S&P Global is a five-segment data and ratings oligopolist — Ratings, Indices, Market Intelligence, Commodity Insights, and Mobility — that converts trust and regulation into 127% ROIC. At $426 (px/IV = 0.955) the price already embodies most of the franchise; only the wide-IV scenario or a credit-cycle dislocation creates a genuine bargain.

Plain English

S&P Global is a referee in financial markets. It rates bonds (so investors trust them), runs the S&P 500 (so ETFs license it), sells research databases to bankers, prices oil and gas trades, and runs CARFAX. Most customers can't easily switch — regulators or contracts demand S&P specifically. The business takes tiny fees on enormous volumes of trades, issuances, and AUM. Returns on capital are extraordinary because the work is mostly software and brand, not factories. The risk: a credit-cycle blowup or Congress could change the rules. Today the price is fair, not cheap.

Thesis

S&P Global owns five franchises that the world's capital markets cannot easily route around: S&P Global Ratings (one of two NRSROs the SEC effectively designates for bond-issuance plumbing), the S&P Dow Jones Indices business (S&P 500, Dow Jones Industrial Average — the most-licensed equity benchmarks on earth), Market Intelligence (Capital IQ, Compustat, RatingsXpress), Commodity Insights (Platts price assessments — the embedded benchmark in physical oil/LNG/petchem contracts), and Mobility (CARFAX, automotiveMastermind). Each is a referee, not a player. Each charges fractional pennies on transactions or AUM that sum to enormous, recurring dollars.

The scorecard validates the franchise economics: ROIC 10y avg = 127.4%, owner earnings TTM = $4.84B, net debt / EBITDA = 1.66x (manageable post-IHS Markit), share count change 10y = +1.21% (essentially flat — buybacks have neutralized stock-comp dilution and one acquisition issuance). The composite score of 59 is held back not by quality but by price: valuation = 14/25, P/E TTM 33.18 vs 10y avg 31.87, and reverse-DCF implied growth of 8.04% — meaning the buyer at $426 is implicitly underwriting another decade of high-single-digit cash-earnings growth.

IV math: low $251 / base $446 / high $672. At $426 the stock trades 4.5% below base IV and 36% below high IV — fair, not cheap. The Buffett move is to wait. Target buy at $335 (25% margin of safety to base IV) opens a meaningful position; $446+ is a trim zone unless the bull case (AI-driven Market Intelligence revenue, Indices ETF AUM compounding) is being demonstrated in the numbers.

Moat

S&P Global is a textbook five-moat case study; the question is not whether the moats exist but whether they are widening or narrowing.

1. Intangibles — regulatory & brand (WIDE). S&P Global Ratings is one of two large Nationally Recognized Statistical Rating Organizations (the other is Moody's; Fitch a distant third). The U.S. SEC's NRSRO designation, plus capital-rule references in Basel and U.S. bank regulation that explicitly cite ratings from designated agencies, mean that most institutional bond issuance is functionally required to carry an S&P or Moody's rating to clear. Damodaran's discussion of legal-protection moats applies almost word-for-word: "Firms may enjoy exclusive rights to produce and market a product because they own the patent rights on the product... or they may have exclusive licensing rights to service a market" [1]. NRSRO status is exactly that — a regulator-granted license to print toll revenue on every new bond issued in the United States. The S&P 500 brand is an even better intangible: the index itself is a trademark. ETF sponsors (SPY, IVV, VOO) pay licensing fees per dollar of AUM tracking it. The brand is so embedded that "the S&P" is a synonym for "the U.S. stock market." Buffett describes exactly this kind of franchise when he writes that Coca-Cola and Gillette "set up a protective moat around their economic castles" [4]; S&P 500 licensing has the same character — a name nobody can replicate at scale because incumbency itself is the asset.

2. Switching costs (WIDE). Capital IQ, Compustat, and RatingsXpress are wired into the workflows of bulge-bracket banks, asset managers, and corporate treasury desks. Migrating a 200-analyst investment-banking floor off Capital IQ to a competitor is a multi-quarter, multi-million-dollar project that nobody does unilaterally. CARFAX is similarly embedded in U.S. used-car dealer software and consumer brand recall. Platts price assessments are written into thousands of bilateral physical-commodity contracts ("Platts Dated Brent" appears as the pricing reference); changing the reference price requires re-papering counterparties globally.

3. Network effects (WIDE on Indices, NARROW on Market Intelligence). The S&P 500's network effect is the cleanest one in finance: every additional dollar of ETF AUM tracking the index increases the index's liquidity and visibility, which makes it the default for the next allocator. ~$15+ trillion is benchmarked to S&P DJI indices. A challenger would have to pry institutional mandates loose one by one against this gravity. Market Intelligence has weaker network effects (data network + workflow), more Switching-cost-driven.

4. Cost advantages — scale economies on data infrastructure (NARROW). SPGI's marginal cost of selling one more Market Intelligence seat or one more index license approaches zero, while a sub-scale competitor must build the same data plumbing. This is real but it is a feature shared with every B2B data company; not unique.

5. Pricing power (WIDE on Ratings & Indices). Ratings revenue per issuance has compounded above CPI for 20+ years; index licensing fees on AUM are typically 2-4 bps and rise as new product structures (sector ETFs, factor ETFs) proliferate.

Competitor stress test ($10B + 5 years). Could $10B and 5 years build a credible #3 ratings agency? No — NRSRO status takes years and the issuer-pays incumbent customer base would not migrate without a regulatory mandate. Could it dethrone the S&P 500? No — incumbency is the moat. Could it carve out a credible Capital IQ rival? Bloomberg already tries (BQuant) and FactSet competes; SPGI's MI seat count keeps growing — switching costs win. Could it beat Platts? No — bilateral contract references are the lock-in. The franchise survives the stress test on four of five legs.

Erosion risk. Real but slow. Indices: a long-tail risk that the SEC, ESMA, or a class of large asset owners (sovereign-wealth pools) push for unlicensed or open-source benchmarks; this would be a multi-decade fight. Ratings: a 2008-style Dodd-Frank-2.0 that strips NRSRO references from regulation would be the structural killer. Market Intelligence: AI-driven competitors (a well-capitalized LLM-native research stack from a Bloomberg or new entrant) is the most credible 5-year erosion threat.

Moat verdict: WIDE.

Management

The core capital-allocation question for SPGI under Doug Peterson and successor Martina Cheung is whether the IHS Markit playbook (closed November 2022, ~$44B all-stock deal) is the last big swing or the template for more.

1. Reinvestment. Organic capex at SPGI is modest because the business is software/data; reinvestment shows up as content investment, technology platforms, and bolt-on M&A. The 127.4% 10-year ROIC says incremental capital deployed inside the franchise has earned spectacular returns. ROIIC of 6.51% over 5 years is the warning flag — incremental returns are decelerating, which is what one would expect after the IHS Markit goodwill landed on the balance sheet and the easy organic cycle turned. This is the single most important number on the management report card.

2. Acquisitions. IHS Markit was a defensible deal: Commodity Insights (Platts + IHS energy data) became a more valuable franchise, and Mobility (CARFAX-led) is genuinely a separable, attractive asset. The 2026 announced sale of Energy's geoscience and petroleum-engineering software portfolio to SLB suggests management is willing to prune — that's a positive signal about discipline. The risk is the next deal: when a high-multiple data company spends stock on a high-multiple data company, the synergies have to be real or the buyer's per-share economics are diluted. So far, the 1.21% 10-year share-count change says management has been unusually disciplined about not overpaying.

3. Debt. Net debt / EBITDA at 1.66x is conservative for a recurring-revenue business with 60%+ EBITDA margins. The company has investment-grade credit (which it must, as a ratings agency — there is reflexive risk if SPGI's own credit deteriorated, and management knows it). Interest coverage is very high; the scorecard reports null but the 1.66x leverage and the segment economics make this a non-issue.

4. Buybacks. This is where SPGI shines. The 10-K filings reference the 2022 and 2025 repurchase programs and four uncapped accelerated share-repurchase programs initiated in 2025 (February, May, August, December). The 1.21% net share count growth over 10 years — through a $44B all-stock acquisition that significantly increased the float — means the buyback machinery has been working aggressively. The Buffett question is always: at what P/IV did they buy? With the stock trading at px/IV = 0.955 today and 10y average P/E of 31.87, recent ASRs have been executed near or slightly above base IV — fine but not bargain repurchases. They are not deploying buybacks counter-cyclically the way a Buffett-grade allocator would; they are deploying them programmatically.

5. Dividends. SPGI has paid and grown a dividend for 50+ years (a Dividend Aristocrat / King). The payout ratio is modest, leaving most owner earnings for buybacks and acquisitions. This is the right balance for a compounder.

Communication quality. Investor-day disclosures are detailed and segment-level; management does not over-promise. The Energy software divestiture announcement was crisp and structurally rational.

Grade. The one drag is buybacks-at-any-price; the one big plus is M&A discipline measured over a decade. Net: this is a B+ allocator, not an A.

Capital allocator: B.

Industry

Buyer power — LOW. SPGI's customers are issuers (corporates and governments needing bond ratings), asset managers and asset owners (paying for index licenses, Capital IQ seats, and Platts), and dealers/brokers (paying for data feeds). No single customer is large enough to dictate terms. Issuers have to be rated to access institutional bond markets at scale — they are price-takers within a narrow band. Asset managers tracking the S&P 500 cannot threaten to leave without losing the benchmark their clients demanded. Buyer power is structurally weak across all five segments.

Supplier power — LOW. SPGI's main "inputs" are talent (analysts, engineers) and technology infrastructure (compute, software). Talent is competitive but not concentrated in any single supplier. Cloud spend is meaningful but commoditized across AWS/Azure/GCP. There is no choke-point supplier that can extract margin from SPGI.

Threat of new entrants — LOW for Ratings and Indices, MEDIUM for Market Intelligence and Mobility. New ratings agencies face NRSRO designation friction, decades of historical default-data accumulation, and an issuer-pays business model that rewards scale [1]. Indices are protected by incumbency-network-effects. Market Intelligence faces real entrant pressure: Bloomberg, FactSet, Refinitiv (LSEG), Pitchbook (Morningstar), and AI-native challengers all want the same data desk. Mobility (CARFAX) faces a long-tail threat from auto OEMs sharing telematics data directly.

Threat of substitutes — MEDIUM-LOW but rising. The credible substitute for ratings is internal credit analysis at large asset managers + machine-learning credit models — a slow grind, not a step-change. The credible substitute for the S&P 500 is direct indexing or alternative passive vehicles (factor, thematic ETFs that bypass SPGI licensing). The credible substitute for Capital IQ is an AI-research stack that can answer the same questions without seat-licensing. Substitute risk is the most dynamic Force right now.

Internal rivalry — LOW in Ratings (duopoly with Moody's), MEDIUM-HIGH in Market Intelligence and Indices. Ratings is structurally a Big-Two with a small Fitch. Indices is more contested (MSCI, FTSE Russell, Bloomberg Indices) but the S&P brand monopolizes large-cap U.S. equity benchmarking. Market Intelligence is competitive every day.

Value-pool location and trajectory. The value pool sits in (a) the metering layer of capital markets (per-issuance ratings fees, per-AUM index license fees) and (b) the workflow layer of finance professionals (per-seat data subscriptions). Both pools are growing modestly above GDP, with index-licensing economics widening as passive AUM continues to compound and Market Intelligence economics being challenged by AI-driven productivity that could collapse seat counts. Net: value pool is growing, but the rate of growth in MI may surprise to the downside as AI adoption matures.

Industry Verdict: Excellent for Ratings and Indices; Good for Commodity Insights; Average for Market Intelligence and Mobility. Weighted by revenue, the consolidated industry verdict is Good trending Excellent if Indices and Ratings keep growing as a share of mix.

Inversion

I am playing a short-seller. I am skeptical of every bull narrative on this page.

1. The single event that kills this. A second 2008 — specifically, a credit cycle in which S&P Global Ratings demonstrably under-rated default risk on a class of widely held instruments (private credit CLOs, sovereign EM debt, AI-infrastructure project finance) — followed by Congressional hearings and a Dodd-Frank-2.0 that strips NRSRO references from bank capital rules and replaces them with internal-models or a public-utility credit-rating bureau. Once you remove regulatory dependence on S&P ratings, issuer-pays economics collapse to commercial-information-services economics, which trade at 8-12x EBITDA, not 20-25x. SPGI's Ratings segment alone could lose half its multiple. This is not a fanciful scenario; the political ground is laid every time there is a credit blow-up. Private credit is currently the largest shadow-rated asset class in history, growing at 15-20% annually, and SPGI is one of the rating agencies anchoring it. A bad cycle there is the kill switch.

2. Why the moat is narrower than bulls think. Three reasons. (a) The Indices moat is a brand moat, and brand moats can be re-platformed when the cost of computing alternative benchmarks goes to zero. Direct indexing already lets large taxable accounts replicate the S&P 500 without paying the licensing fee on the wrapper — Vanguard pays SPGI for the brand, but a sufficiently large RIA platform increasingly does not. (b) The Capital IQ / Market Intelligence moat is a workflow moat, and AI agents are the workflow-killer of our era. A junior analyst who today opens 14 Capital IQ tabs to build a comp set will, in three years, ask an AI agent to do it from public filings + a vector-index of the same data. SPGI's per-seat economics could compress 20-40%. (c) Platts assessments are protected by bilateral contract references, but contract references can be replaced at re-papering events; the LIBOR-to-SOFR transition proved that even the deepest reference rate can be moved when the cost-benefit shifts.

3. Why management is worse than it appears. They have run programmatic ASRs at price/IV ratios above 0.9 throughout 2025 (four programs). A truly capital-disciplined allocator would slow buybacks when the stock is at base-IV and accelerate them only when it dislocates 25%+ below IV. Buying back stock near fair value is fine; calling it disciplined capital allocation is generous. The IHS Markit deal landed enormous goodwill on the balance sheet at peak data-company multiples. The 6.51% ROIIC over 5 years is the proof: incremental capital is no longer earning 30%+; it is earning single-digits because most of it sits in goodwill, not in earning assets.

4. What bulls are extrapolating that won't hold. Bulls extrapolate (a) ratings revenue growth from issuance volume + price/issuance, (b) index-licensing revenue growth from passive-AUM compounding, (c) Market Intelligence growth from cross-sell into Capital IQ, and (d) Mobility and Commodity Insights as long-duration data businesses. The reverse-DCF requires 8.04% growth at constant multiples for the price to make sense. That assumption requires AUM to keep compounding, AI not to compress MI seat counts, and credit cycles not to break the Ratings franchise. Three things going right. Bull cases compound errors when they require all three.

5. Valuation trap (multiple compression / regime change). P/E TTM 33.18 vs 10y avg 31.87 — modestly above its own history. The trap is that the 10y average was set during a zero-rate, passive-AUM-supercycle, AI-naïve era. If we transition to (a) higher real rates (3%+ 10y TIPS), (b) flat-to-declining passive-AUM net flows as Boomers de-accumulate, and (c) AI-driven seat-count compression in MI, the right multiple for SPGI is closer to a high-quality-but-mature data company at 20-22x earnings. On TTM EPS of roughly $13 (back-solving from P/E 33.18 at $426), a 22x multiple yields $286/share. Add a 25% credit-cycle haircut to ratings revenues and the bear scenario is $230-260.

If I am right, the stock could be worth $250 within 5 years.

Lollapalooza Bias Check

Active biases in me as I analyze SPGI right now:

Authority bias (high). This is a Buffett-and-Berkshire-adjacent name (Moody's, the duopoly counterpart, is a long-standing Berkshire holding); the franchise narrative is endorsed by people I respect. I should explicitly notice that the question "would Buffett buy SPGI today?" is not the same as "is SPGI a good Buffett-style business?" Buffett famously refuses to buy Moody's at full price even though he holds it; that distinction is doing real work and I should not conflate quality with value.

Confirmation bias (high). The scorecard already says composite 59 and px/IV 0.955 — both numbers nudge toward Hold/Trim. Once I have that prior, I look harder for reasons to confirm "fair value" than for evidence the stock is genuinely undervalued or overvalued. I should re-examine the IV high case ($672) on its merits before concluding the price is fair.

Anchoring bias (medium). The 10-year P/E average of 31.87 anchors my judgment of "normal" multiple. But that anchor was set during a uniquely favorable interest-rate and passive-AUM regime. Normalizing to a longer history (or to comparable IG-quality data companies) suggests 22-26x is the better anchor. Anchoring on the recent decade overstates the right multiple.

Recency bias (medium). The IHS Markit integration narrative and the 2026 SLB divestiture announcement feel fresh and constructive; that recency is making me grade management more favorably than the 6.51% 5-year ROIIC alone justifies.

Social-proof bias (medium). Every active manager has SPGI in the top quartile of their "high quality compounders" lists. That consensus is itself a contrarian signal — when everybody agrees a name is a compounder at fair value, the marginal incremental buyer at the next 1% of price has to come from somewhere.

Deprival super-reaction (low). I do not own SPGI, so the loss-aversion mechanic is not activating against "selling." If anything, the bias is toward FOMO on a name that has worked.

Incentive bias (low). Acknowledged: writing more flattering analyses gets engagement, and SPGI is a popular name. I should price that in by being slightly more critical than I instinctively want to be — which is what the inversion section forces me to do.

Net effect. Authority + confirmation + anchoring are the three live biases pushing me toward "high-quality, hold at fair value." To counter them, I weight the inversion case more heavily and require a meaningful margin of safety before recommending a buy.

10-Year Outlook

Same fundamental business model in 2036? Mostly yes. Ratings is regulator-anchored and barring a Dodd-Frank-2.0 will look essentially the same. Indices is brand-anchored; the S&P 500 will still be the S&P 500. Commodity Insights/Platts will look the same as long as physical commodity contracts continue to reference benchmark assessments. Market Intelligence is the segment most likely to look meaningfully different — AI-driven research workflows could compress seat counts and shift pricing models from per-seat to per-query or per-API-call. Mobility will look similar but with more telematics-driven data sources.

Customer base larger? Ratings: yes (global bond issuance grows with global GDP and private credit). Indices: yes (passive AUM has compounded at 10%+ for two decades; even decelerating to 5-7%, the AUM base is materially larger by 2036). MI: probably flat to modestly larger in seat count, but with more API-priced revenue. Mobility: yes (vehicle parc grows; CARFAX has further international expansion runway).

Profit per customer higher? Ratings: yes (price/issuance has compounded above CPI). Indices: yes (mix-shift to higher-fee factor and thematic ETFs). MI: uncertain — AI could compress per-seat economics even as enterprise value-per-customer grows. Net direction: modestly higher.

Moat wider? Ratings: probably the same — wide and rule-bound. Indices: probably wider because each year of compounding AUM increases switching costs for asset allocators. MI: probably narrower because of AI-disruption optionality. Net: wider, not by a lot.

Single biggest threat over 10 years. A combined credit-cycle blow-up + Dodd-Frank-2.0 that strips NRSRO references from bank capital rules. Probability over 10 years: low-to-moderate (~10-20%); severity if it occurs: high (multiple compression of 30-50% on the Ratings segment). Secondary threat: AI-driven Market Intelligence seat compression — probability moderate, severity moderate.

Confidence. The franchise is durable enough to underwrite. The price is fair, not generous, and the bear case has real teeth — but I can imagine the business in 2036 with reasonable specificity, the customer base is identifiable, and the value drivers are Buffett-knowable.

CONFIDENCE: medium

Position Guidance

• Recommendation: Hold
• Conviction: medium
• Target buy price: $335 (25% margin of safety to base IV $446; opens a meaningful starter position)
• Aggressive add price: $300 (approaches IV-low $251 with a moat-quality premium)
• Target trim price: $560 (above bull-case IV $672 by ~16% margin; signals multiple expansion that is no longer underwritable)
• Hard sell zone: above $670 (at or above IV high)
• Position sizing: 3-4% on a starter at <$335; up to 6-7% on full conviction at <$300; cap at 8% given regulatory tail risk on Ratings segment
• Time horizon: 10+ years; this is a hold-forever-at-the-right-price compounder, not a trade