69 gigawatts through 2030, secured approximately 13 GW of gas-fired turbine capacity, and reaffirmed a 7-9% earnings growth rate on a $78 billion capital plan with line of sight to $10 billion more. Consensus carries EPS from $6.363 to $9.158 between 2026 and 2030, a 9.5% compound rate. The dividend is $3.78, a 2.94% yield. Beta is 0.504.

The case against the price, which is what decides the verdict.

First, the return is ordinary and the arithmetic says so. Base fair value $137, 6.7% above spot, against a street at $140.55 — so we have essentially no variant perception on the level. Asymmetry is 1.64:1. Add the dividend and the twelve-month expected return is roughly 9.7%, less dilution. That is a fair outcome for a regulated utility and not a reason to buy one.

Second, this company cannot fund itself, and the payload disguises that fact. Trailing free cash flow is MINUS $2,345 million against a payload claiming +$6,814 million and a 12.81% yield. The dividend alone costs $2.0 billion a year. Debt-to-total-capital went from 60.3% to 61.4% in six months on $4.0 billion of net new debt, and 23,543,308 shares — 4.3% of the count — were sold forward in May in a transaction no vendor field records. A five-year plan funded this way is a five-year sequence of opportunities to issue equity at the wrong price.

Third, the most recent quarter missed by 8.1%, GAAP earnings fell from $1,226 million to $713 million, and three separate probable regulatory disallowances were booked in the first half. The company's "pricing power" — the exact phrase in the knowledge base's only claim — was partially refused three times in six months.

Fourth, the knowledge base holds one claim. It is bullish, its demand thesis has been confirmed and nearly tripled by the company's own disclosure, and it is one channel-attributed claim with no named speaker out of 52,021. Breadth of 1 does not carry a verdict in either direction.

Pre-registered KILL criteria — what would take this to Avoid:

• Debt-to-total-capital above 63% in any 10-Q liquidity section, or a negative rating action from any agency.
• A second consecutive quarterly miss forcing the $6.25-$6.55 full-year guide down.
• A material adverse outcome in the 2025 Texas base-rate case beyond the $31 million Pirkey disallowance already booked, or a new disallowance above $100 million in any single proceeding.
• Acceleration of the forward sale agreements at a price below $115, which the 8-K warns may occur "irrespective of the Company's interests, including the Company's need for capital."
• Commercial kilowatt-hour growth falling below 5% year on year in either utility group — the load story is the only reason this is not simply a bond substitute.
• A cut or freeze in the dividend, which at a 65.2% payout ratio and negative free cash flow is not unthinkable if the capital plan accelerates.

Pre-registered UPGRADE conditions — what would take this to Buy — Tactical:

• A price below approximately $115, which is 16.8x the FY2027 consensus and would put the base case above 19% with the dividend on top.
• Full cost recovery granted on new large-load generation in a completed rate case, converting the 69 GW pipeline from capital intensity into earnings acceleration.
• The FY2027 consensus moving above $7.00 on the strength of the raised guide, with the analyst low staying above $6.90.
• Evidence that the capital plan can be funded with less equity than modelled — for example an asset sale, a further minority-interest transaction on the Midwest Transmission Holdings model, or free cash flow turning less negative than the current −$2.3 billion trailing.
• Knowledge-base breadth above 3 with independent confirmation of the load thesis from a second named source.

Where AEP fits in the Synthos Framework Portfolio. No position today. The utilities sleeve has room and this is a credible candidate for it, but at 20.2x forward earnings with a 6.7% base case and a funding gap of this size, the correct action is a watch with a price trigger, not an allocation. Logged as a tracked Synthos call (Hold) as of 2026-08-04 at $128.45, with the fair-value anchors, the kill criteria and the upgrade conditions all gradeable.

Single biggest risk: the funding gap. A $78 billion capital plan meets negative $2.3 billion of trailing free cash flow and 61.4% debt-to-capital. The gap closes with debt and equity, every year, for five years. AEP has already sold 23,543,308 shares forward and raised its authorised share count from 600 million to 900 million. Each issuance is dilution the 7-9% earnings growth rate must first overcome before a shareholder is any better off. The load is real, the pipeline is real, and the demand thesis in the knowledge base has been vindicated at nearly three times its stated magnitude — but a regulated utility converts demand into shareholder value only through a rate base that must be paid for, and this file's most persuasive-looking number, a 12.81% free-cash-flow yield, is an artefact of the wrong row in a cash-flow statement.

Provenance & disclosures

• Traceability: 1 name-level knowledge-base claim names American Electric Power out of 52,021 distilled claims (raw hits 243, surviving a case-sensitive entity re-run 1, entity matches 1, discarded 242; breadth 1, net conviction positive-very-low). The claim: jordi_visser, 2025-08-03, bullish, conviction 75, horizon thesis, speaker: null, speaker_role: independent, channel-attributed — "Utilities won't trade like boring utilities; AEP has pricing power, signed long-term contracts, 24GW new load by 2030 and a raised $70B capital plan — no recession from AI." Its demand thesis has been confirmed and enlarged by the company's own 2026-07-30 disclosure (69 GW, $78 billion); its "pricing power" framing is disputed in Section 6 on the evidence of three probable partial regulatory disallowances booked in the first half of 2026. The 242 discards are homograph and free-text noise — "AEP" as a three-letter token plus deliberately broad sweeps on "utility"/"utilities" caught the entire macro conversation about AI electricity demand across 45 channels led by jordi_visser (25), forward_guidance (16), raoul_pal (15), real_vision (13) and darius_dale (12), none of which names this company. The sweep is discarded in its entirety and said so. No claim carries a management speaker_role; the one that carries independent is judged on content, per house rule, and is independent commentary. No concentration test is meaningful on a single claim. The quote is verbatim from the stored claim text.
• Data as-of: fundamentals — balance sheet, capitalisation, cash flows and segment earnings through 2026-06-30, filing-verified from the 10-Q filed 2026-07-30; second-quarter results, guidance, load data and the GAAP-to-operating bridge from the 8-K filed 2026-07-30 carrying the earnings release as Exhibit 99.1; annual figures for FY2025, FY2024 and FY2023 from the 10-K filed 2026-02-12 · estimates 2026-08-04 · prices 2026-08-04, quote timestamp 1785873568 = 2026-08-04T19:59:28Z ($128.45, +0.10%; 50-DMA $131.60; 200-DMA $126.36; RSI 38.3; MACD −1.17) · knowledge-base claims 2026-08-04. AEP's fiscal year is the calendar year. All figures come from the Synthos vendor data file for AEP or from the SEC filings in the AEP archive; no figure comes from memory, recall or external retrieval.
• Filing archive contents: 10-K filed 2026-02-12 (fiscal 2025 — source of the $8,453M construction-expenditure figure that inverts the free-cash-flow conclusion, and of the $2,783M Midwest Transmission Holdings noncontrolling-interest proceeds); 10-Q filed 2026-05-05 (March 2026 quarter); 10-Q filed 2026-07-30 (June 2026 quarter — source of the capitalisation table, the H1 cash flows and the share count); 8-K filed 2026-05-14, which carries the 23,543,308-share forward sale agreements and underwriting agreement; 8-K filed 2026-07-21 (Item 5.02 — election of directors David S. Marriott and Charles J. Meyers effective 2026-07-20); 8-K filed 2026-07-30, which carries the complete second-quarter earnings release with the raised guidance, the 69 GW load figure, the GAAP-to-operating reconciliation and the kilowatt-hour tables. All carry preserved [TABLE] statement data.
• Where the filings contradicted or corrected the vendor (detailed in Section 7): capitalExpenditure — the vendor reports FY2025 capex of −$130M, which is the 10-K's Acquisitions of Nuclear Fuel line, not its Construction Expenditures line of −$8,453M, a 98.5% understatement; the payload then computed free cash flow by ADDING it, producing +$6,814M and a 12.807% yield where the correct FY2025 figure is MINUS $1,509M and the correct trailing figure MINUS $2,345M (−3.35%), with the addition confirmed by freeCashFlowOperatingCashFlowRatioTTM of 1.1764 and by freeCashFlowPerShareTTM = operatingCashFlowPerShareTTM + capexPerShareTTM exactly; the 2026-05-12/13 forward sale of 23,543,308 shares (4.3% of the count) and the increase in authorised shares from 600,000,000 to 900,000,000, both absent from every vendor field; seg_prod, which drops the Vertically Integrated Utilities and AEP Transmission Holdco segments for FY2024 and FY2025, leaving 57% of revenue unaccounted; seg_geo, which is empty; est.ebitdaAvg/ebitAvg, fixed at exactly 39.61% and 22.65% of revenueAvg in every year from FY2024 to FY2030; weightedAverageShsOutDil for FY2025, which repeats the FY2023 value of 520,206,258 against an actual 544,362,629; and totalDebt, which at $50,241M for FY2025 exceeds the 10-K capitalisation table's $48,830M by $1,411M including $690M of explicitly reported lease obligations. Where vendor and filing AGREED — recorded, because clean checks are findings: the share count matches to 0.006% (10-Q 544,362,629 outstanding against 544,395,442 implied by market capitalisation ÷ price); the enterprise value of $123,067M is within 0.4% of a rebuilt $123,616M — the $1,210M of noncontrolling interests IS omitted per the standard defect, but is nearly offset by $662M of lease obligations wrongly included; netDebtToEBITDATTM of 6.176x reconciles exactly to trailing EBITDA of $8,601M summed from inc_q and is consistent with the filing's 61.4% debt-to-capital, so the derived leverage ratios do NOT contradict the raw block on this name; dividendPerShareTTM of $3.78 reconciles to FY2025 common dividends paid of $2,008M; and tech.max_dd_from_peak of −7.318% coincides with pct_from_hi correctly, the six-year peak falling inside the last twelve months.
• Basis note: analyst estimates and company guidance are on operating (non-GAAP) earnings per share, which exclude mark-to-market hedging, the FERC NOLC Order, regulatory disallowances and asset-sale items. The Q2 2026 bridge is small and disclosed: GAAP $1.31 plus hedging −$0.02, Unified Tracker disallowance +$0.04, wholesale contract agreements +$0.04 and tax effect −$0.01 equals operating $1.36 — a 3.8% adjustment. The prior-year quarter's bridge was 37.6% ($2.29 GAAP to $1.43 operating) on the $480 million FERC NOLC Order, which is why the year-on-year GAAP comparison is misleading and is not used. Forward multiples use consensus epsAvg; FY2025 GAAP EPS of $6.66 is not used as a trailing anchor because it contains the FERC item and transmission-transaction effects.
• Estimate coverage: 11 analysts on FY2026 EPS, 13 on FY2027 — the anchor for all three fair values, with a range of $6.776 to $6.986, a 3.1% spread; 7 on FY2028; 2 on FY2029, which is excluded from every conclusion; 5 on FY2030, used only for the long-horizon growth rate.
• Peer note: the vendor peer set — Dominion, Duke, Consolidated Edison, Entergy, Exelon, National Grid, PSEG, Sempra, Vistra, Xcel — is correctly constructed for this name, which is not always the case in this programme. No peer-multiple comparison is drawn because this file holds no verified current estimates for those names.
• Fair-value caveat: the $108 / $137 / $162 anchors are multiples of the FY2027 consensus EPS distribution — 15.9x the low of $6.776, 20.0x the mean of $6.856, and 23.2x the high of $6.986 — cross-checked against FY2026 and FY2028. Stated arithmetic, not a discounted cash flow. The base is sensitivity-disclosed: 18x gives $123, below spot; 22x gives $151. The multiple band is wider than the entire analyst EPS range, which is the honest statement of where the uncertainty sits on a regulated utility.
• Cash-flow note: every free-cash-flow figure in this dive is recomputed from filed operating cash flow less filed construction expenditures, never taken from the vendor. The vendor's freeCashFlowYieldTTM of 12.807%, priceToFreeCashFlowRatioTTM of 7.808x, evToFreeCashFlowTTM of 13.738x, freeCashFlowPerShareTTM of $16.462, freeCashFlowToEquityTTM of $6,676.9M, capexToDepreciationTTM of 0.371, capexToRevenueTTM of 0.0596, capexToOperatingCashFlowTTM of 0.176, capitalExpenditureCoverageRatioTTM of 5.670x and dividendPaidAndCapexCoverageRatioTTM of 2.247x are not used anywhere in this dive.
• Timing: second-quarter 2026 results were released 2026-07-30, five days before this dive, and MISSED the consensus EPS estimate by 8.1% ($1.36 against $1.48) while full-year guidance was RAISED to $6.25-$6.55 from $6.15-$6.45. The next print is 2026-10-29, 86 days away, on consensus EPS of $1.98 and revenue of $6,552M. The most recent insider filings are dated 2026-08-04, the day of this dive, and are both mechanical tax withholdings on vesting restricted stock units.
• Accessibility note: no information in this dive is conveyed by colour. All emphasis is carried by bold text, table structure and explicit labelling.
• Not investment advice. Independent research, educational and informational only, never personalised. No recommendation to buy, sell or hold any security is made to any person.
• Version: 2026-08-04-full.