SPGI › analyze
SPGI · Analyze
A portfolio-specific passage was removed from the public build.
Verdict: WATCH · conviction 6.0/10 · FV $405–480 (centre ~$440) · Entry $355–385 · Trim 28x fwd
A genuinely excellent business at a fair-to-slightly-cheap price. Under §0 of the framework that is "wait / watch," not "buy." Do not add here. Hold the existing share.
0. Read this first — the vendor snapshot for this name is void
Four of the numbers a screen would show for SPGI today are wrong, and three of them are wrong in the direction that makes the stock look cheaper than it is. S&P Global spun off its Mobility segment on 2026-07-01 as Mobility Global Inc. (NYSE: MBGL), a 1-for-1 distribution, record date 2026-06-15. Recast financials published 2026-07-06. Mobility becomes discontinued operations from Q3 2026 — so every trailing vendor field is still consolidated.
| Vendor field | Shows | Truth | Why |
|---|---|---|---|
PE(fwd) 20.18 |
20.2x — "historically cheap" | 23.2x | Yahoo's implied fwd EPS of $20.24 is a pre-spin figure. Post-spin FY26 guide is $17.50–17.75. Error +14.8% |
| "−25% off the high" | a crash | −21.9% | The $547.82 high included Mobility. MBGL is ~$19.50/sh today → retained value $427.92 |
| Revenue $15.34B | scale | ~$13.6B | Mobility was ~$1.75B. Every CAGR off the Yahoo series is on a stale basis |
P/B 3.82 · BVPS 106.86 · Graham IV $198.75 |
asset backing | meaningless | 60% of equity is IHS Markit goodwill. Tangible book is negative |
EPS(ttm) 16.43 · PE(ttm) 24.86 |
— | ✅ correct | Ties exactly: 14.66 + 8.81 − 7.04 |
And the error repeats across the peer group — see §5. Correcting SPGI alone while leaving
its comps on vendor data produces a worse answer than doing nothing. That mistake was made
and caught during this analysis; it is now
Knowledge/Playbook/pitfall-half-corrected-peer-table-invents-a-discount.md.
1. What the company now is
Four segments, not five. The spin removed the least S&P-like asset — a consumer-data business with no benchmark, no regulatory licence, and a 21.6% margin.
FY2025 recast, ex-Mobility
| Segment | % revenue | % op profit | GAAP margin | Q2'26 growth | Q2'26 adj margin |
|---|---|---|---|---|---|
| Ratings | 34.7% | 48% | 63.8% | +17% | 68% |
| Market Intelligence | 36.2% | 16% | 20.2% | +6% | 36% |
| Indices | 13.6% | 20% | 68.7% | +20% | 72% |
| Commodity Insights (Energy) | 15.4% | 15% | 41.0% | +2% ⚠️ | 48% |
Read that table twice, because it is the whole argument. Market Intelligence is the largest segment by revenue and the smallest but one by profit. Ratings + Indices are 48% of revenue but 68% of profit.
The segment most exposed to AI disruption contributes least to the profit pool. Anyone rating SPGI on revenue weights lets the weak segment set the tone for a company whose profits come from somewhere else. That is very likely the source of the de-rating — and it is also, in part, a real repricing rather than an error (§5).
2. Fundamentals — good business, three new cracks
Free cash flow — and the finding that matters most
Definitional note. SPGI's own FCF subtracts distributions to noncontrolling interests. That is not pedantry: S&P Dow Jones Indices is only ~73%-owned — CME Group holds the rest (verified; JV formed 2012, CME bought Dow Jones' residual 2.6% in 2013). NCI distributions were $321M in FY25. Vendor FCF omits this and overstates owner cash by ~6%.
| FY | OCF | Capex | NCI dist | Owner FCF | FCF/sh | Diluted EPS |
|---|---|---|---|---|---|---|
| 2018 | 2,064 | 113 | 154 | 1,797 | 7.10 | 7.73 |
| 2019 | 2,776 | 115 | 143 | 2,518 | 10.20 | 8.60 |
| 2020 | 3,567 | 76 | 194 | 3,297 | 13.62 | 9.66 |
| 2021 | 3,598 | 35 | 227 | 3,336 | 13.80 | 12.51 |
| 2022 | 2,603 | 89 | 270 | 2,244 | 7.05 | 10.20 |
| 2023 | 3,710 | 143 | 280 | 3,287 | 10.31 | 8.23 |
| 2024 | 5,689 | 124 | 287 | 5,278 | 16.92 | 12.35 |
| 2025 | 5,651 | 195 | 321 | 5,135 ⬇ | 16.83 ⬇ | 14.66 |
🚩 FY2025 owner FCF fell 2.7% ($5,278M → $5,135M; adjusted FCF −3.9%) and FCF/share fell despite a $5B buyback. Revenue grew 8%. Causes: capex +57%, NCI distributions +12%, a $168M severance year. One year is not a trend — but the buyback did not cover the cash decline, and that is the single most important negative in the file.
| FCF CAGR window | Basis | CAGR |
|---|---|---|
| 2018→2025 (7y) | Adjusted — cleanest across the merger | 15.4% |
| 2017→2021 (4y) | Pre-merger standalone | 17.1% |
| 2022→2025 (3y) | Post-merger, adjusted | 11.5% |
| 2020→2025 (5y) | Owner FCF | 9.3% |
| 2022→2025 raw | off the distorted trough | 31.8% — do not quote** |
FY22 owner FCF was 43% below its underlying $3,955M on $1,711M of merger/divestiture cash costs. FY22 is unusable as a CAGR endpoint. The honest number is ~15% over 7–8 years, decelerating to ~9–11.5% on clean recent windows.
Capital allocation — historically excellent, recently mediocre
| FY | $ buyback | Avg price | vs spot $408 |
|---|---|---|---|
| 2016 | 1,123 | 104.09 | +292% |
| 2018 | 1,660 | 180.70 | +126% |
| 2020 | 1,164 | 302.16 | +35% |
| 2021 | 0 | 384.64 | abstained at the all-time high ✅ |
| 2022 | 12,004 | 339.61 | +20% |
| 2024 | 3,301 | 442.05 | −7.6% ❌ |
| 2025 | 5,001 | 488.67 | −16.4% ❌ |
| 2016–25 | $29,795M | $301.28 wtd | +35.6% ✅ |
Abstaining entirely in 2021 at the top is the strongest single piece of evidence of capital discipline in the file. Against that: $8.3B spent across 2024–25 at ~$462 average is ~12% underwater. The 2026 programme is >$7B — over 5% of market cap — running at ~$429 in H1. Buying more as the price falls is the right direction; the 2025 tranche shows they are not price-sensitive in practice.
Capex is 1.3% of revenue. Acquisitions contributed <0.5% of growth — this is a genuinely organic compounder. But FY25 M&A stepped up to $2.02B from ~$300M/yr, a change in behaviour worth watching from the management culture that paid $44B for IHS Markit.
Balance sheet — not a live issue
Net debt $11.0B · net debt/EBITDA ~1.3–1.5x · interest coverage 22.6x · D/A 24% · maturities laddered to 2062. Current ratio 0.95 is not a concern — $4.1B of the current liability is deferred revenue.
⚠️ Flagged disagreement. A "leverage goes to 2.7–2.8x EBITDA" figure surfaced in research. Fundamentals showed contrary arithmetic from the primary balance sheet (~$12.4B YE26 net debt ÷ ~$8.6B EBITDA ≈ 1.45x) and reaching 2.7x would need ~$10B more debt than the funding plan supports. Unresolved, but immaterial — at 22x coverage leverage is not a thesis risk at either figure.
Share count — the merger cost seven years
| Window | Rate |
|---|---|
| Pre-merger 2014→2021 | −1.63%/yr |
| Post-merger 2022→Q2'26 | −1.73%/yr |
| Q2'26 YoY | −3.5% |
IHS Markit added +31.7% shares in one year. Net income compounded 12.5%/yr 2018→2025 while diluted EPS compounded only 9.6% — a 2.9pp/yr per-share drag for seven years. Against §0's "per-share is what I own," this is the central historical criticism of SPGI: the business grew faster than the owner's claim on it.
Per-share — the owner's view
| CAGR 2018→2025 (7y) | |
|---|---|
| Revenue/share | 10.7% |
| Owner FCF/share | 13.1% |
| Diluted EPS | 9.6% |
Good, not exceptional, for a business of this quality — and it is the right lens, because it already absorbs the merger dilution.
Dividend — Aristocrat badge, non-Aristocrat behaviour
| Window | Dividend CAGR |
|---|---|
| 2015→2025 (10y) | 11.3% |
| 2020→2025 (5y) | 7.5% |
| 2022→2025 (3y) | 5.0% |
| 2026 | +1.0% |
🚩 Dividend growth has effectively stopped. The FY24 (+1.1%) and FY26 (+1.0%) raises are token increases preserving a 53-year streak while cash is redirected to buybacks. Yield 0.95%, FCF payout ~22% — perfectly safe, and not an income holding.
This matters directly to the stated rotation toward growth + income. SPGI does not serve the income side of that goal and is actively becoming less able to. It is a total-return and buyback story wearing an Aristocrat badge.
ROIC — report both, they mean different things
| FY | ROIC (with goodwill) | ROIC (ex-goodwill) |
|---|---|---|
| 2022 | 7.5% | 24.9% |
| 2023 | 6.6% | 23.6% |
| 2024 | 9.3% | 35.5% |
| 2025 | 10.6% | 45.5% |
The 2021→2022 collapse (105% → 7.5%) is 100% accounting — $31B of goodwill landed in one quarter while NOPAT rose. Never quote that series without saying so.
- With goodwill 10.6%, rising ~1.5pp/yr — this is the return on the price paid for IHS Markit. It is now above cost of capital. The merger is finally earning its keep.
- Ex-goodwill 45.5%, rising steeply — the return on the operating business. A genuine high-moat number.
- Tangible invested capital is negative ~$5B. The business consumes no tangible capital.
3. Moat — 8.0–8.2/10, and widened by the spin
| Segment | Moat sources (strongest first) | Rating |
|---|---|---|
| Indices | Intangibles (S&P 500 as a consumer brand) → efficient scale → network effects (SPX/VIX derivative liquidity) → switching costs borne by end-investors, not the paying issuer → cost advantage | 9.5 |
| Ratings | Intangibles (NRSRO licence — legal, not reputational) → efficient scale → switching costs (mandates name agencies; cost falls on bond buyers who don't pay) | 9.0 |
| Commodity Insights | Intangibles (Dated Brent written into physical contracts) → network effects → moderate switching costs. No cost advantage | 7.5 |
| Market Intelligence | Switching costs (procedural, dissolvable by an NL interface) → intangibles (proprietary datasets) → scale. No network effect, no efficient scale | 5.0 |
Profit-weighted: 8.15/10. Revenue-weighted: 7.38. Use the profit weighting — owners own cash flow, and MI's revenue weight overstates its economic weight by more than 2x.
The adversarial test — where the moat is thinnest
Market Intelligence, and it is not close. The attack is not to build a better Capital IQ — it is to make Capital IQ a commodity input. Hebbia and Rogo already integrate Capital IQ alongside FactSet, LSEG, PitchBook, Preqin and Daloopa as interchangeable pipes. Daloopa raised a $47M Series C in May 2026 shipping MCP connectors into ChatGPT/Claude/Perplexity. The switching cost in MI is procedural (workflow retraining), and a natural-language layer dissolves exactly that by removing the retraining.
Management has conceded the mechanism, on the Q2 call: "Customers are looking for ways to minimise or manage token expenses, including building solutions in-house," and MI's softness was attributed to "elongated sales cycles from complex AI contract negotiations."
A vendor with pricing power does not have elongated negotiations. Elongation is what happens when the buyer has found a credible alternative and is using it as leverage.
⚠️ The "high-90s retention" figure is worth zero as reassurance — per
pitfall-logo-retention-masks-dollar-retention, it counts accounts, not dollars, and SPGI
does not disclose dollar-based NRR for MI. The absence of that disclosure is itself the
informative fact. But be fair: unlike the TTD case that generated the pitfall (>95% logo
retention against +0.4% revenue — arithmetic that forces wallet loss), SPGI pairs it with
+6% organic. MI is not yet a wallet-loss story; it is one where the instrument that would
detect wallet loss has been withheld.
Commodity Insights is thinner than consensus admits. Argus has completed 14 IOSCO benchmark
reviews and is the reference in several products. IOSCO compliance is a moat-flattener, not a
moat — it wrote down the recipe, so any funded entrant passing a PwC review is procedurally
as legitimate as Platts. Contrast NRSRO, which restricts entry instead. Same word,
"regulated," opposite effect — now captured as
Knowledge/Playbook/pattern-regulation-standardises-or-restricts.md. Corroboration: Q2 Energy
grew +2–3% with "contract renewal challenges prompted flexible pricing."
Indices — the hostile case cannot be made to work. Vanguard's 2012 CRSP switch proved a move is executable, yet no large-cap US core fund has left the S&P 500 in fourteen years. The mechanical cost of switching is only ~$100K; the real barrier is that the benchmark is written into contracts SPGI is not party to and cannot be removed from unilaterally — pension mandates, IPS documents, 401(k) menus, performance fees. Add the SPX/VIX derivative complex (a pure network effect outside the issuer's control) and a consumer-grade brand. While MSCI reported two consecutive quarters of declining average basis points, S&P DJI posted a 13th consecutive record quarter, asset-linked fees +22% on $6.35T linked AUM. Being the price leader is not a weakness when volume compounds like that.
The Indices risk is beta, not competition — ~65% of the segment is a levered claim on equity market levels. A 30% drawdown takes ~20% off segment revenue with no management remedy.
Ratings cyclicality — quantified, and rising
| FY | Ratings rev | YoY | Op profit | YoY | Margin |
|---|---|---|---|---|---|
| 2021 | 4,097 | +13.6% | 2,629 | +18.3% | 64.2% peak |
| 2022 | 3,050 | −25.6% | 1,672 | −36.4% | 54.8% trough |
| 2024 | 4,370 | +31.2% | 2,707 | +45.2% | 61.9% |
| 2025 | 4,724 | +8.1% | 3,013 | +11.3% | 63.8% |
The 2022 rate shock is the cleanest stress test available: 1.42x downside operating leverage, 940bp of margin compression in one year, and Q3'22 transaction revenue −56%.
🚩 The mix is getting more cyclical, not less. Transaction revenue is now 55.7% of Ratings, up from 52.0% a year ago = ~20% of company revenue tied directly to issuance volume. Applying the 2022 template today ≈ −$1.0B operating profit ≈ ~12% of FY26 adjusted EPS. Hold that against any "defensive compounder" framing.
Offsets: ~44% of Ratings is non-transaction surveillance income that did not fall in 2022; ~55–60% of company revenue is subscription or non-transaction.
Disruption — the distinction the whole thesis rests on
What AI commoditises: the description of a fact. Summarising a 10-K, extracting a segment table, building a comp set. Every dollar of MI revenue justified by "it saves an analyst four hours" is impaired on a 5–10 year view.
What AI cannot touch: the creation of a fact by consent. An LLM can tell you everything about the S&P 500. It cannot tell you what the S&P 500 is. That is not a knowledge problem but a coordination problem — the index is authoritative because capital agreed in advance to be bound by it. An AI that builds a demonstrably better 500-stock index has produced a research paper, not a benchmark. Likewise a rating is a licence, not an opinion — AI does not dissolve legal artifacts.
AI attacks the analysis of data. It cannot attack data that exists only because someone with authority declared it. ~83% of SPGI's profit sits in declared-fact businesses.
Three risks the bull framing elides: (1) a monopolist still negotiates with a poorer customer — if AI compresses buy-side budgets 30%, that lands on every line item at renewal, including ones with no substitute; (2) passive's own maturity — when flows normalise, Indices converts from a 20%-growth business to an equity-beta business at a 68% margin; (3) $36.48B of goodwill concentrated in the weakest-moat segment. If MI organic breaks below ~4%, the impairment test gets uncomfortable.
Moat defence — sound strategy, unattractive economics, correctly sized
MI was reorganised into "Kensho Data & Platforms" and "Enterprise Solutions" — read the structure, not the press release: it separates the durable asset (data, feeds, APIs) from the perishable one (the desktop). That is the correct diagnosis of which half survives.
Evidence it is working: Kensho LLM API calls 5x QoQ, AI-connected clients past 500 (+70%), and the load-bearing one — AI-powered MI customers grew ACV 60% faster than average. Plus a disclosed vendor-consolidation renewal at +20% over several years.
⚠️ But that is management-selected data. A 5x from an unstated base is not a number
(pitfall-growth-ratio-against-collapsing-base). "AI customers grow ACV 60% faster" is
selection, not causation — the firms adopting AI fastest are the sophisticated, growing
ones that would have grown ACV faster anyway. And the strategy is intrinsically
deflationary: moving from seat licences to consumption pricing while clients optimise token
spend is right defensively, but the picks-and-shovels layer is worth less per dollar of
client value than the terminal was.
The right frame: MI is now a hedge, not a growth engine. It costs ~16% of the profit pool to hold optionality on being the licensed data supplier to the AI stack. If it works, MI re-rates; if it fails, SPGI loses its smallest profit slice and remains a benchmark monopoly. The >$7B buyback says management would rather shrink the share count than fund a large MI transformation — on this asset base that is the correct call, and a real improvement on the IHS Markit-era instinct.
4. Sentiment — negative narrative, positive fundamentals, positive insider
Why the stock fell — tested, not assumed
| Candidate | Verdict | Evidence |
|---|---|---|
| AI disruption of financial data | ✅ Primary driver | Field-wide; FDS −28.9% at 15.8x with a Hold and a target below spot |
| Feb 2026 guidance cut | ✅ The trigger | 2026-02-10: FY26 guided $19.40–19.65 vs $19.96 consensus → −9.7% in one day. Cause was elective AI capex, not demand |
| Mobility separation | ⚠️ Mechanical only | ~3.5pp of the headline drawdown; also drove fake 7/07 target cuts |
| Ratings slowdown | ❌ Falsified | Ratings +17%, a record quarter; transaction +25%; IG issuance projected >$2T in 2026, the largest on record |
| Broad market de-rate | ❌ Falsified | SPY +13.8% YTD at an all-time high |
| Index fee pressure | ❌ Falsified | Benchmark revenue +15%; Indices guided 12–14%, the fastest division |
| Regulatory | ❌ Immaterial | Recordkeeping fines |
Analyst positioning: 6 Strong Buy / 17 Buy / 1 Hold / 0 Sell — and zero rating changes in twelve months. All 15 analyst actions were price-target adjustments with the rating maintained; five firms raised targets during the decline. The sell side marked to market; it did not change its mind.
The Q2 print was a beat reported as a miss
| Basis | Q2'26 EPS |
|---|---|
| GAAP diluted | $4.12 (+18%) |
| Pro-forma (ex-Mobility) | $4.08 (+26%) |
| Adjusted | $4.83 (+23%) |
| Vendor consensus (pre-spin basis) | $4.75 |
Wires compared pro-forma $4.08 against the unadjusted $4.75 and printed a "14% EPS
miss." The stock fell 5.2% intraday. The correct comparison — adjusted $4.83 vs $4.49 —
was a 7.6% beat, on revenue of $3.678B pro-forma (+11% organic CC) vs $3.64B, with adjusted
operating margin +200bp to 54.3%. Now captured as
Knowledge/Playbook/pitfall-post-spin-consensus-basis-mismatch.md.
Insider activity — the strongest bull signal in the file
⚠️ Pitfall applied — the Feb-2026 "Stock Award (Grant) at price 0.00" entries are RSU
grants, not buying, and are excluded. Yahoo's insider_purchases summary returned all
zeros for SPGI while real buying sat in insider_transactions.
| Date | Insider | Role | Price | Value |
|---|---|---|---|---|
| 2026-02-11 | Hubert Joly | Director | $398.94 | $997,459 |
| 2026-04-29 | Martina Cheung | CEO | $429.93 | $998,297 |
| 2026-04-30 | Robert E. Moritz Jr. | Director | $434.03 | $500,001 |
| 2026-05-01 | Catherine R. Clay | Officer | $431.39 | $1,078,475 |
| Total | ~$3.57M |
The round dollar values are the signature of deliberate fixed-dollar orders — a coordinated show of confidence. Joly bought the day after the guidance collapse, into a −12% three-day fall.
The contrast makes it stronger. In 2025 insiders sold at $500–561 (Peterson $9.63M in May, Tavernier at $555, Saha at $561). In 2026 they bought at $399–434, with zero sales. Correct-direction insider trading is rare and this is a clean instance.
⚠️ The named risk nobody is pricing — circularity
Ratings' 2026 upside is driven by hyperscaler and AI-infrastructure debt issuance (IG transaction revenue +high-20s; AI-related issuance may double to ~$400B). An analyst pressed exactly this on the call and management's answer was appropriately guarded.
SPGI's earnings beat is partly funded by the same AI capex boom that is de-rating its multiple. If AI capex cracks, SPGI is hit twice at once — issuance volume falls and the data-disruption narrative worsens. These are correlated risks, not diversified ones, and it is why 23x should not be treated as a floor.
5. Valuation — the correction that decides the verdict
Peer table, rebuilt on guided EPS
Yahoo's forward EPS is inflated across this entire peer group — 12–15% on every name.
| Name | Price | Verified FY26 guide | Vendor fwd P/E | TRUE fwd P/E | vs SPGI |
|---|---|---|---|---|---|
| SPGI | 408.39 | $17.50–17.75 ✓ | 20.18x | 23.2x | — |
| MCO | 477.24 | $16.50–17.00 ✓ | 25.20x | 28.5x | SPGI −18.7% |
| MSCI | 560.18 | ~$19.50–20.00 ⚠️ est. | 24.90x | ~28.4x | SPGI −18.3% |
| FDS | 277.04 | $17.25–17.75 ✓ | 14.10x | 15.8x | SPGI +46.4% |
The 5.3-turn discount to Moody's is real — it survives correction and is marginally wider than the vendor figure suggested. But so is the 46% premium to FactSet: the market has not capitulated on the mix.
The decisive test — how much of the MCO discount is deserved? Weight SPGI's own segment profit by the multiple each segment earns (Ratings 51.3% @ 28.5x · Indices 15.6% @ 28.4x · Commodity Insights 16.0% @ ~20x · MI 17.1% @ 15.8x):
Mix-fair blended multiple ≈ 25.0x. SPGI trades at 23.2x — about 7% below what its own mix justifies.
Neither "unjustified 5-turn gap" nor "fully explained by mix." The market has correctly re-rated SPGI for owning a FactSet-like half, then taken ~7% more off the top. That residual is the opportunity, and it is single-digit.
Historical band — mind the adjusted/GAAP trap
Built from unadjusted closes per pitfall-adjusted-close-breaks-multiple-bands:
| Year | Close | GAAP P/E | Adj P/E |
|---|---|---|---|
| 2019 | 258.33 | 30.0 | 27.1 |
| 2020 | 311.00 | 32.2 | 26.6 |
| 2021 | 446.48 | 35.7 | 32.6 |
| 2024 | 471.17 | 38.2 | 30.0 |
| 2025 | 494.41 | 33.7 | 27.7 |
⚠️ Do not compare today's 23.2x adjusted forward to the ~33.7x GAAP band — GAAP EPS post-2022 carries ~$1B/yr of IHS Markit amortisation that pre-2022 GAAP does not. That comparison manufactures ~30% of false cheapness. The valid adjusted band is 26.6–32.6x, median 27.7x; SPGI sits at ~24.5x trailing adjusted — about 12% below median.
Caveat limiting the band's authority: pre-2022 SPGI was ~85% Ratings+Indices, a better mix than today's, and 2019–21 sits in ZIRP. Mean reversion to 27–32x is not automatic — the business that earned those multiples no longer exists.
Models
| Model | Result | Weight |
|---|---|---|
| SOTP | $381 / $438 / $496 (cons/base/gen) | 40% |
| Bogle ER | Bear +6.0% · Base +9.9% · Bull +13.3% /yr | 25% |
| Adjusted band | ~12% below median | 20% |
| Reverse-DCF | requires ~8.5% owner-FCF growth for 10yr | 15% |
| DYT | $473 — a price-decline artifact | ~5%, confirmatory only |
| Graham | $198.75 | 0% — inapplicable |
Graham is refused, not merely discounted. Ex-goodwill BVPS is −$18.07; ex-all-intangibles −$59.17. SPGI has negative tangible book. The $198.75 values the accounting residue of a 2022 stock deal as if it were asset backing, on a business with 71% gross margins and $195M of capex. It gets zero weight and must not anchor the range.
DYT inverts here — the yield rose because the price fell, not because the dividend was
raised (pitfall-dyt-inverts-when-price-caused-the-yield). At a 23% payout the dividend is a
rounding error in the return.
Reverse-DCF is the sobering one. On corrected post-spin owner FCF of ~$4.57B (not the $5.1B a vendor feed implies), $408.39 requires ~8.5% annual owner-FCF growth for a decade — up from ~7% on the naive base — against a company that just delivered −2.7%.
The SOTP crux — and why the popular bull argument is circular
| Case | Implied MI value | Implied multiple |
|---|---|---|
| Conservative | $23.7B | 19.6x EBIT, 4.5x rev — above FactSet |
| Base | $8.2B | 6.8x EBIT, 1.6x rev |
| Generous | −$7.4B | negative |
"The market is giving you Market Intelligence for free" is not a fact about the market. It is a restatement of the assumption that Ratings and Indices deserve full peer multiples. Across a defensible span of comps the implied MI value runs from −$7B to +$24B. Under conservative multiples the market pays more than a FactSet multiple for MI.
A material correction inside the SOTP: the prior pass capitalised profit SPGI does not own. S&P DJI is 73%-owned, so Indices EBIT of $1.51B is only $1.10B economically SPGI's — and the tie-out is exact (27% of $1.51B, taxed at 22%, = $318M vs the $321M actually distributed). This removed ~$9–11B of EV and moved the SOTP range down $15–20/share — more than the peer correction added.
6. Synthesis — weighted verdict
Two corrections landed during this analysis and they cut in opposite directions.
Pushing up: the discount to Moody's is 5.3 turns, not 2.0. SPGI is decisively not priced as a FactSet — it carries a 46% premium to FDS.
Pushing down: (a) the SOTP fell $15–20/share once S&P DJI's 27% minority was removed from the crown jewel; (b) the reverse-DCF bar rose from 7.0% to 8.5% on corrected owner FCF, against an FY25 that delivered −2.7%; (c) only ~7% of the 18.7% MCO discount is unexplained by mix.
They offset almost exactly. Net: a good business roughly 7–8% below fair value with a falling owner-FCF base. Under §0 — great company + fairly priced = wait/watch — that is not a buy.
Bull vs bear, fairly stated
A portfolio-specific passage was removed from the public build.
The honest tension
The bull case and the bear case are the same trade. AI is simultaneously funding the ratings beat and compressing the multiple. Going long here bets that S&P's proprietary data survives commoditisation and that the AI capex cycle holds for 18 months. Those are correlated risks, not diversified ones.
Verdict
WATCH · conviction 6.0/10
- Fair value $405–480 (centre ~$440, +8%)
- Entry $355–385 (20.1–21.8x fwd) — anchored at/below the conservative SOTP of $381. Under conservative comps you are overpaying at today's price. The 52-week low of $361 shows the zone is reachable.
- Trim 28x fwd — where SPGI reaches parity with MCO and MSCI. With a materially weaker profit mix than either, parity is where it becomes expensive.
Had only the peer correction landed, this would be 6.5 and ACCUMULATE-on-weakness. The NCI and owner-FCF findings are what hold it at WATCH.
A portfolio-specific passage was removed from the public build.
7. Data gaps — flagged explicitly
- MSCI's FY26 guide unverified — the ~28.4x is annualised from Q2 adjusted EPS of $4.94. The MSCI gap is softer than the MCO gap; treat accordingly.
- MIS-standalone 24.7x EV/EBIT depends on an assumed 16x for Moody's Analytics and MCO's ~55/45 split.
- Exact debt maturity schedule by year not retrieved — the 10-K footnote table did not survive HTML extraction. Tenors (2026–2037, 2052, 2062) and near-term ST debt ($2.57B) are sourced. Not load-bearing at 22x coverage.
- Leverage disagreement unresolved — ~1.45x (primary-source arithmetic) vs a 2.7–2.8x figure found in research. Immaterial to the verdict.
- FY23→FY24 Ratings +31.2% spans a 2025 segment recast; treat that single YoY as ±3pp. The 2021→2022 stress figures are within one release and are clean.
- Pre-2016 history unusable — FY2015 OCF of $356M reflects the $1.6B DOJ ratings settlement.
- Post-spin FCF (~$4.57–5.1B) is estimated by pro-rating Mobility out of FY25; no clean post-spin FCF guide obtained.
- A second divestiture is in flight — $120M of Energy geoscience/petroleum-engineering software held for sale at 2026-06-30. Re-check segment comparability at Q3.
- roic.ai not used — free tier caps history at 2 years on this account, confirming
pitfall-roicai-free-plan-caps-history-at-two-years. History built from SEC XBRL companyfacts (CIK 0000064040) and primary 8-K releases. - FY25 restated adjusted EPS ~$15.88 is inferred, not read from a filing. ⚠️ Note the FY26 guide of $17.50–17.75 sits below the $17.83 originally reported for FY25 — anyone comparing those directly will wrongly conclude earnings are shrinking. On the restated base it is +11% growth.
8. What would change the verdict
🟢 Upgrade toward ACCUMULATE: - Price into $355–385 without a thesis break - Owner FCF/share returns to growth in FY26 — the single most important open question - Commodity Insights back to the guided 6–8% without renewal discounting - SPGI discloses dollar-based NRR ≥100% for Market Intelligence
🔴 Downgrade / break: - MI organic constant-currency below ~4% for two consecutive quarters, or a disclosed dollar-NRR below 100% - Owner FCF declines a second year - Goodwill impairment taken against Market Intelligence - An AI-capex crack hitting IG/structured issuance volumes - Buyback funded by debt beyond the stated 2.0–2.5x end-2027 path
Agents: Fundamentals · Moat · Sentiment · Valuation. Phase 3 debate round run on the forward-P/E peer comparison — the manager's initial correction was itself wrong and was overturned by Fundamentals, then verified against Moody's primary filing.