Solaris Energy Infrastructure, Inc. (NYSE: SEI) — A Sand-Logistics Microcap Reborn as a Levered Bet on Renting Turbines to Hyperscalers, at Its Richest-Ever Price

Independent fundamental research. The analytical body (Sections 1–15) takes no position and names no price target; the sole exception is the clearly-labeled Claude’s Take block below.

⚡ Claude’s Take

This is the author’s own independent opinion and general information only. It is not investment advice and is not a recommendation to buy or sell any security. The analytical body (Sections 1–15) below takes no position, names no price target, and remains recommendation-free; only this block states a view.

Verdict: HOLD / AVOID-adding-here — a genuinely transformed, contracted-cash-flow growth story wrapped in a levered, serially-dilutive, hyper-concentrated capital monster, at the 92nd percentile of its own valuation history. Not a clean short. Constructive re-entry zone ~$45–58. Tag: “They’re renting the picks in a gold rush — with borrowed money, and the early backers are selling into your bid.”

Two years ago Solaris was a $6 oilfield sand-silo company that earned a negative operating margin in 2021. Today it is a ~$5.5B-equity behind-the-meter power platform that rents natural-gas turbines to three investment-grade hyperscalers under 10–15-year contracts, with revenue that nearly doubled to $622M in FY2025 and a management scenario in which 3.1 GW of secured capacity “could well exceed $1 billion” of annual adjusted EBITDA. The transformation is real, not a story — Power Solutions went from 6% to 54% of revenue in one year at a 57% segment-EBITDA margin, the contracts are signed, the counterparties are creditworthy, and the demand driver (AI compute racing ahead of a grid that cannot interconnect it) is the most durable secular tailwind in the market. I do not doubt the demand.

What I doubt is the price, the balance sheet, and the durability of the returns — all three at once, which is why this is a HOLD and not a buy. Unlike its net-cash data-center-power cousins Powell (POWL) and Argan (AGX), Solaris is funding this build with a torrent of external capital: FY2025 capex was $647M against operating cash flow of $209M, free cash flow was −$438M, net debt went from $201M to $1.25B, it has issued ~$900M of convertible notes plus a fresh $1.3B of 6.375% senior notes in May 2026, and the CFO says $1B+ more capital must be raised in 2026–27. The equity is a moving target: the screen shows ~58M Class A shares, but the true fully-diluted economic count including Class B units, the GESA shares, and in-the-money converts is ~90–95M — so the real market cap is closer to $6.5B than the $4.2B a naïve screen implies, and it grows every quarter. Layer on a Tax Receivable Agreement that skims 85% of tax savings to pre-IPO insiders, a single customer at 47% of FY25 revenue (52% in Q1’26), a single turbine supplier at 74% of payables, and a compensation plan with no return-on-capital metric at all (relative-TSR PSUs that paid the founder $13.9M as the stock quadrupled), and you have a business whose contracted cash flows are high-quality but whose incremental-capital economics are exactly the >20% returns that, per Marathon’s capital-cycle logic, are drawing a flood of competitors (VoltaGrid’s 2.3 GW Oracle deal, ProEnergy, and management’s own-named “speculators in the turbine queue”) into the trade. The moat is turbine-slot scarcity and an operating track record — a first-mover timing advantage, not a structural franchise — and it is set to erode as OEM capacity expands 25–35%/yr from 2026.

The most honest tell is in the ownership register: while the founder and one director made small, genuine open-market buys (director A.J. Teague at ~$73), the pre-IPO sponsors — Yorktown, J Turbines, KTR — distributed $650M+ of stock into the rally through eight secondary offerings. Founders who built this to sell (Zartler did exactly that with Aris Water, sold to Western Midstream in October 2025) are handing the paper to the public at the top of the valuation range. The framing is a crowded, high-beta momentum trade at a capital-cycle peak (beta 1.97, RS-12m +133%, +135% annualized over the last year, already ~17% off its June-2026 high of $86), not a value setup and not yet a falling knife. I won’t short it — contracted IG cash flows, S&P SmallCap 600 index buying, sell-side euphoria (Needham $97, Wolfe Outperform), and the sheer violence of the move make the two-sided risk brutal for shorts. But at ~22x a still-ramping ~$336M adjusted-EBITDA run-rate — or ~7.5x a $1B number that requires another $1B+ of dilutive capital and flawless execution to reach — the stock already pays for the success. Conviction: medium. The single fact that flips me bullish: two or three additional 10-year IG contracts self-funded from internal cash and the Stateline distribution rather than fresh equity, proving the flywheel can turn without diluting — plus a maintained 20%+ unlevered return on the newest capital. The single fact that flips me bearish (toward avoid-entirely): a customer deferral/renegotiation on the 47%-concentration contract, a convertible/senior-notes refinancing at a materially worse cost as the capital cycle turns, or a gross-margin roll-down that signals the scarcity rent is competing away while the leverage and the share count are still climbing.

📈 Stock Price Action — Five-Year Event Map

Solaris is a five-year, ~12x round trip from oilfield microcap to AI-power momentum leader. From a $5–8 range that persisted through 2019–2023 (a no-growth sand-logistics business), the stock troughed near $6.18 (Jan 2024), inflected on the July-2024 Mobile Energy Rentals pivot, and compounded to an all-time high of $86.19 (June 18, 2026) before pulling back to $71.57 (July 10, 2026) — roughly 17% off the high, with a 52-week range of about $20 to $86. The move is a genuine fundamental re-rating (revenue 2x, EBITDA margin +13 points) layered with momentum and index-inclusion flows. Price moves below are FACT; attributed drivers are INTERPRETATION.

Cycle narrative. (1) For half a decade Solaris was an unremarkable oilfield-logistics name trading on completions activity. (2) The July-2024 MER acquisition — mostly a seller rollover into Class B units — created the Power segment and re-rated the equity off its lows. (3) By December 2024 the market grasped the data-center-power TAM, and the rename crystallized the story. (4) The April-2025 macro drawdown hit this high-beta name hard, then reversed. (5–7) A cadence of ever-larger multi-hundred-megawatt, 10–15-year hyperscaler contracts — Stateline, Hatchbo, and a third >600 MW deal — plus a Q1’26 beat and the $1.3B note deal drove the parabolic move to $86. (8) The most recent leg is a normal momentum give-back into a “measured” Q3 guide and continued equity issuance, cushioned by forced index buying. The stock now sits in the upper third of its all-time range, discounting substantial future execution.

1. Executive Summary

Solaris Energy Infrastructure (NYSE: SEI), Houston, is a two-segment energy-infrastructure company in the middle of one of the most complete business-model transformations in the market. Solaris Power Solutions (54% of FY25 revenue, growing explosively) owns and operates a fleet of natural-gas turbines that it leases, turnkey and behind-the-meter, to data-center/hyperscaler customers under long-duration contracts — a “molecule-to-electron” model spanning generation plus balance-of-plant (transformers, switchgear, SCRs, e-houses, last-mile gas). Solaris Logistics Solutions (46% of revenue, flat-to-declining) is the legacy oilfield business: mobile sand-silo and top-fill systems, last-mile logistics, and software for well completions — a cash-generative but structurally mature franchise being deliberately milked to fund the power build.

The numbers tell the story. Revenue: $159M (2021) → $320M (2022) → $293M (2023) → $313M (2024) → $622M (2025), with Q1’26 at $196M (a ~$785M run-rate). Adjusted EBITDA margin expanded from ~23% to ~35%. Power Solutions alone went from $38.6M of revenue in FY24 to $333.5M in FY25 at a 57% segment-EBITDA margin. The company has ~3.1 GW of secured turbine capacity, >2 GW under 10–15-year contracts with three investment-grade technology companies, and a management scenario in which the full 3.1 GW “could well exceed $1 billion” of annual adjusted EBITDA.

The tension is equally stark. This growth is enormously capital-intensive and externally funded: FY25 capex was $647M, free cash flow was −$438M, net debt rose from $201M to $1.25B, and the company has stacked ~$900M of convertible notes plus $1.3B of new senior notes on the balance sheet — with $1B+ more capital to raise in 2026–27. It is a Class A/Class B Up-C with a Tax Receivable Agreement paying 85% of tax savings to pre-IPO insiders, and the true fully-diluted economic share count (~90–95M) is roughly double the Class A screen count. Customer concentration is extreme (one customer 47% of FY25 revenue, 52% in Q1’26), as is supplier concentration (one turbine OEM = 74% of payables). Returns on capital are, so far, ordinary (ROIC ~7% as freshly-deployed assets lag), the moat is turbine-slot scarcity rather than a structural franchise, and pre-IPO sponsors have sold $650M+ into the rally. At ~$71.57 the stock trades at the 92nd percentile of its own valuation history (P/S 97th, P/B 95th) — roughly 22x a still-ramping adjusted-EBITDA run-rate. The business quality of the contracted stream is high; the quality of the incremental-capital return — and whether the balance sheet and share count can bear the build — is the open question the market is being asked to underwrite at a rich price.

2. Business Overview

What the company does. Solaris makes money two ways, in two reportable segments.

Solaris Power Solutions is the growth engine and now the majority of the business. Solaris buys natural-gas turbines from OEMs and provides turnkey, behind-the-meter (off-grid) electric power to large loads — overwhelmingly AI data centers, plus some oil-and-gas, utility, and commercial/industrial customers. It is an asset owner-operator and systems integrator, not a turbine manufacturer. The economic model has two layers (FY25 10-K revenue-recognition note; Q1’26 call):

• A “dry lease” — a fixed monthly rent for the turbine, recognized straight-line as an ASC 842 operating lease, that begins at contract commencement even before the turbine physically spins. This is why a contract “goes under rent at the beginning of the year” but only ramps physically as the data center energizes.
• A “wet lease” / fired-hour charge — a usage-based charge that layers on once units operate. Turbine engine cores are depreciated units-of-production over 30,000 fired hours, reflecting mechanical wear.

Around the turbine, Solaris increasingly sells balance-of-plant — transformers, switchgear, SCR emissions catalyst systems, e-houses, power distribution/conditioning, last-mile gas delivery, and O&M. Management calls this “molecule to electron” and quantifies a 20–50% uplift to capital deployed and EBITDA per site when it delivers the full scope. Capital cost is roughly $1.1M/MW ($1,100/kW), against a management-cited ~$3,500/kW for a new large-frame grid build — the cost/speed argument for behind-the-meter power.

Solaris Logistics Solutions is the legacy oilfield business — mobile sand-silo systems, AutoHopper AI-controlled sand delivery, top-fill equipment, Railtronix/Solaris software, and last-mile trucking/mobilization for well completions. Revenue is a mix of equipment leasing (93 “fully utilized systems” in FY25; 104 average in Q1’26) plus growing last-mile/ancillary services. Customers are E&P and oilfield-service firms under master service agreements.

Revenue segmentation (FY25 10-K):

Recurring vs. non-recurring. Power is contracted, long-duration (10–15-year terms), and take-or-pay-like (fixed dry-lease rent), with an advance rental prepayment (e.g., $45.4M booked as deferred revenue on the Hatchbo contract) — high-visibility. Logistics is recurring but cyclical, tied to US completions activity and oil price. The company directs essentially all incremental capital to Power (capex $639M vs $7M), a clear statement of where it is going.

Verdict: A coherent two-part machine — mature oilfield cash and operating culture subsidizing a contracted, high-margin power build — but the enterprise is now fundamentally a bet on behind-the-meter data-center power at a moment of extraordinary turbine scarcity.

3. Industry Dynamics

The demand driver is the most powerful secular story in the market, and it is real. AI compute is being built faster than the electric grid can connect it. Grid-interconnection queues have lengthened to multi-year waits; residential electricity-affordability politics make it hard to route incremental load through regulated utilities without public backlash. The binding constraint for a hyperscaler is speed-to-compute — a data center that cannot get power is a stranded multi-billion-dollar asset. Hyperscalers are therefore being forced (management’s word, repeatedly) into contracting for behind-the-meter, “bring-your-own-power,” island-mode generation that sidesteps the grid. This is not a preference cycle; it is a structural workaround to a physical bottleneck, and it is large — gigawatts per hyperscaler.

The supply chokepoint is the crux of both the bull and bear case. Large-frame gas-turbine supply is acutely constrained:

• GE Vernova’s gas backlog reached ~100 GW in Q1’26, with 2026–2027 delivery slots largely sold out.
• Siemens Energy reports an ~€136B order book with ~60% of gas-turbine orders data-center-tied.
• Mitsubishi, Siemens, and GE together are ~two-thirds of global gas-turbine capacity, each expanding output only 25–35%/yr from 2026; lead times have stretched to 5–8 years.

Solaris sits below the OEMs and beside the EPC contractors in the value chain: it does not make turbines; it secures scarce delivery slots, owns the assets, integrates the balance-of-plant, and operates the plants. Its economic edge today is that it holds slots and a live operating track record while the OEM queue is multi-year — a genuinely valuable position right now.

But apply the Marathon capital-cycle lens honestly. Sustained >20% unlevered returns on a buy-and-lease asset, in an industry where the OEMs have publicly committed to 25–35%/yr capacity expansion, is precisely the signal that draws capital. It already has: VoltaGrid signed Oracle for 2.3 GW; ProEnergy is deploying aeroderivative “bridge power”; PE-backed “powered-land” developers are proliferating; and management itself describes “speculators in the queue” who bought turbine slots hoping to become “a mini-SEI” — Solaris literally bought a distressed queue (the NovaLT16/Colusa slots) off one such party in March 2026. When turbine supply normalizes (2027–2029), the slot-scarcity rent compresses and the return on new capital mean-reverts toward the cost of capital.

Structural verdict: A large, fast-growing, genuinely attractive end-market — but a mid/late-capital-cycle one for a deployer of new capital. The next ~3–5 years favor an incumbent asset-owner with secured slots; the abnormal returns are a scarcity rent, not a permanent industry feature. Solaris’s task is to convert a timing advantage into a structural one (operating scale, switching costs, service embeddedness) before the cycle turns.

4. Competitive Position

The competitive set is real and crowding fast. In power: VoltaGrid (closest large-scale peer; Oracle 2.3 GW), ProEnergy (aeroderivative bridge power), Caterpillar/Solar and Cummins gensets, Aggreko and Generac (mobile/distributed generation), Bloom Energy (fuel cells), the hyperscalers’ own in-house power teams, utility bridge-power offerings, future SMRs, and the “speculators” holding turbine slots without operating capability. In logistics: Atlas Energy Solutions (which bought Hi-Crush in 2024 and dominates integrated proppant logistics with ~30% Permian share), US Silica, Covia, and others.

Name the moat honestly — there is no durable structural moat today. Solaris is not an OEM; it holds no unique resource, no network effect, and no proprietary generation technology (its patents sit in the logistics silo business). What it has is a bundle of first-mover, capital-deployment advantages, ranked by durability:

1. Secured turbine delivery slots — real and valuable, but a timing/capital advantage, not a barrier to entry. Erodes as OEM capacity expands.
2. Operating track record / uptime — the most defensible piece. Management repeatedly leans on demonstrated uptime to win contracts and negotiate SLAs (“you can point to actual operations”). This is a reputation/execution advantage — real, but replicable over time by well-capitalized entrants.
3. Balance-of-plant + O&M integration — the “molecule-to-electron” turnkey scope, deepened by the SCR-manufacturing investment, the power-distribution acquisitions, and the July-2026 GESA deal (adding Baseload Power + Pro-Per Energy O&M). This is the most promising durable differentiator: an in-house service capability for a large installed base of turbines that need overhaul every 30,000 fired hours raises switching costs and is hard to assemble quickly. It is early and unproven.
4. Contract captivity — 10–15-year contracts with three IG customers create genuine customer captivity for the life of each contract (a Greenwald demand-side advantage). But captivity that must be re-purchased with ~$1.1M/MW of fresh capital for every incremental megawatt is a series of good contracts, not a self-reinforcing moat.

Greenwald classification: contingent demand-side captivity (contract life) plus a nascent local-scale/service advantage; no barrier to entry on new capacity. If a moat claim cannot be tied to a financial outcome that deteriorates without it, it is not a moat — and here, if turbine supply normalizes and a VoltaGrid/ProEnergy/PE entrant matches uptime, Solaris’s >20% unlevered return on new capital compresses. The contracted base is protected; the growth returns are the capital-cycle question.

Verdict: A crowded and rapidly-crowding market with a real but time-limited first-mover/execution advantage — not yet a durable moat. The credible path to durability (O&M/service/switching-cost embeddedness) exists but is unproven. Rate it a narrow, contingent moat over contract life; no structural moat on new capital.

5. Growth History and Forward Opportunities

Historical growth is spectacular but almost entirely acquired and capital-driven, not organic in the traditional sense. Revenue nearly doubled in FY25 to $622M, but the driver is the MER acquisition (which created Power) plus the aggressive purchase and deployment of turbines. Power Solutions revenue went from $38.6M to $333.5M in a single year as the fleet scaled from ~230 MW average (FY24) to ~630 MW (FY25) to >900 MW operated (Q1’26). This is “growth by balance sheet” — every incremental megawatt of revenue requires ~$1.1M of capital, funded externally.

Forward opportunities are large and, unusually for a story stock, substantially contracted:

• >2 GW under 10–15-year contracts with three investment-grade tech companies. The three anchors: the Stateline JV (900 MW, 7-yr, 50.1%/49.9% JV), the Hatchbo Agreement (>500 MW, 10-yr + 5-yr option, ramping from Q1’27, with a $45.4M advance prepayment), and a third >600 MW contract (April 2026, 10-yr + 5-yr option, energizing late 2026).
• ~3.1 GW secured capacity after the March-2026 Genco and NovaLT16 slot acquisitions — leaving ~1 GW of secured-but-not-yet-contracted capacity to place.
• Scope expansion — management is in “advanced negotiations” to add balance-of-plant scope and incremental generation to already-signed contracts, plus adjacencies (distribution-only projects, consulting, a hyperscaler “mobile distributed compute” pilot). The stated 20–50% BoP uplift is being underwritten at the low end in guidance, leaving upside.
• Management’s pro-forma scenario: the full 3.1 GW deployed “could well exceed $1 billion” of annual adjusted EBITDA, with further upside from scope expansion.

Quality assessment. The contracted portion of the growth is genuinely high-quality: long-duration, high-margin, take-or-pay-like, IG counterparties. But three caveats keep it from being unambiguously high-quality: (1) it is externally funded — $1B+ more capital needed — so per-share value depends on financing terms and dilution; (2) it is concentration-laden — a handful of customers, one at 47–52%; and (3) it is being booked at the top of a supply-driven return cycle. Verdict: high-quality growth in accounting terms (contracted, high-margin), high-risk in economic terms (capital-intensive, concentrated, dilutive, cyclically-timed).

6. Financial Quality

Revenue and margins. The multi-year trajectory (ROIC/EDGAR): revenue $159M/$320M/$293M/$313M/$622M FY21–25, Q1’26 $196M. Gross margin climbed 10% → 22% → 27% → 26% → 32% (FY25), reaching 37% in Q1’26 as higher-margin power mix grew. EBITDA margin (GAAP) 17% → 23% → 30% → 30% → 35%, with Q1’26 at 37%. Adjusted EBITDA (segment sum) was $278M in FY25; Q1’26 adjusted EBITDA was $84M (+79% YoY), with Q2’26 guided to $83–93M and initial Q3’26 to $80–95M — a ~$350M+ and rising annualized run-rate. The margin structure genuinely improves with the power mix; this is not a low-quality top line.

The cash-flow reality is the heart of the skeptical case. Operating cash flow was healthy at $209M in FY25 — but capex was $647M, producing free cash flow of −$438M (−$129M in FY24; the company has been FCF-negative in four of the last five years as it builds). This is a capital-consuming enterprise by design: net PP&E went from $649M (YE24) to $2,240M (Q1’26); gross fixed assets from $852M to $2,530M. The build is financed by debt and equity issuance, not internal cash.

Balance sheet — leverage is climbing fast and the structure is complex. Net debt rose from $201M (YE24) to $1,252M (Q1’26); net-debt/equity ~113%. On top of the 3/31/26 debt stack — two convertible notes ($155M 4.75% due 2030 at a $26.39 conversion; $747.5M 0.25% due 2031 at $57.20, with $65.6M of capped calls), a $300M bridge term loan, the non-recourse Stateline project loan (~$260M drawn of a $518.5M facility), and Genco-related equipment debt — the company issued $1.3B of 6.375% senior notes due 2031 in May 2026, taking pro-forma gross debt to ~$2.4B+. Liquidity is adequate for now (cash $344M at Q1’26 plus undrawn facilities plus the note proceeds), but the model is explicitly dependent on continued access to capital markets.

Returns on capital are, so far, ordinary. ROIC (ROIC.ai) was ~7.4% in FY25 and 5.2% in FY24 — below any reasonable cost of capital for a levered, high-beta name — because a huge slug of capital was just deployed and has not yet earned. Reported ROE (135% in FY25) is a meaningless artifact of the Up-C structure’s tiny GAAP common book value (per-share book of $0.64 vs tangible book of ~$15.82); ignore it. The bull case is that stabilized ROIC on the contracted fleet reaches the mid-teens as dry-lease revenue ramps against deployed assets; that is unproven and is the number that matters most.

Quality-of-earnings flags. (1) GAAP net income is small and noisy — FY25 net income to common was just $30M (EPS $0.72 basic / $0.61 diluted) despite $135M of operating income, because of heavy NCI (~$28M to Class B) and large, volatile non-operating items ($62M of non-operating income in FY25 including ~$41M of “other,” and a $34.7M other-non-operating item in Q4’25) that appear to be fair-value/remeasurement in nature — GAAP EPS is not a clean earnings gauge here; use adjusted EBITDA and, more importantly, cash returns on the fleet. (2) SBC was $23.4M in FY25 (up from $10.6M), a real and growing dilution cost. (3) The dry-lease convention front-loads revenue recognition relative to cash energization, so reported power revenue can lead physical utilization.

Verdict: Economics do improve with the power mix (gross and EBITDA margins are structurally higher), but the business is deeply cash-consumptive, increasingly levered, and has not yet demonstrated an above-cost-of-capital return on the capital it is deploying. The quality of the contracted revenue stream is high; the quality of the balance sheet and the returns on incremental capital is the concern.

7. Capital Allocation

The strategy is single-minded: raise external capital and deploy it into turbines at a targeted >20% unlevered return. Judged on its own terms it has created enormous equity value — but the capital-allocation quality is genuinely mixed, and several signals warrant skepticism.

M&A ledger (all disciplined in rationale, aggressive in pace and dilution):

• MER (Sept 2024, $323.1M): $186.4M seller rollover into 16.46M Class B units (at $11.32) + $136.7M cash. Created the Power segment; no earn-out. In hindsight a franchise-making deal.
• HVMVLV (Aug 2025, $59.7M): power control/distribution (SCR/switchgear); 2% of FY25 revenue.
• Genco Power Solutions (Mar 2026, $484.3M): asset acquisition (no goodwill), including 4.18M Class A shares + cash + assumed equipment debt; +400 MW.
• NovaLT16 slots (Mar 2026, $66.9M + $64.3M to clear Colusa’s overdue Baker Hughes invoices): effectively a distressed-queue takeover; +~500 MW, and a supplier-diversification move (toward Baker Hughes/NovaLT away from Solar/Cat).
• GESA (Jul 2026, ~$55M cash/debt + 2.88M Class A shares): O&M capability.

Funding — the aggressive, dilutive part. The build has been financed with ~$900M of converts, a $1.3B senior-note issue, project debt, and repeated equity: eight 424B5 secondary offerings and stock consideration in most acquisitions. Fully-diluted economic units have gone from ~29M (2023) to ~76M (mid-2026) before convert dilution (~90–95M loaded). The share count is a moving target that grows with every deal and financing.

Two red flags.

1. The dividend contradiction. Solaris maintained its $0.12/quarter ($0.48/year, ~$34M including LLC distributions) dividend — reaffirmed February 2026 — while burning $438M of FCF and preparing to raise $1B+. Paying cash out to shareholders while simultaneously diluting and levering to fund capex is capital-allocation theater; the dividend is a legacy-oilfield artifact that should arguably be redirected to the build.
2. The incentive structure ignores capital efficiency. Executive compensation uses relative-TSR PSUs and RSAs with no ROIC or return-on-capital metric at all. In a business whose entire thesis rests on earning >20% on deployed capital, rewarding management purely on stock price and TSR — Zartler’s pay rose from $3.82M (FY24) to $13.87M (FY25) as the stock quadrupled — incentivizes growth and multiple, not return discipline. This is the classic Marathon warning sign: management paid to deploy capital, not to deploy it well.

Insider behavior — a revealing split. Management and one director made small but genuine open-market buys (director A.J. Teague bought ~$500k at ~$73; Zartler made five buys; Ramachandran bought at $25). But the pre-IPO sponsors distributed $650M+ into the rally: J Turbines (4.0M shares at $30.30), KTR (4.0M at $29.50 + 2.0M at $70.75), and Yorktown-affiliated director W. Howard Keenan (2.0M at $74.50). The people who financed the pivot are handing the paper to the public near the top; the operators are nibbling. Zartler — a serial build-and-monetize operator who founded and sold Aris Water to Western Midstream in October 2025 — owns ~7.6%.

Governance notes. One-vote-per-share for both classes (no super-voting); Yorktown/Zartler/KTR control ~20% of votes combined; a Tax Receivable Agreement pays 85% of cash tax savings to pre-IPO holders (~$75M liability, accelerable on change-of-control); related-party leases with Zartler entities are minor. A CFO change (Ramachandran → Tompsett) landed at the exact moment of the largest financing program in company history.

Verdict: Management has allocated capital in a way that created enormous value and made franchise-defining acquisitions — but on quality, the picture is mixed to cautionary: aggressive dilution, a contradictory dividend, a TRA that leaks value to insiders, an incentive plan with no capital-efficiency metric, and sponsors exiting into the rally. Rate it “bold and so-far-successful, but structured to favor growth and insiders over per-share return discipline.”

8. Changes and Headwinds — Last Two Years

Strategic changes (transformational). The MER acquisition (2024) created the Power segment; the company renamed itself (Sept 2024); it added Co-CEO Amanda Brock (ex-Aris Water CEO) in October 2025 alongside founder Zartler; and it executed a rapid-fire series of power contracts (Stateline, Hatchbo, the third >600 MW deal) and capacity acquisitions (Genco, NovaLT16 slots, HVMVLV, GESA). It transitioned from a single-segment oilfield-logistics company to a two-segment, power-led infrastructure platform in under two years.

Financing changes. Two convertible-note issues (2025), a $300M bridge (Mar 2026), a $1.3B senior-note issue (May 2026), the non-recourse Stateline facility, and eight equity secondaries — a wholesale rebuild of the capital structure from ~$300M net debt to ~$2.4B+ gross.

Leadership/board. Co-CEO structure installed; CFO handoff (Ramachandran to President; Tompsett to CFO); two shareholder-derivative suits (Oct/Dec 2025) filed and voluntarily dismissed; a director drew ~21% withheld votes at the 2026 annual meeting.

Headwinds and watch-items. (1) The Q3’26 guide was deliberately “measured,” with management cautioning that “the market may have gotten a little exuberant” about the pace of rollout — an unusually candid check on the momentum. (2) Turbine delivery timing swings quarterly results and is not fully in the company’s control. (3) The capital-markets dependence is the dominant headwind: any tightening in credit or equity markets directly constrains the growth plan. (4) Rising interest cost — the step from 0.25%/4.75% converts to 6.375% senior notes signals the rising marginal cost of the build.

Verdict: The changes have strengthened the business’s strategic position and earnings trajectory but weakened its balance-sheet resilience and raised its dependence on external capital and a few customers. Net: a stronger franchise carried on a more fragile financial structure.

9. Risk Analysis

Catastrophic-loss scenario: a combination of a closed capital-markets window, a deferral by the 47%-concentration customer, and a rising refinancing cost as the capital cycle turns — hitting the growth, the leverage, and the multiple simultaneously. Probability is low but non-trivial given the leverage and concentration; the equity is levered to it.

10. Valuation Discussion (Embedded Expectations)

No price target, no recommendation — this section frames what the market is underwriting.

Start with the correct denominator, because the screen lies. The naïve market cap (Class A ~58M shares × $71.57 ≈ $4.2B) materially understates the true economic value. Including Class B units and the GESA shares, economic units are ~76.4M; loading in-the-money converts (~5.9M at $26.39 + ~13.1M at $57.20) pushes the fully-diluted count to ~90–95M. True economic equity value is therefore ~$5.5–6.5B, not $4.2B. Adding net debt (~$1.25B at Q1’26, higher pro-forma after the $1.3B notes), the fully-loaded enterprise value is roughly $7–8B.

Multiples (fully-loaded):

• EV/TTM adjusted EBITDA (~$336M annualizing Q1’26): ~21–24x.
• EV/2027E adjusted EBITDA (a plausible ~$450–550M as contracts ramp): ~13–17x.
• EV / management’s “$1B+” pro-forma (full 3.1 GW, ~2028+): ~7–8x — but that number requires another $1B+ of dilutive capital and flawless execution to reach.
• Own-history percentile (AZI): composite 92.5th, P/S 97th, P/B 95th — the richest valuation in the company’s history.

What must be true to justify ~$71.57. At a ~$7–8B fully-loaded EV, the market is capitalizing a number well above the current ~$336M run-rate — roughly midway between today’s run-rate and the pro-forma $1B. Embedded is: (1) the full 3.1 GW gets contracted and deployed on schedule with IG counterparties honoring 10–15-year terms; (2) incremental capital is raised without catastrophic dilution or a punitive cost of capital; (3) >20% unlevered returns hold on new capital despite the capital cycle; and (4) the scope-expansion (BoP) upside materializes. That is a demanding, largely-priced set of assumptions.

Scenario frame (illustrative, fully-diluted ~90M units):

• Bear: capital cycle turns / a customer defers / financing tightens. Stabilized adjusted EBITDA plateaus near ~$350–400M; the market applies a ~9–10x levered-power multiple → EV ~$3.5–4B → equity ~$2.3–2.8B → ~$25–32/share.
• Base: contracts ramp to ~$500M adjusted EBITDA by 2027–28 at an ~11–12x multiple → EV ~$5.5–6B → equity ~$4.2–4.8B → ~$47–55/share.
• Bull: full 3.1 GW to ~$1B adjusted EBITDA with scope upside, financed accretively, at ~9–10x → EV ~$9–10B → equity ~$7.5–8.5B → ~$85–95/share — roughly today’s price plus the June high, i.e., the bull case is close to fully priced.

Comparables framing. Same-theme peers Powell (POWL) and Argan (AGX) trade at their own richest-ever multiples (~35x and ~59x EV/EBITDA) — but both are net-cash, asset-light businesses. Solaris is the levered, asset-heavy version: it should trade at a lower EV/EBITDA multiple than an asset-light equipment/EPC peer (it carries the capital risk on its own balance sheet), which is why the ~21–24x fully-loaded multiple is arguably more stretched than the headline suggests. Levered contracted-power asset owners (IPPs, midstream) typically clear at 8–12x EV/EBITDA on stabilized cash flows.

Verdict: The market is underwriting a near-flawless execution of the full pipeline, financed without significant per-share damage, at returns that hold through a turning capital cycle. The contracted base plausibly supports a valuation in the mid-$40s–low-$50s; the current price additionally capitalizes the un-contracted ~1 GW and the scope-expansion optionality at close to full value.

11. Variant Perception

Consensus view. Sell-side is euphoric — Needham initiated Buy ($97), Wolfe Outperform, and the “earnings will grow several-fold” narrative dominates. Consensus treats SEI as a scarce, contracted, secular-growth pure-play on AI power, with the 10–15-year IG contracts de-risking the story and the $1B pro-forma EBITDA as a credible target.

The factor tape says something subtler. FactorsToday still classifies SEI among gas-compression and oilfield-services names (Archrock, Kodiak Gas, Natural Gas Services, Select Energy) — not among data-center-power peers. The quantitative regime has not yet re-coded the stock as “AI infrastructure.” Loadings are strong momentum (~0.9–1.0) and oil-price-sensitive, deeply anti-value (Value −0.6 to −1.3), negative quality, and high market beta (1.97). Risk-adjusted, it screens as a high-return, high-volatility momentum vehicle: +135% annualized over the past year, but a −55% max drawdown in its history — the amplitude cuts both ways. This is a crowded momentum trade riding a real fundamental re-rating, positioned where a growth stumble would force both a factor-rotation and a multiple compression.

Strongest bull case. The demand is the most durable secular force in markets; the contracts are signed with creditworthy hyperscalers on decade-plus terms; the turbine-scarcity window is multi-year; Solaris has a genuine operating track record and is vertically integrating the balance-of-plant and O&M to deepen switching costs; and management’s $1B EBITDA scenario, if hit, makes today’s ~$7–8B EV look cheap at ~7–8x. Index inclusion and forced buying add a technical bid.

Strongest bear case. It is a levered, serially-dilutive, hyper-concentrated capital consumer earning ordinary returns (ROIC ~7%) so far, at the 92nd percentile of its own valuation, whose entire edge is a temporary scarcity rent that the capital cycle is actively competing away, whose incentive plan rewards growth not return, whose insiders are selling $650M into the rally, and which must raise $1B+ more — any hiccup in customers, financing, or the cycle re-rates a high-beta name violently.

The 3–5 assumptions that matter most: (1) Will the 47%-concentration customer(s) honor and expand? (2) Can the $1B+ of remaining capital be raised without crushing per-share value? (3) Do >20% unlevered returns survive the OEM capacity expansion? (4) Does the O&M/BoP layer actually create durable switching costs? (5) Does stabilized ROIC reach the mid-teens the model needs?

What would falsify each side. Bull falsified by: a customer deferral, a financing at punitive terms, or a gross-margin roll-down signaling the scarcity rent is competing away. Bear falsified by: two or three additional 10-year IG contracts self-funded from internal cash + the Stateline distribution (no fresh equity), with maintained 20%+ unlevered returns on the newest capital — proof the flywheel turns without dilution.

12. Fact vs. Interpretation

13. Open Questions

1. Who is the 47% customer, and what are the termination/renegotiation provisions? The single largest thesis risk is un-named.
2. What is the exact take-or-pay floor on the dry-lease contracts if a data center is delayed or a customer’s AI capex plan changes?
3. How will the remaining $1B+ be raised — more converts, senior notes at 6.375%+, project debt, or equity — and at what dilution/cost?
4. What stabilized unlevered return are the newest contracts (Hatchbo, the >600 MW deal) actually underwritten at, net of the rising OEM turbine cost management flagged?
5. How much of the $1B pro-forma EBITDA is contracted vs. speculative on the un-placed ~1 GW?
6. Will the dividend be cut/redirected to fund the build, or maintained as an income signal while diluting?
7. What is the real all-in cost of the balance-of-plant scope and does the 20–50% uplift survive competition?
8. How durable is the Solar/Caterpillar supplier relationship as SEI diversifies to competing OEMs?

14. What Must Be True

Bull case — what must be true:

• The full ~3.1 GW gets contracted and deployed on schedule; IG counterparties honor and expand 10–15-year terms.
• The remaining $1B+ is raised accretively — internal cash + Stateline distributions + reasonably-priced debt — without a dilutive equity flood.

• 20% unlevered returns survive the OEM capacity expansion; stabilized fleet ROIC reaches the mid-teens.

• The O&M/BoP integration creates durable switching costs, converting a timing edge into a structural one.
• Falsification test: if, over the next 12–18 months, Solaris raises the bulk of its remaining capital via equity/high-cost debt AND/OR gross margin on new contracts rolls down toward the mid-20s, the accretive-self-funding bull thesis is broken.

Bear case — what must be true:

• Turbine supply normalizes and competitor entry (VoltaGrid/ProEnergy/PE) competes the scarcity rent away; new-capital returns fall toward the cost of capital.
• Capital-markets dependence bites — a financing at punitive terms or a customer deferral stalls the ramp.
• The multiple compresses from the 92nd percentile as growth decelerates; the high-beta equity re-rates violently.
• Falsification test: if Solaris signs two or three additional 10-year IG contracts self-funded from internal cash and the Stateline distribution (no material new equity), with maintained 20%+ unlevered returns on the newest capital, the “capital-cycle-peak / can’t-self-fund” bear thesis is broken.

15. Source Appendix

See the separate Source Appendix (Appendix B in the combined report) for the full citation list. Primary sources: SEI FY2025 10-K (filed 2026-02-27) and prior 10-Ks (FY21–24); 10-Qs through Q1 2026; DEFM14A (Aug 2024); DEF 14A (2026); Form 4 / 424B5 corpus; SEI Q1 2026 earnings call transcript (2026-04-28, via ROIC.ai); ROIC.ai financial data; market news feeds and own-history valuation percentiles; FactorsToday factor model; daily price history; company press releases (MER, HVMVLV, Genco, NovaLT16, GESA); GE Vernova and Siemens Energy filings for turbine-market context; same-theme peer analyses (Powell Industries / POWL; Argan / AGX).

APPENDIX A — Standard Diligence Questionnaire

Solaris Energy Infrastructure, Inc. (NYSE: SEI) — as of 2026-07-11

Supplemental diligence checklist. Answers grounded in the underlying analysis; Fact / Interpretation / Assumption labeled where it matters.

General

What thoughtful questions have other investors asked about this company? The central debates: (1) Are the >20% unlevered turbine returns durable, or a temporary scarcity rent that the capital cycle competes away? (2) Can the company fund $1B+ of remaining capex without crushing per-share value? (3) How real and how sticky is the “molecule-to-electron” balance-of-plant moat? (4) What is the true fully-diluted share count and per-share value given the Up-C + converts? (5) Is the 47%-customer concentration a de-risking feature (contracted IG cash flow) or a single point of failure? (6) Is this a secular compounder or a founder’s build-and-monetize vehicle (cf. Zartler’s Aris Water sale)?

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? Interpretation: Margins and multiples are at a high; returns on capital (ROIC ~7%) are depressed by freshly-deployed, not-yet-earning assets. The Logistics segment’s EBITDA is off its peak (declining $115M→$98M→$89M). Power is early-ramp, not peak. So: valuation high, contracted-EBITDA early, oilfield-cash mid-cycle.

Driven by external environment or internal action? Both. External: the AI-driven behind-the-meter power boom and turbine scarcity. Internal: aggressive M&A and capital deployment. The revenue doubling is internally-engineered (buying and deploying turbines) atop an external demand wave.

How stable are revenues? Power is contracted and long-duration (10–15-yr, take-or-pay-like dry leases) — high visibility once signed, though ramp timing is lumpy. Logistics is cyclical to US completions/oil. Consolidated stability is improving as Power grows.

Outlook / market size. Very large and growing — behind-the-meter data-center power is gigawatt-scale per hyperscaler, constrained by a 5–8-year turbine backlog. Domestic (US) focus.

Business Quality & Competitive Moat

Is the industry getting more or less competitive? More. VoltaGrid (Oracle 2.3 GW), ProEnergy, PE-backed developers, and “speculators in the turbine queue” are entering; OEMs are expanding capacity 25–35%/yr. Classic Marathon capital cycle.

How profitable is the business (ROIC, ROE)? ROIC ~7.4% (FY25) — below cost of capital, distorted low by just-deployed assets; the model requires stabilized mid-teens ROIC, unproven. Reported ROE (135%) is a meaningless Up-C artifact (tiny GAAP book). Segment EBITDA margins are strong (Power 57%, Logistics 31%).

How profitable is the industry / barriers to entry? High current returns (>20% unlevered) but low structural barriers on new capacity — the edge is turbine-slot scarcity and execution, not a franchise. Barriers erode as OEM supply normalizes.

Can the business be easily understood? Moderately. The power model (own turbines, lease them turnkey) is simple; the Up-C structure, TRA, convert stack, VIE (Stateline), and dry/wet-lease accounting add real complexity.

Undermined by foreign low-cost labor? No — US-sited, physical infrastructure with local O&M.

Do brands matter? No consumer brand; reputation/track-record (“uptime”) is the relevant intangible in winning hyperscaler contracts.

Nature of competition / switching costs. Competition on speed-to-power, secured slots, operating reliability, and turnkey scope. Switching costs are real within a 10–15-year contract (customer captivity) but must be re-bought with capital for each new megawatt.

Financial Condition & Balance Sheet

Assets not fully recognized? Secured turbine delivery slots and the contracted backlog have value beyond book. Conversely, the un-placed ~1 GW of secured capacity is capital at risk if not contracted.

Off-balance-sheet liabilities? The TRA (~$75M, pays 85% of tax savings to insiders, accelerable on change-of-control). Operating-lease and purchase commitments for turbines. Stateline is consolidated (VIE) with non-recourse project debt (~$260M drawn).

How conservative is the accounting? Interpretation: Aggressive in one respect — the dry-lease convention recognizes power revenue at contract commencement before physical energization. GAAP net income is small and noisy (heavy NCI + volatile fair-value/non-operating items). Use adjusted EBITDA and cash returns, not GAAP EPS.

CapEx-hungry? Extremely — the defining feature. FY25 capex $647M vs $209M operating cash flow; ~$1.1M/MW; $1B+ more planned. This is a capital-consuming business by design.

Capital Allocation & Management

FCF generation / use / philosophy? FCF is deeply negative (−$438M FY25) by design; the philosophy is “raise external capital, deploy into >20%-return turbines.” All incremental capital goes to Power.

Significant acquisitions? Yes, serially: MER ($323M, 2024), HVMVLV ($60M), Genco ($484M), NovaLT16 slots ($67M+), GESA ($55M) — mostly funded with stock + debt.

Buying back / issuing shares? Issuing heavily — eight secondaries, stock M&A consideration, converts. Economic units ~29M (2023) → ~76M (2026), fully-diluted ~90–95M. No buybacks.

Insiders issuing to themselves / selling? Sponsors (Yorktown, J Turbines, KTR) sold $650M+ into the rally via secondaries; management made small open-market buys. Zartler owns ~7.6%.

Compensation policy / motivations. Relative-TSR PSUs + RSAs; no ROIC metric. Zartler FY25 pay $13.87M (up from $3.82M) tracking the stock. Motivation is growth/TSR, not capital efficiency — a governance concern. Founder has a build-and-monetize track record (Aris Water → Western Midstream, 2025).

Valuation & Market Data

ADR / MLP / K-1? No. It is an Up-C (Class A common + Class B units in Solaris LLC) with a TRA — so Class A holders own shares (Form 1099, not K-1), but the structure passes economics/tax attributes to pre-IPO Class B holders. Not an MLP.

Dividend policy? $0.12/quarter ($0.48/year, ~0.67% yield), maintained despite negative FCF — a legacy-oilfield artifact and, arguably, a capital-allocation contradiction while raising $1B+.

How profitable / is net income diverging from cash flow? GAAP net income ($30M FY25) is far below operating cash flow ($209M) due to D&A on the large fleet, NCI, and non-operating items — a normal divergence for a capital-heavy asset owner. Adjusted EBITDA ($278M) is the better gauge.

Risks & Downside

What would cause the stock to decline? A customer deferral/renegotiation (47% concentration), a financing at punitive terms, a gross-margin roll-down signaling scarcity-rent erosion, a growth-deceleration multiple compression (92nd-percentile valuation, beta 1.97), or a broad risk-off move in high-beta AI-infrastructure names.

Catastrophic-loss risk? Non-trivial but low-probability: a simultaneous closed capital-markets window + major-customer default + refinancing spike. The leverage and concentration make the equity levered to a bad-case convergence.

Chance of total loss? Low near-term — contracted IG cash flows, real assets, and adequate current liquidity. But it is a levered, capital-dependent growth story; a severe cycle turn could impair the equity meaningfully (the stock has a −55% historical max drawdown).

Recent News & Events

Has the environment changed recently? Yes, continuously: two new hyperscaler contracts (>1 GW, Feb + Apr 2026), Genco + NovaLT16 slot acquisitions (Mar 2026), $1.3B senior notes (May 2026), GESA acquisition (Jul 6, 2026), S&P SmallCap 600 inclusion (Jul 9, 2026), and bullish sell-side initiations (Needham $97, Wolfe Outperform). Momentum is intense; the tape is euphoric.

Accounting-policy changes? None material flagged beyond the evolving power-revenue conventions (dry/wet lease, turbine-core units-of-production depreciation).

New markets / facilities / management? New end-market (data-center power) is the whole story; Co-CEO Amanda Brock added (Oct 2025); CFO change (Ramachandran → Tompsett, Feb 2026); rapid geographic/fleet expansion.

APPENDIX B — Source Appendix

Solaris Energy Infrastructure, Inc. (NYSE: SEI) — as of 2026-07-11

Primary sources prioritized. All financial figures reconciled to SEC filings where possible; third-party aggregated data (ROIC.ai, AZI, FactorsToday) cross-checked against filings.

Primary — SEC filings (CIK 0001697500)

• FY2025 Form 10-K (filed 2026-02-27), sei-20251231.htm — segment revenue/EBITDA, customer & supplier concentration notes, convertible-note terms, Stateline VIE (Note 3), Up-C/TRA disclosures, dividend policy, share counts (53.1M Class A / 15.35M Class B as of 2026-02-19). https://www.sec.gov/Archives/edgar/data/1697500/000162828026012501/sei-20251231.htm
• FY2021–FY2024 Form 10-Ks (soi-2021…, soi-2022…, soi-2023…, sei-2024…) — historical (pre-pivot) financials and segment history.
• Form 10-Q Q1 2026 (period 2026-03-31, filed 2026-05-01) — Q1 financials, updated debt stack, top-customer 52%, supplier 74% of AP, Hatchbo advance prepayment.
• Form 10-Qs FY2021–FY2025 — quarterly trajectory.
• DEFM14A (special meeting 2024-08-30) — MER acquisition: issuance of 16,464,778 Class B shares, rename to Solaris Energy Infrastructure, LTIP increase.
• DEF 14A (2026 annual meeting) — executive compensation (relative-TSR PSUs + RSAs, no ROIC metric), board, related-party (Solaris Energy Management).
• Form 4 / 424B5 corpus (129 Form 4s; 8 secondary offerings) — insider activity: management open-market buys (Teague ~$73, Zartler, Ramachandran) vs sponsor distributions (J Turbines 4.0M @ $30.30; KTR 4.0M @ $29.50 + 2.0M @ $70.75; Keenan/Yorktown 2.0M @ $74.50).
• 8-K filings — Genco & NovaLT16 slot acquisitions (2026-03-16), $1.3B 6.375% senior notes due 2031 (2026-05-12), GESA acquisition (2026-07-06), quarterly earnings releases.

Primary — Company / IR

• SEI Q1 2026 earnings call transcript (2026-04-28), via ROIC.ai — 3.1 GW secured, >2 GW under 10–15-yr contracts with 3 IG customers, “$1B+” pro-forma adjusted EBITDA scenario, >20% unlevered return target, ~$1.1M/MW vs ~$3.5M/MW grid, $1B+ additional capital 2026–27, Q2/Q3 guidance, dry/wet-lease conventions.
• Company press releases — Mobile Energy Rentals acquisition (2024-07-09, businesswire); HVMVLV (2025-08); Genco/NovaLT16 (2026-03-16); Hatchbo Agreement (2026-02-12); third >600 MW contract (2026-04-24); GESA (2026-07-06, businesswire). https://www.businesswire.com/news/home/20240709800442/en/ ; https://www.businesswire.com/news/home/20260706403615/en/
• Company website / IR — Solaris Energy Infrastructure. https://www.solaris-energy.com

Quantitative data providers (cross-checked to filings)

• ROIC.ai MCP — income statement, balance sheet, cash flow, profitability ratios, per-share, enterprise value, valuation multiples, company profile (annual FY20–25 + quarterly through Q1’26). Third-party aggregated; reconciled to 10-K/10-Q.
• Market news & valuation data — own-history valuation percentiles (composite 92.5th; P/E 84.8th, P/B 95.4th, P/S 97.2nd as of 2026-07-10); news feed (Needham Buy $97 6/29; Wolfe Outperform 7/6; S&P SmallCap 600 add 7/9; GESA deal 7/6); daily price CSV (52-wk ~$20–86; $71.57 on 7/10; beta ~1.97).
• FactorsToday — factor loadings (Momentum ~0.9–1.0, Value −0.6 to −1.3, DividendYield high, OilPrice positive, market beta 1.43–1.97; R² ~0.31–0.39); leaderboard (y1 return +135% ann., y1 Sharpe 1.80, max drawdown −55%); stock-info (RS-12m +133%); related stocks (Archrock, Kodiak Gas, Natural Gas Services, Select Energy — gas-compression/OFS cluster).

Industry / third-party context

• GE Vernova 1Q26 8-K — ~100 GW gas backlog, 2026–27 slots sold out. https://www.sec.gov/Archives/edgar/data/0001996810/000199681026000063/gevpressrelease1q26.htm
• Siemens Energy — ~€136B order book, ~60% of gas-turbine orders data-center-tied (public reporting).
• IEEFA (Oct 2025) and trade press — turbine-shortage, 5–8-yr lead times, OEM capacity expansion 25–35%/yr.
• Competitor context — VoltaGrid–Oracle 2.3 GW; ProEnergy aeroderivative bridge power; Atlas Energy Solutions / Hi-Crush (proppant logistics) — public reporting.

Peer context (cross-read)

• Powell Industries (POWL) and Argan (AGX) — same data-center gas-power theme at richest-ever multiples; used for peer valuation framing (both net-cash, asset-light — the structural contrast with levered, asset-heavy SEI). Public filings and market data.

Analytical frameworks

• investment-research-frameworks skill — Greenwald & Kahn Competition Demystified (moat-type taxonomy; barriers to entry; ROIC/share-stability tests) and Marathon Capital Returns (supply-side capital-cycle; high returns attract capital and mean-revert). Applied throughout the Competitive Position, Industry, and Capital Allocation sections.