NiSource

Phase A — Understand the business

Lens 1 · Company Overview

NiSource Inc. (NYSE: NI, incorporated DE, HQ Merrillville, Indiana — the 325,000 sq ft Southlake Complex) is a fully regulated US gas & electric utility holding company serving ~4.0 million customers across six states; 478,533,171 shares out and 14,211 holders of record as of 2026-02-04. Two reportable segments:

• Columbia Operations — natural-gas local distribution companies (LDCs) across Ohio, Pennsylvania, Virginia, Kentucky and Maryland. FY2025 operating revenue $3,343.3M, operating income $895.1M, ~2.43 million gas customers.
• NIPSCO Operations — Northern Indiana Public Service Company, a combined gas + electric utility in northern Indiana. FY2025 operating revenue $3,308.5M, operating income $938.1M; ~0.5 million electric customers + ~0.88 million gas customers.

Consolidated FY2025: operating revenue $6,642.2M, operating income $1,835.3M, net income to NiSource $929.5M (total net income $1,012.6M incl. NCI), diluted EPS $1.95.

How it actually makes money: the classic regulated-utility model — earn an allowed return on rate base (invested capital) set by state commissions (the IURC in Indiana; PUCO and counterparts in the Columbia states). The commodity itself (gas, fuel, purchased power) is a pass-through tracker — the majority of energy costs in both segments "are tracked costs that are passed through directly to the customer, resulting in an equal and offsetting amount". So the revenue line swings ~$1.2B YoY on gas prices and weather (2025 rev +$1,187M vs 2024) with essentially zero net-income impact; what drives earnings is rate-base growth and tracker recovery on capital programs. FY2025 net income grew on "new rates from base rate proceedings and regulatory capital programs" — Columbia +$178.9M and NIPSCO +$324.2M of revenue uplift. In 2025 the company added 24,000 net customers, took rate orders in four cases (Columbia MD/PA/VA + NIPSCO Electric), and invested $1.6B in core infrastructure modernization.

Contract structure / key terms: historically pure cost-of-service regulation with automatic recovery riders (IRP, CEP, TDSIC, SAVE, FMCA — see Lens 4). The structural break is the ADS Contract — a 15-year, commercially-negotiated electric-supply agreement entered Sept 2025 between NIPSCO and Amazon Data Services (a wholly-owned Amazon.com subsidiary, with Amazon.com Inc. — investment-grade — guaranteeing ADS's payment obligations). Capacity commitment begins 2027 and ramps to 2,400 MW by end-2032; NIPSCO will construct up to 3,000 MW of dispatchable generation to serve it. Crucially, the charges are "structured to provide us with a return of our invested capital over the fifteen-year initial term" plus an unlevered IRR within a defined range that NiSource expects, over the contract life, to EXCEED the realized return of NIPSCO's current electric operations. That last clause is the whole bull case in one sentence — and it is the company's own disclosure, not a sell-side estimate.

Capital allocation profile: capital-intensive, FCF-negative-by-design. FY2025 cash from operations $2,362.3M vs capex $4,051.6M — the ~$1.7B gap (plus dividends) is plugged with debt + equity + minority-interest capital. Pays a quarterly dividend of $0.300/share ($1.20 annualised), declared Jan 22 2026, ~2.5-2.6% yield. No buybacks (zero shares repurchased in Q4 2025) — every dollar and then some goes into rate base.

Lens 2 · Supply Chain

A regulated utility's "supply chain" is the physical fuel-and-asset chain plus the construction/equipment chain. Named stakeholders:

Upstream (fuel & commodity):

• Natural gas — procured on behalf of customers under state-approved Gas Cost Adjustment (GCA) trackers; pass-through, no margin. NIPSCO holds pipeline service agreements (capacity/transport/storage) expiring 2030-2044, $2,829.1M of contractual pipeline obligations.
• Coal — NIPSCO still runs coal at R.M. Schahfer (Units 17 & 18) and Michigan City; coal hauled by three rail operators under minimum-payment contracts through 2026-2028. Schahfer was forced to keep running 90 days past its planned retirement by a DOE §202(c) emergency order (issued Dec 2025, through March 2026) to meet an "energy emergency across MISO's North and Central regions"; the order authorizes cost recovery under 16 U.S.C. §824a(c) and the 10-K warns it "could continue to be issued every 90 days… for the foreseeable future".
• Purchased power / capacity — sourced via the MISO day-ahead/real-time market plus renewable PPAs. As of YE2025, owned renewable + storage in service total 1,950 MW + 101 MW nameplate, plus 1,200 MW of renewable PPAs in service (expiring 2038-2045). In 2025 NIPSCO's own units met 55.4% of system load; the rest came from PPAs + MISO.

Midstream (the company): NIPSCO transmission (~3,000 circuit miles, 69kV-765kV), 65 transmission + 240 distribution substations, ~312,500 poles; functional control of transmission ceded to MISO under the Open Access Transmission Tariff.

Construction / equipment chain (the growth bottleneck):

• GenCo Generation Assets — GenCo plans two 1,300 MW CCGTs + 400 MW of battery storage, COD between 2028 and 2032, aggregate Contract Assets (generation + related transmission) ~$7 billion.
• EPC + turbine suppliers — GenCo has signed EPC contracts and equipment-supply contracts "including a contract to acquire turbines"; EPC terms provide some cost-overrun protection, with residual overruns "shared by ADS and NIPSCO (for transmission) and GenCo (for generation)".
• Renewable developers under Build-Transfer Agreements (BTAs) — Dunn's Bridge II, Fairbanks, Gibson, Appleseed, Carpenter, Templeton — paid in milestone tranches (e.g. Fairbanks: $336.6M mechanical + $141.4M substantial + $3.6M final completion in 2025).
• Long-lead equipment — $373.8M of advance deposits made in 2025 to secure long-lead items for data-center generation; the 10-K flags "elevated material and supply costs… driven by increased demand and tariffs".

Downstream (end customers): ~4M residential/commercial/industrial ratepayers (captive, exclusive electric service territories) + the new hyperscaler large-load segment (Amazon via ADS, Alphabet). NIPSCO electric is unusually steel-exposed: ~49.3% of industrial MWh sales go to steel-related industries — a cyclical tail unusual for a regulated utility.

Chokepoints / single-source dependencies: (1) MISO interconnection queue + resource-accreditation rules — the 10-K states "MISO's resource accreditations for renewables and storage remain uncertain," which "may [make it] necessary to evaluate changes to our previously communicated resource timelines"; (2) EPC/turbine/equipment lead times and tariffs for the data-center build; (3) the IURC as the single regulatory gatekeeper for cost recovery in Indiana — every data-center contract and the related NIPSCO↔GenCo PPA is "pending IURC approval."

Lens 3 · Competitive Advantages (moats)

The moat is the regulated-utility moat, sharpened by one genuinely differentiated asset.

• Regulated monopoly / exclusive franchise (the base moat). NIPSCO electric customers "do not have the ability to choose their electric supplier". Gas LDCs face theoretical "choice" programs but in practice own the pipe — the company keeps the distribution margin regardless of who supplies the molecule. Switching cost for a captive ratepayer is effectively infinite; bargaining power over customers is structurally high (rates set by commission, not market).
• Constructive multi-jurisdiction regulatory framework with automatic recovery riders. Columbia of Ohio (IRP/CEP), NIPSCO (TDSIC, GCT, FMCA), Columbia of Virginia (SAVE), Columbia of Kentucky (SMRP) — a stack of trackers that recover capital between rate cases. This converts a ~$21B base-capex plan into largely pre-approved, recoverable rate base. Spreading across six states diversifies regulatory risk (no single-commission concentration).
• The differentiated asset — first-mover hyperscaler power supply via the ring-fenced GenCo, on terms that beat the regulated return. NIPSCO sits on excess transmission infrastructure in a corridor hyperscalers want (Northern Indiana: cheap land, supportive politics, MISO interconnection). The 10-K is explicit that the ADS Contract's unlevered IRR is designed to land in a defined range that "result[s] in an overall realized return greater than that of NIPSCO's current electric operations, driven by execution and financing" — i.e. the growth capital earns more per dollar than the rate-base business it sits beside, which is the rarest thing in utility-land. The structure also makes GenCo/NIPSCO Holdings II "the exclusive vehicles for all power, storage and generation requirements for data center customers within NIPSCO's service territory" so long as Blackstone holds ≥14.9% — an exclusivity right over a scarce resource (deliverable at-scale power) in a land-rush market.
• Capital scale + a marquee capital partner. Blackstone Infrastructure's minority stakes (NIPSCO Holdings II + Generation Holdings II) bring committed third-party equity — up to $1.325B (GenCo) plus increased NIPSCO Holdings II mandatory contributions — that lets NiSource fund a >$7B data-center build while protecting investment-grade ratings and limiting common-equity dilution.

Bargaining power over suppliers: moderate. As a large, investment-grade, multi-state buyer NiSource has scale, but it is a price-taker on gas/fuel (pass-through anyway) and currently a price-taker on EPC/turbines/equipment in a tariff-inflated, demand-squeezed construction market.

Durability verdict: the base moat is wide and boring (regulation). The data-center moat is real and better-than-regulated on return, but it is contractual, not structural — it rests on the ADS Contract's terms and the hyperscalers' continued demand, both of which can change (Lens 13). The exclusivity covers NIPSCO's territory only — it is no defence against Texas or Virginia winning the next campus.

Lens 4 · Segments

All figures ``.

By segment (operating income — the number that matters since revenue is gas-price noise):

NIPSCO is the engine: operating income compounded from $541.7M (2023) → $938.1M (2025), a ~31% 2-yr CAGR, driven by rate relief (+$324.2M revenue from new rates/DSM in 2025) and depreciation recovery on the generation transition.

Within NIPSCO (sub-lines):

• NIPSCO Electric revenue $2,208.9M (2025) vs $1,913.6M (2024), +15.4%; industrial ~$581.3M of that, steel-heavy. Sales 16,538.8 GWh.
• NIPSCO Gas revenue $1,099.6M (2025) vs $838.4M (2024), +31.2% — but heavily weather/commodity-driven (HDD 5,936 vs 4,975).

By geography: ~100% US. Columbia = OH/PA/VA/KY/MD; NIPSCO = northern Indiana. Indiana is the strategic growth state (data centers, steel, the entire generation transition).

Drivers of the trend (accelerating): the consolidated op-income +26% is not organic demand — gas/electric volumes are roughly flat-to-weather. It is rate-base monetisation: new base-rate orders, capital trackers, and depreciation recovery on the ~$4B/yr capex. This is the whole bull case in one line — earnings growth comes from spending capital the commission lets you recover, and the data-center contracts let you spend far more capital (at a higher return) than a normal LDC ever could.

---

Phase B — Measure performance

Lens 5 · Earnings Result (latest print: Q1 2026, filed 2026-05-06)

Q1 2026 actuals:

• Total operating revenue $2,363.1M (vs $2,183.2M Q1'25, +8.2%)
• Operating income $819.2M (vs $759.4M, +7.9%)
• Net income to NiSource $507.1M (vs $474.8M)
• Diluted EPS $1.06 (vs $1.00 GAAP Q1'25, +6%); adjusted EPS $1.06, +8% YoY
• Diluted avg shares 480.9M (vs 472.5M) — ~1.8% share-count growth YoY (the ATM dilution, visible)

vs consensus: adjusted EPS $1.06 beat the ~$1.05 estimate; revenue of ~$2.36B missed a higher Street figure — but revenue is a pass-through/weather artefact, so the miss is cosmetic and the EPS beat is what matters.

What drove it: rate relief + capital recovery (O&M and D&A both up on the larger asset base).

The flag worth circling — interest expense. Q1 2026 interest expense, net $191.6M vs $132.8M in Q1'25 — +44% YoY. This is the cost of the debt-funded capex ramp arriving in the P&L, and it is the single most important line to track quarter-over-quarter, because the whole thesis is "can rate-base earnings (at the new, higher data-center return) outrun the rising interest + share count of funding them?" Corroborating evidence the build is debt-funded in real time: on May 11 2026 NiSource priced $1.25B of senior notes — $500M at 4.750% due 2031 + $750M at 5.300% due 2036 (net ~$1.24B, general corporate purposes). Note the ~4.75-5.30% coupons sit well above the legacy book — the marginal cost of capital funding this growth is materially higher than the average, which is exactly why the interest line is inflecting.

Guidance / tone: management reaffirmed FY2026 adjusted EPS guidance of $2.02–$2.07 and raised the long-term adjusted-EPS CAGR to 9-10% through 2033 (from 8-9%), saying it tracks toward the high end through 2030, underpinned by 9-11% rate-base growth. Tone is confidently expansionary.

Balance-sheet flags: total consolidated indebtedness $16,213.5M at YE2025; debt-to-cap ~51% vs a 70% covenant ceiling; net available liquidity ~$2.0B; cash only $110.1M (utilities run thin cash). Ratings BBB+/Baa2/BBB, all stable. A downgrade triggers $150.2M of collateral posting.

Market reaction: stock rose ~1.7% on the Q1 print to ~$49 premarket, "reflecting investor confidence in the company's strategic partnerships with Amazon and Alphabet".

Lens 6 · Earnings Calls (sentiment trend)

No transcripts on disk (transcripts/ empty) — this lens is ``.

Management focus (Q1 2026 call): the story has fully re-centred on the GenCo data-center engine. CEO Lloyd Yates and CFO Shawn Anderson framed the quarter around: (1) the raised 9-10% CAGR; (2) the contracted hyperscaler pipeline; (3) a "balanced funding plan through 2030." CCO Michael Luhrs anchored the $1.4B customer-savings mechanism as a "defined mechanism within each special contract".

The pipeline they want you to focus on: ~4 GW signed (incl. the 340 MW Alphabet agreement + 400 MW Amazon expansion), 3 GW in strategic negotiation, 2 GW in development → ~9 GW total potential. CFO Anderson stressed guidance "only includes signed customer contracts" — the negotiation pipeline is upside not in the numbers.

Quantified GenCo earnings ramp: GenCo EPS contribution $0.25–$0.35 in 2030, $0.40–$0.60 in 2033; annual equity issuance $400–$600M via ATM; target FFO-to-debt 14-16%.

Tone shift over time: the multi-quarter arc is a deliberate re-rating campaign — from "steady 6-8% Indiana utility" (the post-Massachusetts-exit identity) to "9-10% AI-infrastructure compounder." Recurring new phrases: "ring-fenced GenCo," "special contracts," "customer savings," "balanced funding." What they've stopped emphasising: the legacy gas-LDC modernization story (still the cash base, now the supporting actor).

Lens 7 · Comps

Peer set: mid-cap, predominantly-regulated US gas/electric utilities with rate-base growth stories. Multiples are ``, dated ~June 2026; where a clean figure wasn't sourced I mark n/a rather than fabricate.

EV/Sales, EV/EBIT, trailing P/E, 5-yr avg ROE: NI trailing P/E ~23-24x; a clean sector EV/EBITDA and a 5-yr ROE series were not cleanly sourced for the full peer set — n/a rather than guess. (NI structural ROE for a regulated utility runs ~9-11% on allowed returns; treat as `` directional only, not a sourced figure — and note the data-center book is disclosed to earn above that.)

Read: NI trades at the top of the regulated-utility forward-P/E band (~22-23x) — a premium to DTE (~18x) and WEC (~20.5x), roughly level with AEE/LNT. The premium is the market pricing the 9-10% CAGR vs peers' 6-8% plus the disclosed above-regulated return on the incremental data-center capital. That is defensible: ~150-300bps of extra growth at a higher incremental ROE justifies a ~2-4 turn P/E premium. But the comps also say the easy money (re-rating from a peer multiple) is largely made — sell-side consensus price target is only ~$47.25-47.91 (avg "Buy", range ~$44-50; Barclays $49 OW), i.e. ~flat-to-low-single-digit upside on the Street. From here it's an execution/compounding story, not a cheap-stock story.

Lens 8 · Stock-Price Catalysts (last ~5 years, >5% moves)

Mostly ``; the structural events corroborated by filings.

• 2018-2020 — Greater Lawrence / Merrimack Valley gas disaster. Columbia Gas of Massachusetts over-pressurised a low-pressure system (Sept 2018), killing one person, injuring ~two dozen, damaging 100+ buildings; NTSB blamed "deficiencies in management and oversight." Settlements: $143M class action + a ~$53M federal fine + a $56M state penalty; NiSource pled out and exited Massachusetts, selling Columbia Gas of MA to Eversource for $1.1B (2020). The defining negative catalyst of the prior cycle and the reason the company de-risked into a pure Columbia+NIPSCO footprint. Pattern lesson: gas-distribution catastrophe is the tail that actually moves this stock hard to the downside.
• 2020-2023 — generation-transition / rate-case grind. Coal-retirement plans (2018/2021/2024 IRPs), renewable BTAs, serial base-rate orders → steady, unexciting re-rating. Stock behaved like a bond proxy — sensitive to the 10-year and Indiana rate-case outcomes.
• Dec 2023 — Blackstone NIPSCO minority deal. Blackstone bought into NIPSCO Holdings II, validating asset value and pre-funding capex.
• 2025 (mid-year) — the data-center pivot. The ADS (Amazon) Contract (Sept 2025) + GenCo formation + Blackstone's Generation Holdings II 19.9% stake (Oct 2025) reframed the story. "NiSource stock jumps 3% on data center deals".
• Apr 16 2026 — Alphabet/Google deal + Amazon expansion. New long-term supply agreement with an Alphabet subsidiary (service summer 2026) + expanded Amazon agreement; ~$1.4B of customer savings announced; stock rose on the print.
• May 6 2026 — Q1 beat + CAGR raise to 9-10%. Stock ~+1.7%.
• NEAR-TERM (June 2026) — IURC approval of the foundational ADS settlement is expected by ~June 2026, with the newer Alphabet/Amazon deals on a 90-120 day expedited review. This is the next discrete de-risking event — approval converts the whole GenCo construct from "pending IURC approval" to live, and is the gating catalyst on the contracted EPS ramp.

What the market actually reacts to: historically, gas-safety catastrophe (down, hard) and rate-case/rate-base news (slow grind). Newly, hyperscaler contract announcements + the CAGR guidance are the up-catalysts — the stock has acquired an AI-infrastructure beta it never used to have. Management itself warns the stock "may experience increased volatility… as a result of any actual or perceived slowdown in the adoption of artificial intelligence technology".

---

Phase C — Judge people & books

Lens 9 · Management

• CEO — Lloyd M. Yates (President & CEO since Feb 2022). Career utility operator (ex-Duke Energy, ex-PNM Resources board); came in post-Massachusetts to run the de-risked, growth-pivot phase. Track record: executed the safety-first reset, the Blackstone partnerships, and the data-center pivot — arguably one of the better strategic moves in mid-cap utility-land (turning excess Indiana transmission into a hyperscaler annuity that earns above the allowed return). Skews professional manager / capital-allocation operator, not founder.
• CFO — Shawn Anderson (EVP & CFO) — owns finance, accounting, strategy, risk; architect of the "balanced funding plan" (ATM + Blackstone minority + debt) and the FFO/debt 14-16% discipline.
• Tenure & skin in the game: professional-manager ownership — insider holdings are modest (typical for a regulated utility); insider-transactions.csv not on disk, so n/a on precise insider %. No founder block.
• Capital-allocation history: the defining decisions — (1) exiting Massachusetts after the disaster (right call, removed a tail risk); (2) bringing Blackstone in as a minority partner rather than diluting common holders to fund NIPSCO/GenCo (shareholder-friendly given the alternative); (3) structuring data-center deals as customer-savings-positive (~$1.4B of bill credits over 15 years) to keep regulators and ratepayers aligned — a genuinely clever political/regulatory hedge; and (4) negotiating the ADS economics to an unlevered IRR the 10-K says should exceed the regulated return, i.e. growth that's accretive to blended ROE, not dilutive. They reinvest, they don't buy back (the $4B capex consumes everything and then some).
• Red flags on management: none acute. The Massachusetts disaster predates Yates and was resolved. Comp and related-party items are standard-utility; no AAER/LR (Lens 10). The honest critique is execution risk on a vastly larger, more complex capital program — they are now running a ~$28B build with hyperscaler counterparties, EPC/turbine supply chains and MISO interconnection, a structurally harder job than running LDC rate cases, and the team's data-center-construction track record is by definition short.

Lens 10 · Forensic Red Flags

Forensic-analyst pass across the income statement, balance sheet, and cash flow. Net assessment: LOW forensic risk — a transparent, audited, rate-regulated utility with clean trackers and no enforcement history. The real risks are structural (leverage, off-balance-sheet, ASC 980), not fraud.

• Revenue recognition: low risk — commodity revenue is a tracked pass-through matched to expense; the watch item is the ADS Contract's accounting under ASC 980 (rate-regulated). The 10-K explicitly warns that significant construction overruns, capacity shortfalls, liquidated-damages events, or early termination "could either preclude ongoing application of ASC Topic 980 or result in an immediate disallowance and impairment of the Contract Assets". That is the single biggest accounting tail — a ~$7B asset base whose regulatory-asset treatment hinges on the contract performing and on NIPSCO/GenCo not over-spending versus the contemplated cost.
• Cash flow vs earnings: CFO $2,362.3M comfortably exceeds net income $1,012.6M (D&A-heavy utility), so earnings quality is fine — but FCF is deeply negative by design (capex $4.05B >> CFO), so the story is entirely about access to capital, not cash generation. Investing outflow $4,524.1M in 2025 (+~$1.3B YoY).
• Receivables/inventory: no flag — gas-cost over/under-recovery is a regulatory-asset mechanism, not a collectibility problem.
• Leverage / off-balance-sheet: the area to watch. Total debt $16.2B, plus finance leases (~$274.0M excluded from the LT-debt figure), plus PPA termination charges "that could be material," plus BTA guarantees, plus the VIE consolidation of Generation Holdings II. Management openly says it "may consider… off-balance-sheet arrangements in the form of BTAs to support maintenance of our investment grade credit ratings" — i.e. structuring to keep leverage metrics inside ratings thresholds. Legitimate, but it means the headline debt understates total economic obligations. Total contractual obligations $32,086.3M. The May-2026 $1.25B notes (4.75%/5.30%) are the leading edge of years more issuance.
• Minority interest / VIE complexity: Blackstone's stakes create real complexity — $83.1M of FY2025 net income attributable to NCI; large minority-interest contributions flowed in over the period. NiSource is attributed 80.1% of Generation Holdings II P&L; Blackstone gets 19.9%, two of seven board seats, and approval rights over major actions. Not a red flag, but a governance-sharing the equity story must respect.
• SBC / non-GAAP: modest; the adjusted-EPS bridge (GAAP $1.95 → adjusted) is standard utility normalisation. No evidence SBC flatters the non-GAAP number materially.
• Goodwill/intangibles: annual goodwill test (May 1 2025) passed a qualitative "step 0".

Regulatory findings (required sub-section).

• SEC Litigation Releases / AAERs: none. Verified via SEC EDGAR EFTS (LR + AAER), period 2021-06-20 → 2026-06-20 — total_sec_findings: 0.
• 10-K Item 3 (Legal Proceedings): cross-references Note 19.C, which states only routine claims and that the ultimate outcome of such proceedings is expected to be, individually or in aggregate, "not material at this time". No securities class action, no active enforcement.
• Environmental: routine. Total environmental remediation liability $82.6M (YE2025); MGP (manufactured-gas-plant) liability $75.5M across 41 identified sites, mostly rate-recoverable; $48.9M accrued in 2025 for the EPA Legacy CCR Rule (coal-ash). All within normal utility bounds and expected to be recovered through rates.
• Non-SEC enforcement (web): the material historical item is the 2018 Merrimack Valley gas disaster — federal criminal plea, ~$53M fine, $143M class action, $56M state penalty, MA exit. Resolved; no open federal action found in current search.
• Conclusion: No material current regulatory or legal findings — verified via SEC EDGAR EFTS (LR, AAER → 0), web search, and 10-K Item 3 / Note 19 as of 2026-06-20. The forensic risk in NiSource is operational-catastrophe + ASC-980 impairment, not accounting integrity.

---

Phase D — Project & stress-test

Lens 11 · Forward Projection (FY2026 → FY2028)

Built bottom-up from management's own guidance + the disclosed GenCo ramp. Output is ``; every input labelled. No forecast.ts logged (watchlist breadth mode — only log when genuinely committing the base case).

Anchors:

• FY2026 adjusted-EPS guidance $2.02–$2.07, midpoint $2.045.
• Long-term adjusted-EPS CAGR 9-10% through 2033, tracking high end through 2030; supported by 9-11% rate-base growth and a ~$28.0B 2026-30 consolidated capital plan (~$19.4B/$21B base utility + ~$7.0-7.6B GenCo/data-center).
• GenCo EPS contribution $0.25-0.35 (2030), $0.40-0.60 (2033).
• Equity issuance $400-600M/yr (~1-2.5% annual share-count dilution at ~$47) ` on the dilution %]. Marginal debt cost ~4.75-5.30%.

Base case (deliver guidance midpoint, then ~8.5% CAGR — middle of base + GenCo uplift):

• FY2026 EPS ≈ $2.05
• FY2027 EPS ≈ $2.22
• FY2028 EPS ≈ $2.41

Bull case (high-end ~10% CAGR, GenCo pipeline converts faster, rates supportive, IRR realises above-regulated):

• FY2026 $2.07 → FY2027 $2.28 → FY2028 $2.51

Bear case (base-plan-only ~6%, GenCo slips/partly impaired, higher interest drag + more equity dilution than planned):

• FY2026 $2.02 → FY2027 $2.13 → FY2028 $2.24

Valuation cross-check: at ~$47.2 and base FY2026 $2.05, NI trades ~23x current-year, ~21x FY2027. For 8-10% EPS growth + a ~2.5% yield, that is full but not absurd — total-return math ~10-12.5%/yr if the CAGR delivers and the multiple holds. The multiple is the risk: a de-rate to a peer ~19x on any data-center wobble is a 15-18% price hit even with EPS intact. And sell-side targets ($47.25-50) already sit roughly at spot, so the Street agrees the cheap re-rate is done.

(Brier forecast not logged this run. If committing the base case later: NI FY27 adj EPS >= $2.20, p≈0.60, resolves 2027-12-31.)

Lens 12 · Bull vs Bear

Bull case. NiSource is a regulated-utility chassis with an AI-infrastructure turbocharger, and the turbocharger is de-risked in four ways most "AI power" stories aren't: (1) it's housed in a ring-fenced GenCo with a Blackstone partner pre-committing capital ($1.325B+), so common holders aren't bearing the full build; (2) the contracts are structured customer-savings-positive (~$1.4B of bill credits over 15 years), buying regulatory and political durability the pure-merchant data-center plays lack; (3) it has contractual exclusivity on data-center power in NIPSCO's territory and a ~9 GW pipeline against ~4 GW signed — visible, contracted growth, not hopium; and (4) — the underrated one — the ADS economics are disclosed to earn an unlevered IRR above NIPSCO's regulated return, so the growth capital is accretive to blended ROE rather than just bigger. The result is a 9-10% EPS CAGR with utility-grade risk, on top of a ~$21B base plan that grows rate base 9-11% regardless of how the data-center upside lands. The secular tailwind (US power demand inflecting after two decades flat, hyperscaler land-rush, reshoring) is real and Indiana is squarely in its path. Upside surprise if the 3 GW negotiation pipeline converts into guidance, and a discrete near-term de-risk if the IURC approves the ADS settlement (~June 2026).

Bear case (permanent-impairment risks).

1. Hyperscaler concentration + the termination/reduction option. The Contract Assets, fully delivered, are "approximately equivalent to the generating capacity of all NIPSCO's existing generating assets" — and ADS can terminate for convenience, or exercise a one-time option (by March 31 2029) to halve committed capacity to 1,200 MW from January 31 2032. Reimbursement on termination/reduction is capped at signing-date cost estimates and depends on ADS/Amazon honouring it — so cost overruns on a cancelled or halved project land on NiSource. A single counterparty deciding AI capex was overbuilt could strand a multi-billion-dollar asset and trigger an ASC 980 impairment.
2. Funding the build in a higher-for-longer world. Interest expense already +44% YoY (Q1'26) and the marginal note now costs ~4.75-5.30%. The plan needs continuous debt + $400-600M/yr equity while holding FFO/debt 14-16% and investment grade. A capital-markets shock, a downgrade ($150.2M collateral call) or an AI-sentiment-driven equity de-rate makes the equity portion more dilutive exactly when it's most needed — the 10-K names this explicitly.
3. Execution / construction risk at unprecedented scale. ~$28B over five years; EPC + turbines in a tariff-inflated market; MISO interconnection/accreditation uncertainty; and a coal plant the DOE won't let them close on schedule (Schahfer §202(c), renewable every 90 days). Delays trigger ADS liquidated damages that offset billings.

Pre-mortem (18 months out, thesis broke): AI-capex digestion arrives, a hyperscaler signals it will not expand (or pre-commits to the Mar-2029 reduction option), the Street re-rates NI from ~22x back to a ~18-19x peer multiple, the GenCo EPS contribution gets pushed out, and the stock is $38-40 even though base-utility EPS is fine. The break is multiple compression on sentiment, not an earnings collapse.

Are multiples too high? At the top of the peer band (~22-23x fwd), the easy re-rate is done — but not bubble-priced for a credible 9-10% grower whose incremental capital earns above the allowed return. Risk is asymmetric to the downside on sentiment.

Contrarian view (what the market is refusing to see): consensus treats NI as an "AI-power" stock and gives it AI-beta — but the downside is far more protected than a pure data-center name because ~70-80% of the rate base is boring recoverable LDC/utility capital, and the data-center deals are bill-credit-positive, so even a regulator/ratepayer backlash cuts the other way (customers like these deals). The asymmetry the market underrates is that NI keeps a 6-8% utility floor even if every speculative gigawatt evaporates — you are paid a 2.5% dividend to hold a free option on the AI-power upside.

Lens 13 · Devil's Advocate (short-seller)

Dismantling the bull case.

• What structurally breaks the model: the entire incremental-growth premium rests on counterparties NiSource does not control and a contract that lets them leave or halve. "Customer concentration risk" is the 10-K's own header. NIPSCO is roughly doubling its generation for one cohort of hyperscalers in one technology cycle (AI) that "developed rapidly… and continue to develop" — management literally "cannot predict the rate at which… these emerging technologies will be broadly adopted." You are underwriting Amazon's and Google's capex plans through an Indiana utility.
• Where revenue is concentrated / what shifts: the speculative EPS ($0.25-0.60 by 2030-33) is concentrated in GenCo / the ADS + Alphabet contracts. If AI-data-center power demand plateaus (efficiency gains, a model-scaling slowdown, on-site generation/SMRs becoming preferred — the 10-K names "competing energy technologies could become a preferred source"), the pipeline (3 GW negotiating, 2 GW developing) never converts and the 9-10% CAGR quietly reverts to the 6-8% base. Note the "above-regulated IRR" is expected, not guaranteed — "our realized return may be impacted by construction costs, operating performance, financing costs and other variables"; higher-for-longer financing eats that spread first.
• Why the moat may be weaker than bulls think: the "exclusivity" is only over NIPSCO's service territory. Hyperscalers choose sites across the US and abroad; NiSource "face[s] competition from utilities and other energy sources across the United States and abroad" and customers weigh "the ability of utilities or power providers to deliver electricity quickly and at scale". No moat against Texas/ERCOT or Virginia/PJM winning the next campus.
• Most dangerous competitor bulls underestimate: not another utility — it's behind-the-meter / on-site generation and SMRs, plus the broader set of states competing for the same load. If hyperscalers vertically integrate power, the utility-as-landlord thesis erodes.
• Worst capital-allocation / accounting concerns: the VIE/off-balance-sheet structuring to defend ratings and the ASC 980 impairment cliff — a structure explicitly engineered to keep leverage metrics inside thresholds is, by construction, financially fragile to a shock. And the Merrimack Valley precedent: gas-distribution operational catastrophe is a real, demonstrated tail for this exact company.
• What must hold for today's price: (1) ADS/Alphabet don't terminate or de-scope (and ADS doesn't pre-signal the Mar-2029 halving); (2) the build comes in on-time/on-budget despite tariffs and MISO; (3) capital markets stay open at investment grade; (4) the multiple stays ~22x. Break any one and you have a ~$38-40 stock.
• If growth disappoints 20-30%: EPS CAGR to ~6% + a de-rate to ~18x → roughly $38-40, ~15-20% downside, with the 2.5% dividend as the only cushion.
• The single permanent-impairment scenario: a hyperscaler terminates/halves the ADS Contract mid-build, reimbursement caps don't cover overruns, ASC 980 treatment is lost, and NiSource books a multi-billion impairment on stranded generation it can't re-contract. Plausibility: low-to-moderate over the build window — the contracts have caps, an Amazon.com parent guarantee, and AI demand is currently insatiable — but it is the real fat tail, and it's not in any base case.

Lens 14 · Management Questions (ordered by information value)

1. Under the ADS one-time capacity-reduction option (to 1,200 MW from Jan 2032) and the for-convenience termination right — what is the dollar gap between the signing-date reimbursement caps and your current all-in Contract-Asset cost estimate, i.e. how much capital is genuinely at risk if Amazon walks or halves mid-build?
2. The 10-K says the ADS unlevered IRR is expected to exceed NIPSCO's regulated electric return — what is the range, and how much does ~4.75-5.30% marginal debt + the planned equity issuance compress that spread before it's no longer accretive to blended ROE?
3. For the Contract Assets, what specific events would discontinue ASC 980 treatment, and what is the maximum impairment you'd recognise in that scenario?
4. How firm are the hyperscalers' commitments economically beyond the Amazon.com parent guarantee — are there take-or-pay floors, and what credit protections survive a counterparty downturn?
5. The 3 GW "in negotiation" — what's your realistic conversion rate and timeline, and what's not in the 9-10% CAGR that could be?
6. What FFO/debt and equity-issuance trajectory keeps you investment-grade if you convert even half the 3 GW pipeline — and at what point does the funding plan require common equity beyond the $400-600M/yr ATM?
7. With Schahfer under a renewing DOE §202(c) order and MISO accreditation uncertain, what is your downside case for generation-transition timing and cost recovery?
8. EPC + turbines in a tariff-inflated market — what cost-overrun protection do you actually have, and how much of any overrun is contractually shared with ADS vs borne by GenCo/NIPSCO?
9. How do you keep the IURC and ratepayers aligned as the data-center capital base grows — does the ~$1.4B savings mechanism scale with every new contract, and is there backlash risk?
10. Blackstone holds two of seven Generation Holdings II board seats + major-action approval rights — what decisions can they block, and how does that constrain strategic flexibility?
11. Steel is ~49% of NIPSCO industrial electric sales — how do you think about that cyclical exposure layered on top of data-center concentration?
12. What's the plan B for stranded generation if a campus is cancelled — can the two 1,300 MW CCGTs + 400 MW storage serve the existing grid, and at what return haircut?
13. How much off-balance-sheet obligation (BTAs, PPAs with material termination charges, ~$274M finance leases) sits outside the headline $16.2B debt, and how do rating agencies treat it?
14. Given the Merrimack Valley precedent, what has structurally changed in safety/oversight since 2018?
15. At ~22-23x forward you're priced at the top of the peer band with the Street's target roughly at spot — what EPS CAGR justifies that, and what's your message to holders if the multiple compresses to peers on an AI-sentiment wobble?

---