The Goldman Sachs Group, Inc. (NYSE: GS) — A Genuinely Better Goldman, Priced as if the Up-Cycle Never Ends

Report date: 2026-06-10 | Price (ref.): ~$1,001 | Shares out: ~295M | Market cap: ~$296B Sector: Financials — Capital Markets (Investment Banking & Brokerage) | CIK: 0000886982 | FYE: December

An independent fundamental-research note. Primary sources: SEC EDGAR filings (FY2025 10-K filed 2026-02-25; Q1 2026 results; 8-K and Form 3/4 corpus), Goldman Sachs earnings-call and conference transcripts (Q1 2026, Q4 2025, Q3 2025, Bernstein May-2026), and public market data. All figures reconciled to filings; management commentary treated as hypothesis. Except for the clearly-labeled “Claude’s Take” block below, this note contains no investment recommendation and no price target — the body discusses valuation only as embedded expectations and scenarios.

⚡ Claude’s Take

This block is the author’s own subjective opinion and general information only — not investment advice. The analytical body below carries no position and no price target.

Verdict: HOLD / accumulate-on-weakness — a real, durable quality upgrade in the franchise meeting a textbook cyclical-peak multiple. Not a short (the cycle and the buyback are with you), but no margin of safety at ~$1,000. “A genuinely better Goldman, priced for the up-cycle to never end.”

Directional valuation zone (the author’s own view): Goldman today is a structurally mid-teens-ROTE franchise that just printed a 21.3% ROTE quarter (Q1 2026) into the teeth of a record M&A and trading cycle, and the stock — at ~$1,001, ~2.8x book / ~3.0x tangible book / ~17.5x trailing earnings — sits at the 98th percentile of its own ten-year valuation range. A reverse-Gordon back-solve says ~3.0x TBV embeds a permanent ~22% ROTE — i.e., the cyclical peak as the new normal — versus management’s own through-the-cycle target of “mid-teens.” On the mid-teens (~16–17%) ROTE the firm guides to, a justified multiple is closer to ~2.0–2.3x tangible book, which on ~$335 TBV/share implies an accumulation band of ~$680–$770 (≈ 11–13x normalized EPS). I’d call genuine value below ~$650 (≈2.0x TBV, a level last seen in early-2025) and would trim enthusiasm above ~$1,050. At today’s price the forward return is roughly “earn the mid-teens ROTE minus multiple compression” plus the ~1.8% dividend — decent if earnings durability beats the skeptics, thin if the cycle rolls.

The framing is quality-re-rating-meets-cyclical-peak, not value and not pure momentum. What the market is pricing correctly: this is a materially de-risked, higher-quality Goldman than the 2018-vintage one — the money-losing consumer experiment (Marcus, GreenSky, GM/Apple cards) is all but gone; Asset & Wealth Management is now a $3.6T-AUS, fee-led engine with a credible alternatives build ($750B fee-paying alts targeted by 2030); “durable” financing revenue has roughly doubled to $11.4B (37% of FICC + equities); and a friendlier Basel III endgame / G-SIB reproposal hands Goldman a capital tailwind it is already spending via record buybacks. That deserves a premium to the sub-1x-book Goldman of the 2010s. What I think it’s pricing incorrectly: it capitalizes a double cyclical peak — advisory revenue +89% YoY, equities trading at records, and a net credit-loss benefit in 2025 (a $1.1B reserve release, the opposite of normal) — as the durable run-rate. M&A volumes near the 2021 record do not compound forever; trading reverts; and provisions can only normalize upward from a release. Goldman is the most cyclical earnings stream in the megabank complex (beta 1.31, half of revenue is trading + banking), and it is trading at its richest-ever multiple on its best-ever earnings. That is the combination you fade, not chase.

Conviction: medium. Flips bullish if ROTE demonstrably holds ~17–18% through a banking/trading down-quarter and a credit normalization (proving the AWM/financing durability thesis is structural, not cyclical) — or simply on price, in the ~$680–770 zone. Flips bearish (toward trim) if the multiple pushes past ~3.2x TBV on still-peak earnings, or if a capital-markets air-pocket collides with private-credit losses while the stock sits at 3x book. You own this one on weakness, not here.

1. Executive Summary

The Goldman Sachs Group is the premier global investment bank and trading house, and FY2025 was the best year in its history as a public company: $58.28 billion of net revenue (+9% YoY), $17.18 billion of net income, diluted EPS of $51.32 (+27%), a 15.0% return on equity and 16.0% return on tangible equity, on $1.81 trillion of assets and $125.0 billion of shareholders’ equity (FY2025 10-K, filed 2026-02-25). The first quarter of 2026 was stronger still — a 19.8% ROE / 21.3% ROTE quarter, the second-best revenue and EPS in the firm’s history. After a decade in which Goldman was the perennial “value trap” of the megabanks — chronically sub-1x book, dinged for an earnings stream too cyclical and capital-intensive to deserve a premium — the stock has re-rated hard, from roughly tangible book in 2022–23 to ~3.0x tangible book today at ~$1,001, near its all-time high and at the 98th percentile of its own ten-year valuation range.

Two things are simultaneously true, and the investment debate is entirely about which dominates. First, the franchise is genuinely better and more durable than it was: the disastrous consumer-banking detour (Marcus deposits aside, the lending businesses — GreenSky, the GM and Apple co-brand cards — destroyed capital and are now exited or held-for-sale); Asset & Wealth Management has been rebuilt into a $3.6 trillion-AUS, fee-led business with 33 consecutive quarters of long-term fee-based inflows and a credible alternatives platform; and Goldman has deliberately grown its “durable” revenue (management/wealth fees plus FICC and equities financing, which together hit a record $11.4 billion in 2025, 37% of total FICC + equities revenue). Second, FY2025 and Q1 2026 sit at or near a cyclical peak: investment-banking advisory revenue rose 89% YoY in Q1 2026; equities trading is at records; and the firm booked a net credit-loss benefit of $1.11 billion in 2025 (an Apple Card reserve release) rather than a normal provision. Peak banking + peak trading + a credit tailwind, capitalized at a peak multiple, is the textbook setup an analyst is paid to flag.

The moat is real but narrower and more cyclical than a JPMorgan’s. Goldman’s advantages are genuine — a top-two-globally M&A and equity-underwriting franchise with decades of relationship intangibles, a scaled markets/financing operation, and a fast-growing capital-light asset manager — but they are advisory-and-trading-led, which means the revenue is inherently lumpier and more correlated to capital-markets activity than a deposit-funded universal bank’s. Goldman does not have JPMorgan’s $2.5T cheap-deposit moat or Charles Schwab’s captive-retail float; its $501B deposit base is real and growing but is largely a funding tool, not a profit engine. The result: Goldman out-earns Citi but structurally under-earns JPMorgan on through-cycle ROTCE, and trades — for the first time in a decade — at a richer tangible-book multiple than at almost any point in its history.

Capital allocation is now a clear strength. Management returned $16.78 billion to common shareholders in 2025 ($12.36B buybacks + $4.42B dividends), shrank the share count ~6.6% in under two years (315.8M → 295.0M), raised the dividend to a $18/share forward run-rate (from $9.00 in 2022), and bought back a record $5 billion in Q1 2026 alone — drawing CET1 down to 12.5% deliberately, against a 11.4% requirement, because a friendlier regulatory regime (a “neutral” Basel III endgame, a G-SIB surcharge reproposal, a 320bp improvement in the stress-capital buffer) is freeing capital. The buyback is the single most reliable lever in the thesis. This report takes no position and sets no price target; the body that follows analyzes Goldman as embedded expectations, scenarios, and risks.

2. Business Overview

The Goldman Sachs Group, Inc. (NYSE: GS), founded in 1869 and headquartered in New York, is a global financial institution serving corporations, financial institutions, governments and high-net-worth individuals. As of December 31, 2025 it carried $1.81 trillion of total assets, $501 billion of deposits, $125.0 billion of total shareholders’ equity, and employed ~47,400 people (FY2025 10-K). It is the smallest of the US “Big Four” universal banks by assets (versus JPMorgan’s $4.42T) but is the single most capital-markets-concentrated of them — the purest large-cap expression of investment banking, trading, and (increasingly) asset management in the public markets.

Three reportable segments

Following the 2022–2024 strategic reset, Goldman reports three segments. The FY2025 results:

(Source: FY2025 10-K MD&A. Platform Solutions is presented as a runoff/transition segment; the GM card was transitioned in August 2025 and the Apple Card portfolio was moved to held-for-sale in January 2026.)

Global Banking & Markets (~71% of net revenue) is the heart of Goldman and the source of both its prestige and its cyclicality. The segment decomposes into roughly four revenue engines: (1) Investment banking — FY2025 fees of $9.34 billion (+21%), itself split across advisory (M&A, the highest-margin and most prestige-driven line, where Goldman is #1 globally), equity underwriting (IPOs, follow-ons, convertibles), and debt underwriting; (2) FICC — fixed income, currency and commodities, $7.24 billion of market-making revenue, the most balance-intensive trading line; (3) Equities — $10.77 billion of market-making revenue plus $5.03 billion of commissions and fees, where Goldman’s prime-brokerage franchise is a genuine scale leader; and (4) Financing — secured lending, prime balances, and Capital Solutions origination embedded across FICC and equities. The mix matters: advisory and underwriting are event-driven and the most cyclical; market-making is volatility-driven; financing is balance-driven and durable. Management’s strategic project is to grow (4) faster than (1)–(2), structurally lowering the segment’s cyclicality — and the data shows it working (financing a record $11.4B, +17% CAGR since 2021). Critically, GBM now embeds a large and growing financing business — secured lending, prime brokerage balances, structured credit, and “Capital Solutions” (private-credit origination and warehouse financing) — which management classifies as durable because it is balance-driven and recurring rather than transaction-driven. FICC + equities financing reached a record $11.4 billion in 2025, ~37% of combined FICC + equities revenue, growing at a ~17% CAGR since 2021. INTERPRETATION: this is the most important quality-improvement story inside the most cyclical segment — Goldman is deliberately raising the floor under a famously volatile revenue stream.

Asset & Wealth Management (~29% of net revenue) is the strategic pivot. AWM manages $3.606 trillion of total assets under supervision (AUS) as of December 2025 (up from $3.137T in 2024 and $2.812T in 2023 — ~$224B of net inflows plus ~$245B of net market appreciation in 2025 alone), having taken in $224 billion of net inflows in the year ($168B long-term, $56B liquidity products). The long-term inflow streak — 33 consecutive quarters — is the single best evidence that AWM is a genuine flows franchise rather than a beta-rider, because it persisted through the 2022–23 market drawdown. Wealth flows are running ahead of target (7–9% annualized versus the 5% long-term goal), and the third-party (RIA/intermediary) wealth channel grew AUS ~25% in Q1 2026 — a deliberate distribution-broadening beyond the legacy UHNW-direct model. Revenue splits across Management and other fees (the recurring, capital-light core — running at a $3.1B/quarter rate exiting Q1 2026, +14% YoY), Private banking and lending (spread income on UHNW lending and deposits), Equity and debt investments (on-balance-sheet principal, which Goldman is shrinking to a more capital-light model), and Incentive fees (carried interest on alternatives). Management has set explicit, aggressive targets: a 30% pretax margin (raised from 25%), a high-teens return on the segment, $750 billion of fee-paying alternatives AUS by 2030 (from ~$430B today), and $1 billion of annual incentive fees. AWM is the leg of the franchise the market is being asked to pay a structurally higher multiple for.

Platform Solutions is the graveyard of the consumer strategy and is being wound down. The GM co-brand card was transitioned to a new issuer in August 2025; the Apple Card portfolio was moved to held-for-sale in January 2026 (the transition is expected to take ~24 months and was net EPS-accretive ~$0.46 in Q4 2025 because a $2.5B reserve release exceeded the lost revenue). The economic message is clear: Goldman is exiting sub-scale, high-loss consumer lending and redeploying that capital into GBM financing and AWM. Transaction banking (corporate cash management / deposits) remains.

Funding, net interest income, and the deposit base

Goldman funds a $1.81 trillion balance sheet with $501 billion of deposits (up from $433B in 2024 and $387B in 2022 — a deliberate, multi-year build), plus unsecured long-term debt, secured financings, and shareholders’ equity. The deposit growth is strategically important but easy to over-read: a large slice is consumer/Marcus savings and brokered/private-bank deposits that are more rate-sensitive and less sticky than JPMorgan’s retail operating accounts, and transaction-banking deposits that are operationally sticky but modest in margin. Net interest income is a secondary earnings driver for Goldman (unlike a retail bank where it is the core), and in 2025–26 it has been pressured by deposit-cost competition — management flagged that the AWM private-banking NIM headwind “will persist for much of 2026.” INTERPRETATION: Goldman’s deposit franchise is a funding achievement, not the profit annuity that a true deposit moat provides; this is the single biggest structural reason it under-earns JPMorgan through the cycle.

How Goldman makes money

Unlike a retail bank, Goldman’s economics are ~50% fee/advisory + trading and ~50% spread/financing/principal, with the fee/trading half dominant and volatile. The firm earns: advisory and underwriting fees (event-driven, highest-margin, most cyclical); trading spreads and commissions (volume-driven, volatile); financing spreads (balance-driven, durable); recurring management fees on AUS (capital-light, the prized annuity); net interest income on its lending and deposit book; and principal investment gains. The defining structural fact is that a far larger share of Goldman’s revenue is tied to the level of capital-markets activity than at any deposit-funded peer — which is exactly why the stock is the highest-beta name in the megabank group (β ≈ 1.31).

Scale and headcount

Goldman generated $58.3B of net revenue from ~47,400 employees — roughly $1.23 million of net revenue per head, far above a retail bank (JPMorgan’s ~318,000 staff produce ~$0.58M/head) because Goldman has no branch network and is concentrated in high-value advisory, trading and asset management. Headcount fell from a ~49,000+ peak as the consumer business unwound and “One Goldman Sachs 3.0” (AI-enabled operating model) took hold — a deliberate efficiency drive toward the 60% efficiency-ratio target. The flip side of high revenue-per-head is high cost-per-head: compensation is ~31.8% of net revenue, the largest single expense line and the reason the efficiency ratio sits well above a deposit-funded peer’s.

3. Industry Dynamics

Goldman competes in three overlapping industries, each with distinct structure.

1. Investment banking (advisory + underwriting). This is a global oligopoly at the top — a handful of bulge-bracket banks (Goldman, Morgan Stanley, JPMorgan, Bank of America, plus elite boutiques like Evercore, Centerview, Lazard in advisory) capture the bulk of large-cap M&A and equity-capital-markets fees. Barriers to entry at the mega-deal level are genuinely high: the business runs on relationships, reputation, league-table credibility, balance sheet, and senior banker talent — intangibles that cannot be bought quickly. Goldman has been #1 or #2 in global announced M&A for decades and reports a ~$300B league-table lead in 2026 YTD, “its largest lead ever at this point in the year” (Bernstein, May-2026). But the industry is brutally cyclical — fee pools swing 40–60% peak-to-trough with the M&A and IPO cycle, which is itself a function of CEO confidence, financing conditions, and equity valuations. The 2022–2023 trough (post-rate-shock M&A freeze) cut Goldman’s IB fees roughly in half from the 2021 peak. Verdict: structurally attractive oligopoly, viciously cyclical revenue.

2. Markets / trading (FICC + equities). A scale-and-technology oligopoly that has consolidated in Goldman’s favor over fifteen years — European banks (Deutsche, Credit Suisse/UBS, Barclays) have retreated, leaving the US bulge bracket (Goldman, JPMorgan, Morgan Stanley, BofA, Citi) to capture share. This is a “winner-take-most” business (management’s own phrase): the scaled players with the best technology, balance sheet, and prime-brokerage franchises win the flow, and the financing/prime balances are sticky. Equities financing in particular is a genuine moat — multi-year prime-brokerage relationships with the largest hedge funds. But trading revenue is volatility-dependent: it booms in dislocated markets (2020, 2022, 2025) and fades in calm ones. Verdict: a consolidating, structurally improving oligopoly, but revenue keyed to market volatility.

3. Asset & wealth management. A scale-and-flows business with a powerful secular tailwind toward alternatives (private equity, private credit, real assets, infrastructure) where fees are higher and stickier than in commoditized public-market beta. The asset-management industry overall is bifurcating: passive/ETF beta is a fee-compression race to zero (BlackRock/Vanguard win on scale), while alternatives and active outcome-oriented strategies retain pricing power. Goldman is positioning AWM at the high-fee end (UHNW wealth + alternatives), which is the correct strategic call. But it is entering a crowded field — Blackstone, Apollo, Ares, KKR, Brookfield, Blue Owl and the other alt managers are larger and more focused in private markets, and BlackRock/Morgan Stanley dwarf Goldman in wealth. Verdict: structurally the most attractive of the three industries, but Goldman is a credible challenger, not the leader.

Sizing the fee pools (why the cyclicality is so violent). Global investment-banking fees (advisory + ECM + DCM) form a pool that the industry estimates at roughly $80–110B in a normal year, swinging to ~$130B+ at a 2021-style peak and collapsing toward ~$70B in a 2022–23 trough — a ~40–50% peak-to-trough range that flows almost directly to the bottom line because the cost base (bankers) is semi-fixed. Goldman’s ~8–9% share of that pool means its advisory/underwriting revenue is a leveraged bet on the level of global deal and issuance activity, which is in turn a function of CEO confidence, financing spreads, and equity valuations — none of which Goldman controls. Trading fee pools (FICC + equities) are larger (~$150–200B industry-wide) and more volatility-driven: they expand in dislocated markets (2020, 2022, 2025) and contract in calm ones. This is why a single peak year tells you almost nothing about the through-cycle earnings power — and why capitalizing a peak at a peak multiple is the analytical trap.

The private-credit / alternatives capital cycle (Marathon lens). The most important supply-side dynamic right now is the flood of capital into private credit and alternatives. Returns there have been attractive, a friendlier regulatory regime is freeing bank balance sheets to lend, and every major player — Goldman included — is racing to originate. By Marathon’s logic, abundant capital chasing high returns is the reliable precursor to those returns mean-reverting: spreads compress, underwriting standards loosen, and the next default cycle reveals which books were built on discipline versus momentum. Goldman’s entire private-credit/financing growth has occurred inside a benign-credit regime that has not yet turned. That the CEO spent his most airtime in Q1 2026 defending the asset class — “this has been a very long credit cycle… when you do have a cycle turn in a recession, you will see higher losses across the space” — is itself a late-cycle tell.

Cross-cutting: regulation. As a US G-SIB, Goldman is subject to the most demanding capital, liquidity, stress-testing (CCAR / stress-capital-buffer) and resolution requirements in finance. For a decade this was a pure headwind that suppressed ROE and forced the sub-1x-book valuation. The regime is now easing at the margin: management cites a “more neutral” Basel III endgame finalization, a G-SIB surcharge reproposal, greater CCAR transparency, and a 320bp improvement in its own stress-capital buffer since 2024. This is a genuine, quantifiable tailwind to capital return — and a key reason the multiple has re-rated. The capital-cycle lens (Marathon): high returns and a friendlier regulatory regime are attracting capital back into capital-markets and private-credit businesses across the industry — a classic late-cycle signal that competition for the next dollar of financing/origination revenue will intensify and compress returns. The reproposal that frees Goldman’s capital frees every competitor’s too.

4. Competitive Position

The moat is real, but it is an advisory-trading-intangibles moat, not a deposit moat — and that distinction governs the entire valuation debate.

Where Goldman genuinely has an advantage:

• Advisory franchise / relationship intangibles (the strongest leg). Goldman’s #1 global M&A position is a decades-compounded asset: CEO and board relationships, sector expertise, a brand that signals quality on a deal, and a senior-banker bench that competitors cannot replicate quickly. The financial proof: a ~$300B league-table lead in 2026 YTD and advisory revenue +89% YoY in Q1 2026, achieved while raising rather than cutting price. This is a Greenwald “customer-captivity + intangibles” advantage — a CEO doing a transformational deal does not shop on price, and the switching cost is reputational. If this moat deteriorated, advisory market share and fee realization would visibly fall; they are doing the opposite.

• Scaled markets + financing flywheel. Goldman’s prime-brokerage and financing balances are sticky multi-year relationships with the largest hedge funds and corporates. The record $11.4B of FICC + equities financing revenue (37% of the total, +17% CAGR since 2021) is the durable, balance-driven core that survives a trading downturn. This is a genuine economies-of-scale advantage: the marginal cost of intermediating another dollar of flow on an already-built platform is low, and scale begets the best flow.

• Capital Solutions Group + the AI-infrastructure financing wave. Goldman formed its Capital Solutions Group in 2024 to combine advisory, financing and private-credit origination. Management quantifies the cross-sell: in deals where Goldman both advises and finances, total revenue runs ~140% of the advisory fee alone. With ~$700–800 billion of AI-infrastructure financing needs forming over the coming years (management’s framing), this is a credible adjacency that leverages the advisory moat into balance-sheet revenue.

Where Goldman is disadvantaged versus the best universal banks:

• No cheap-deposit moat. JPMorgan funds itself with $2.5T of sticky, low-cost retail and operating deposits earning a structural ~2.7% deposit margin; Goldman’s $501B deposit base is a wholesale-ish funding tool, not a high-margin annuity. This is the reason Goldman structurally under-earns JPMorgan on through-cycle ROTCE despite comparable franchise quality in capital markets.
• Earnings volatility. Half of revenue is advisory + trading. Goldman’s ROE swung from 23% (2021 peak) to ~10–11% (2022–23 trough) to 15% (2025). The market historically — and rationally — pays a lower multiple for a more volatile earnings stream; the current 3x-TBV multiple implicitly argues the volatility has been tamed. That is the bull’s central, and unproven, claim.
• Challenger, not leader, in AWM. Against the focused alt-manager pure-plays (Blackstone, Apollo, Ares, Blue Owl) and the wealth giants (Morgan Stanley, BlackRock), Goldman is building credibly but from behind.

Direct peer comparison (FY2025 / latest):

Greenwald market-share-stability test. The cleanest test of a genuine moat is whether market share is stable over time (stability implies customer captivity and barriers; churn implies a competitive commodity). Goldman passes in advisory — it has held a top-two global M&A position for decades and is currently extending its league-table lead to a record ~$300B — and passes in scaled equities/prime, where it has gained share as European banks retreated. It is weaker but improving in FICC (share gains, but a more commoditized, balance-intensive business) and unproven in AWM alternatives (it is gaining share, but from a challenger position against entrenched leaders). On the Greenwald taxonomy, Goldman’s advantage is best described as customer captivity + intangibles (reputation/relationships) in advisory, reinforced by economies of scale in markets/financing — a real composite moat, but one concentrated in the most cyclical revenue and absent the demand-side captivity of a sticky deposit base that anchors a JPMorgan.

ROIC / ROE-durability test. A moat must show up in returns that persist. Goldman’s ROE has been genuinely volatile — ~23% (2021) → ~10–11% (2022–23) → 15% (2025) → 19.8% (Q1-26) — which is the signature of a cyclical advantage, not a stable one. JPMorgan, by contrast, held ~17–20% ROTCE through the same cycle. The bull’s claim is that the durable-revenue build has now narrowed Goldman’s ROE band upward and tightened it; the data is consistent with the floor having risen (the 2022–23 trough was shallower than prior cycles) but does not yet prove the band has compressed to JPMorgan-like stability. This is the single most important unresolved question in the whole thesis (see the body).

Verdict: a real, durable advantage in advisory and scaled markets/financing, plus a credible (not dominant) AWM build — but a narrower and more cyclical moat than JPMorgan’s, lacking the cheap-deposit engine. Goldman is a high-quality franchise; it is not the highest-quality bank, and it is now priced as though it were.

5. Growth History and Forward Opportunities

History — a violently cyclical line with a rising underlying trend. Net revenue ran $44.6B (2020) → a $59.3B COVID-stimulus/SPAC peak (2021) → $47.4B (2022) → $46.3B (2023 trough) → $53.5B (2024) → $58.28B (2025). EPS tells the cyclicality even more starkly: $30.06 (2022) → $22.87 (2023) → $40.54 (2024) → $51.32 (2025). The 2023 trough reflected the M&A freeze and the consumer-strategy write-downs (GreenSky impairment, Marcus/card reserves); the 2024–2025 recovery reflects the capital-markets rebound plus the cleanup of the consumer drag. INTERPRETATION: the trend line is genuinely rising — Goldman’s 2025 record is ~25% above the prior (2021) non-stimulus peak on revenue — but the path is a sawtooth, and 2025 is a tooth-top, not a mid-cycle point.

Forward growth opportunities, ranked by quality:

1. AWM fee growth (highest quality). Long-term fee-based net inflows have now run for 33 consecutive quarters; management/other fees compound mid-teens; the alternatives platform targets $750B fee-paying AUS by 2030 (from ~$430B) with $75–100B raised annually. This is the capital-light, recurring growth the market wants and is paying for. Risk: it is a multi-year build against entrenched competitors, and incentive fees are themselves cyclical.

2. Financing / Capital Solutions (durable, balance-driven). The $11.4B durable-financing base growing ~17% is the structural floor-raiser. The AI-infrastructure financing wave is a real, large adjacency. Risk: private-credit growth carries credit risk that has not yet been tested by a default cycle (see the body).

3. M&A / IB recovery (high quality but cyclical). Management is explicitly calling a multi-year up-cycle — backlog at a four-year high, corporate M&A +62% YoY, advisory revenue +89% in Q1 2026, and a view that 2021’s record is “not the ceiling.” If correct, IB fees have room to run. Risk: this is precisely the most cyclical and least predictable revenue; calling a multi-year up-cycle near a record is the bull’s most extrapolative claim. Notably, private-equity sponsor activity has lagged (“PE is down 4%… that’s the reverse of what we’ve seen”), and the IB recovery is so far corporate-led — meaning the ~$1T of sponsor dry powder is a future catalyst that has not yet fired.

4. Buyback-driven EPS growth (mechanical, reliable). Even with flat revenue, a 4–6%/year share-count reduction compounds EPS. This is the most dependable growth lever and is underpinned by the regulatory capital release.

A forward earnings bridge (illustrative, not a forecast). Decomposing where the next few years of EPS could come from clarifies the quality mix. Starting from ~$52–55 of normalized EPS power: (a) buyback mechanically adds ~4–6%/year to EPS at the current pace, the most reliable contributor; (b) AWM fee growth (management/other fees compounding mid-teens, alternatives scaling toward $750B) adds a steady, capital-light layer; © durable financing growth (~17% historically) raises the base; (d) IB/markets is the swing factor — if management’s multi-year up-cycle call is right, it adds materially, but it is also the line that could subtract 20–40% in a down year; and (e) credit normalization is a near-term headwind (the 2025 benefit reverses). Netting these, a reasonable read is that Goldman can compound EPS at a high-single-digit rate through the cycle from durable sources (buyback + fees + financing), with IB/markets adding cyclical upside and downside on top. The market, at ~17.5x trailing / 98th-percentile multiple, appears to be paying for the cyclical-upside scenario as the base case.

Verdict: a mix of genuinely high-quality growth (AWM fees, durable financing, buyback) and high-quality-but-cyclical growth (IB/markets). The durable components are real and improving; the headline 2025–26 growth rate is flattered by a cyclical banking/trading peak that will not compound linearly.

6. Financial Quality

Returns. FY2025 delivered ROE 15.0% and ROTE 16.0% (up from 12.7% / ~13.5% in 2024), and Q1 2026 spiked to 19.8% / 21.3%. Management’s through-the-cycle target is mid-teens ROE (and GBM “mid-teens through the cycle”). INTERPRETATION: the FY2025 number is at the high end of the through-cycle target and the Q1 2026 number is well above it — i.e., current returns are running hot relative to what management itself guides as sustainable. A 16% ROTE business is a good business; it is not a 21% structural business, and the valuation (next section) implicitly assumes the latter.

Revenue quality. The $58.28B (+9%) is high-magnitude but mixed-quality: GBM +18% is the cyclical engine (IB fees +21%, equities at records); AWM +2% is the steadier, lower-growth annuity this year (held back by NIM compression in private banking). The encouraging structural signal is the rising share of “durable” revenue (management fees + financing = a record $11.4B financing alone), which management says has “doubled” and “raised the floor meaningfully.” That claim is directionally supported by the data but unproven through a downturn.

Provision / credit — a key quality flag. FY2025 booked a net credit-loss benefit of $1.11 billion, versus a $1.35B charge in 2024 — a ~$2.5B favorable swing driven largely by the Apple Card reserve release as that book moved to held-for-sale. This is a non-repeatable tailwind that flattered 2025 earnings. Q1 2026 already turned back to a $315M provision build (single-name impairments + macro overlay). Normalizing credit alone takes a couple of dollars off the “clean” EPS run-rate — an important adjustment before capitalizing 2025 as a base.

Efficiency. The FY2025 efficiency ratio was 64.4% (operating expenses / net revenue), versus management’s 60% target and well above JPMorgan’s ~52%. Goldman is structurally less efficient because it is people-heavy (compensation ratio ~31.8% of net revenue) and lacks the operating leverage of a deposit-funded retail bank. The 60% goal is credible but not yet hit; the AI/“One Goldman Sachs 3.0” program (40,000 staff on the GS AI Assistant, a claimed ~20% software-development productivity gain) is the lever, but management has declined to quantify the savings — appropriately skeptical territory.

Multi-year return and book-value walk. The compounding story is real even through the cyclicality. Net revenue: $47.4B (2022) → $46.3B (2023) → $53.5B (2024) → $58.3B (2025). Net income: $11.3B → $8.5B → $14.3B → $17.2B. Diluted EPS: $30.06 → $22.87 → $40.54 → $51.32. ROE: ~10% (2022) → ~7.5% (2023 trough, depressed by consumer write-downs) → 12.7% (2024) → 15.0% (2025). Tangible book per share has compounded steadily — $316.02 (2024) → $335.49 (2025) — despite returning $16.8B to shareholders, the hallmark of a business out-earning its distributions. INTERPRETATION: the underlying trend is genuine upward compounding, but the slope is heavily influenced by where in the cycle you measure — the 2023 trough (ROE ~7.5%) and the 2025 peak (15%) bracket a through-cycle reality closer to the mid-teens guide than the 19.8% Q1-26 print.

Comp ratio and operating leverage. Compensation and benefits ran ~31.8% of net revenue in 2025 (down ~100bp YoY in Q1 2026), the disciplined end of Goldman’s historical 33–36% range — management has held comp growth below revenue growth, the source of positive operating leverage in an up-cycle. The risk is the reverse: in a down-cycle, revenue falls faster than comp can be cut (you cannot fire bankers fast enough to protect a collapsing fee year), which is precisely why bank-trough margins are so much worse than peak margins. The 60% efficiency target assumes both continued comp discipline and the AI productivity unlock landing.

Balance sheet / capital. $1.81T assets; $125.0B total equity (~$114B common); CET1 14.3% Standardized / 15.1% Advanced at year-end 2025, drawn deliberately to 12.5% after the record Q1 2026 buyback (still ~110bp above the 11.4% requirement). Deposits grew to $501B. Book value per common share $357.60 (+6.2%); tangible book $335.49. Liquidity and capital are ample; the firm is releasing capital, not building it — a posture only justified if the regulatory easing is durable. Goldman’s supplementary leverage ratio and liquidity coverage ratio remain comfortably above requirements; the binding constraint is the stress-capital buffer (CCAR), which the 320bp improvement has materially loosened.

Verdict: economics are good and improving, but 2025 is a high-water mark. The returns are running above the through-cycle target, a credit benefit (not charge) flattered the year, efficiency is still above target, and the durable-revenue thesis — while genuinely progressing — has not been stress-tested. Goldman’s economics improve with scale in AWM and financing; they do not compound smoothly in the banking/trading core.

7. Capital Allocation

This is the clearest strength in the thesis, and the most reliable plank under the stock.

Shareholder returns. Goldman returned $16.78 billion to common shareholders in 2025 — $12.36B of buybacks + $4.42B of dividends — and bought back a record $5 billion in Q1 2026 alone ($6.4B total return that quarter). Share count fell from 315.8M (July 2024) to 295.0M (April 2026), ~6.6% in under two years, and the buyback has been accelerating as regulatory capital frees up. The dividend has risen from $9.00/share (2022) to a ~$18/share forward run-rate ($4.50/quarter, raised in Q4 2025, “+50% from a year ago”), at a conservative ~22–25% payout. INTERPRETATION: management is using the regulatory-capital tailwind to shrink the share count aggressively at ~3x tangible book — which is the one caveat (buying back rich stock is less value-accretive than buying back cheap stock, the discipline JPMorgan’s Dimon explicitly applies; Goldman is buying because it can, capital-wise, more than because the stock is cheap).

M&A / strategic capital — a story of discipline after indiscipline. The 2016–2021 consumer push (Marcus, GreenSky acquired for ~$2.2B in 2021, the Apple and GM card partnerships) was a genuine capital-allocation error — it consumed capital and management attention and produced multi-billion-dollar losses. Quantifying the damage: the consumer/Platform Solutions effort accumulated cumulative pre-tax losses estimated in the billions across 2020–2023 (the segment alone lost ~$3B+ pre-tax over its life), GreenSky was acquired for ~$2.2B in 2021 and sold in 2024 at a loss, and the 2023 ROE trough (~7.5%) was substantially a consumer-driven write-down year. This is not a rounding error — it is the clearest evidence that this management team is capable of value-destroying strategic mistakes, and it happened within the past five years. The credit to current management is the speed and decisiveness of the exit: GreenSky sold, GM card transitioned (Aug 2025), Apple Card moved to held-for-sale (Jan 2026), with capital redeployed into higher-returning GBM financing and AWM. The lesson cuts both ways — it demonstrates management will correct mistakes, but it earns a permanent skeptic’s discount on any future “strategic” foray, and it tempers how much credit one should extend to management’s current multi-year bullish framing.

Reinvestment. Capital is flowing to (a) GBM financing / Capital Solutions (durable, balance-driven), (b) the AWM alternatives build (capital-light fee growth), and © technology / AI. The on-balance-sheet principal investment book is being deliberately shrunk in favor of a capital-light third-party model — a correct, ROE-accretive shift.

Compensation & incentives. Comp ratio ~31.8% of net revenue (down ~100bp YoY in Q1 2026) — high in absolute terms (Goldman is a talent business) but trending the right way. The proxy ties senior compensation to ROE/ROTE, book-value growth, and relative TSR. Insider Form 4 activity in 2026 is the routine post-vesting cluster (April equity-award settlement; sell-to-cover and grants), with no signal-bearing open-market purchases — typical for a megabank and not a tell either way.

Verdict: management has allocated capital intelligently recently — disciplined exit of the consumer error, aggressive and growing shareholder returns, a capital-light strategic pivot — but the 2016–2021 consumer detour is a real blemish, and the current buyback is being executed at a premium multiple because capital is available, not because the stock is cheap. Net: a genuine strength, with an asterisk.

8. Changes and Headwinds — Last Two Years

Strategic changes (mostly positive):

• Exit of consumer lending — GreenSky divested; GM card transitioned (Aug 2025); Apple Card portfolio held-for-sale (Jan 2026). Removes a multi-year drag and reserve volatility. The Apple savings program is being retained/serviced (no transition agreement), a residual.
• AWM re-targeting — pretax-margin target raised to 30%, return target to high-teens, alternatives target $750B fee-paying by 2030, $1B incentive fees. A clear, capital-light strategic priority.
• Capital Solutions Group (2024) — combining advisory + financing + private credit; positioned for the AI-infrastructure financing wave.
• One Goldman Sachs 3.0 (AI operating model) — firmwide AI deployment; productivity claims as yet unquantified.

Capital / regulatory developments (positive):

• A “more neutral” Basel III endgame finalization, a G-SIB surcharge reproposal, greater CCAR transparency, and a 320bp improvement in the stress-capital buffer since 2024 — collectively freeing capital and funding the accelerated buyback and dividend.

Headwinds / risks that have emerged:

• Private-credit scrutiny. Management spent unusual airtime in Q1 2026 defending private credit against “an enormous amount of negative sentiment,” conceding that “this has been a very long credit cycle… when you do have a cycle turn in a recession, you will see higher losses across the space.” Goldman’s private-credit and financing growth has occurred entirely within a benign-credit regime; it is the single least-tested part of the franchise.
• NIM compression in AWM private banking/lending — deposit-cost headwinds management says “will persist for much of 2026,” capping the AWM spread businesses.
• Macro / geopolitical — Q1 2026 commentary turned more cautious (Middle East conflict, energy prices, AI-driven software disruption), tempering the IPO/sponsor outlook.
• The cyclical setup itself — 2025’s credit benefit won’t repeat (Q1 2026 already a $315M build); advisory +89% YoY and record equities are cyclically elevated.

Verdict: the structural changes strengthen the franchise (cleaner, more durable, better capital posture); the cyclical and private-credit headwinds weaken the durability of the 2025 earnings level. Net: a better business sitting on a more fragile earnings peak.

9. Risk Analysis (Risk Matrix)

Catastrophic-loss assessment: low probability of a total loss (G-SIB capital/liquidity buffers, regulatory backstop), but Goldman carries genuinely higher tail risk than a deposit-funded bank — a large trading book, prime-brokerage counterparty exposure, and a fast-growing, untested private-credit book. The realistic severe-downside scenario is not insolvency but a cyclical earnings halving plus a multiple de-rate — the combination that took the stock from ~$420 to ~$290 in 2022.

10. Valuation Discussion (Embedded Expectations)

No price target; no recommendation. This section frames what the current price requires.

Where the stock trades. At ~$1,001, Goldman is valued at ~2.80x book ($357.60) / ~3.0x tangible book ($335.49) / ~17.5x trailing EPS (~$54.78–$57.52 TTM) / ~15.7x forward EPS (~$65 next-year estimate), with a ~1.8% dividend yield. On the firm’s own ten-year history, this is the 98th percentile composite (P/E 97th, P/B 99.7th, P/S 99.7th on its own valuation history) — i.e., Goldman has essentially never been more expensive on its own metrics. For a decade the stock lived below tangible book; it now trades at 3x.

The embedded-expectations math (the crux). Use a simple Gordon-style justified multiple, P/TBV = (ROTE − g) / (COE − g). At a ~10% cost of equity and ~4% long-run growth:

• 3.0x TBV back-solves to a permanent ~22% ROTE — essentially the Q1 2026 peak (21.3%) extrapolated forever.
• Management’s through-the-cycle target is mid-teens ROE / ~16–17% ROTE. Plug 16.5% in: justified P/TBV ≈ (0.165 − 0.04)/(0.10 − 0.04) ≈ 2.08x, ≈ ~$700/share on $335 TBV.
• A more generous durable-18% ROTE (crediting the AWM/financing quality upgrade): P/TBV ≈ 2.33x ≈ ~$780.
• The bear (mid-cycle 14% ROTE, modest de-rate to 1.6x TBV): ≈ ~$540 — roughly the 52-week low.

INTERPRETATION: the current price is not absurd if you believe Goldman now structurally earns ~20%+ ROTE — i.e., that the franchise has been permanently re-rated to JPMorgan-like durability. But that is the exact claim the cyclical setup undermines: 2025/Q1-26 returns are inflated by peak banking, peak trading, and a credit benefit, against a management guide of “mid-teens.” The multiple capitalizes the peak as the mean.

Scenario analysis (illustrative, ~2-year EPS power):

Even the bull case (durable high-teens ROTE, multiple held) sits roughly at today’s price — meaning the stock is discounting the bull-to-momentum band already. The asymmetry is unfavorable: limited upside if everything goes right, ~30–50% downside if the cycle reverts.

Shareholder yield as a sanity check. At ~$296B market cap, Goldman’s 2025 capital return of $16.78B is a ~5.7% total shareholder yield (~1.8% dividend + ~4% buyback) — attractive in absolute terms and the most reliable component of the forward return. But two cautions: (a) the buyback is being executed at ~3x tangible book, so each dollar retires less book value than it would at a discount (buying $1 of TBV costs ~$3) — a capital-return strength but a value-creation mediocrity; and (b) the yield is funded by a regulatory-capital release that is policy-dependent. A reverse-DCF that credits the ~5.7% yield plus mid-single-digit EPS growth and a held multiple gets you a high-single-digit forward return; the same math with any multiple compression from the 98th percentile turns negative quickly.

Relative comps. Goldman at ~3.0x TBV / 16% ROTE sits richer-than-JPMorgan-on-TBV (~2.85x / ~20% ROTCE) despite lower through-cycle returns and higher earnings volatility — a hard combination to defend on fundamentals. The bull’s retort is that Goldman’s growth mix (AWM fees, alternatives, financing) is improving faster off a lower base, justifying multiple convergence; the bear’s is that JPMorgan earns a higher, steadier return and carries a heavier capital tax, so Goldman should trade at a discount, not a premium, on tangible book. Morgan Stanley trades at a similar/higher TBV multiple but on a more durable wealth-fee mix (wealth management is ~half of MS revenue, far stickier than trading). Citi at ~0.9x book is the cyclical-value alternative for investors who want financials exposure with a margin of safety rather than a quality premium. On a quality-adjusted basis (return level × return stability ÷ multiple), Goldman screens as the most fully-priced of the large IB-exposed banks — which does not make it a short (the cycle and buyback are tailwinds), but does make the entry point unattractive.

Verdict: the price embeds a permanent peak-ROTE outcome. What must be true to justify ~$1,000 is that Goldman has structurally become a ~20% ROTE business — not the mid-teens one management guides to. The evidence for a durable quality upgrade is real but incomplete; the evidence that 2025/26 is a cyclical high is strong.

11. Variant Perception

Consensus view. Sell-side is moderately constructive but not euphoric (AZI: ~5 strong-buy / 3 buy / 14 hold / 1 sell; mean target ~$935, below the current ~$1,001). The consensus narrative: “Goldman has been transformed — cleaner business mix, durable revenue, AWM inflows, regulatory tailwind, aggressive buyback — and deserves its re-rating.” The hold-heavy rating distribution and a mean target below spot suggest the Street thinks the quality story is right but the price has run ahead of it.

The strongest bull case. Goldman is a genuinely de-risked, higher-quality franchise: the consumer error is excised; durable revenue (fees + financing) has doubled to ~$11–12B and is structurally floor-raising the most cyclical segment; AWM is a credible $3.6T, fee-led, alternatives-growing engine; the regulatory regime is easing, freeing capital for a self-reinforcing buyback that compounds EPS 4–6%/year; and a multi-year M&A up-cycle (backlog at four-year highs, ~$1T of sponsor dry powder yet to fire, an AI-infrastructure financing super-cycle) means the “peak” is actually a floor for a higher plateau. On this view, ~20% ROTE is the new structural normal and 3x TBV is fair.

The strongest bear case. This is the most cyclical earnings stream in the megabank complex, printing its best-ever numbers — advisory +89% YoY, record equities, a credit benefit not a charge — at its richest-ever multiple (98th percentile own-history, 3x TBV). Management itself guides to “mid-teens” ROE, ~4–5 points below the Q1-26 print the multiple capitalizes. Half the revenue mean-reverts; the credit benefit is non-repeatable (Q1-26 already a build); private credit is untested by a default cycle; and the buyback — the bull’s reliable lever — is being executed at 3x book, destroying optionality. The capital-cycle lens warns that the same regulatory easing freeing Goldman’s capital is freeing every competitor’s, attracting capital into financing/private-credit and compressing the very returns being extrapolated. On this view, fair value is ~2.0–2.3x TBV (~$680–780), and the stock discounts perfection.

The 3–5 assumptions that decide it:

1. Is through-cycle ROTE ~16% (management’s guide) or ~20% (the multiple’s implication)? The single most important question.
2. Is the durable-revenue / AWM quality upgrade structural enough to dampen the historical cyclicality? (Floor-raising, yes; cyclicality-eliminating, unproven.)
3. Does the M&A/trading cycle have multiple years left, or is 2025–26 the top?
4. Does the regulatory capital tailwind persist (funding the buyback), or partially reverse?
5. Does private credit/financing survive its first real default cycle without a material loss?

What would falsify each side: Bull falsified if ROTE drops back to low-teens in the next banking/trading down-quarter and the multiple de-rates — proving 2025/26 was a cyclical tooth-top. Bear falsified if ROTE holds ~18%+ through a capital-markets soft patch and a credit normalization, proving the durability thesis structural rather than cyclical.

12. Fact vs. Interpretation Table

13. Open Questions

1. What is the true through-cycle ROTE once banking/trading normalize and credit reverts from a benefit to a normal charge — 16% (guide) or closer to the 20% the price implies?
2. How much of the record $11.4B “durable” financing revenue survives a trading downturn and a credit cycle — is the floor genuinely raised, or is “durable” partly a benign-regime artifact?
3. Does the regulatory capital tailwind persist through a political/policy cycle, and what happens to the buyback pace if it reverses?
4. Can AWM actually close the gap to Blackstone/Apollo/Blue Owl in alternatives, or is $750B-by-2030 a stretch target that the multiple has already paid for?
5. What is the real, quantified AI/efficiency benefit — management has pointedly declined to put numbers on “One Goldman Sachs 3.0”; is the 60% efficiency target reachable?
6. Succession — Solomon’s tenure has been eventful (consumer reversal, partner departures); what is the bench and timeline?
7. How concentrated is the private-credit/prime book in a few large counterparties (Archegos-type tail risk)?

14. What Must Be True

For the bull case (Goldman deserves ~3x TBV / sustains today’s price):

• Through-cycle ROTE must be structurally ~18–20%, not the mid-teens management guides to — i.e., the durable-revenue + AWM + financing upgrade must have permanently lifted returns and dampened the historical cyclicality.
• The M&A/trading cycle must have multiple years of runway (sponsor dry powder fires, AI-infra financing super-cycle materializes), so 2025–26 is a plateau not a peak.
• The regulatory tailwind must persist, funding a continued 4–6%/year buyback.
• Private credit/financing must avoid a material loss through its first default cycle.
• Falsification test: if, in the next capital-markets soft patch or credit normalization, ROTE falls back to low-teens and the multiple de-rates below ~2.3x TBV, the bull thesis (durable ~20% returns) is falsified — 2025/26 was a cyclical tooth-top, and the 3x multiple was a peak-on-peak error.

For the bear case (Goldman is a peak-earnings, peak-multiple fade):

• 2025/Q1-26 must prove to be a cyclical high — advisory and trading revert, the credit benefit reverses to a charge, and ROTE settles back toward the mid-teens guide.
• The multiple must de-rate from the 98th percentile toward a historically normal ~1.5–2.3x TBV.
• Falsification test: if ROTE holds ~18%+ through a banking/trading down-quarter and a credit normalization — demonstrating the cyclicality has genuinely been tamed — the bear thesis is falsified, the quality re-rating is validated as structural, and ~3x TBV is defensible.

15. Source Appendix

(Consolidated in the full Source Appendix below. Primary sources: Goldman Sachs FY2025 Form 10-K (filed 2026-02-25); EDGAR XBRL (RevenuesNetOfInterestExpense, NetIncomeLoss, StockholdersEquity, EarningsPerShareDiluted, CommonStockDividendsPerShareDeclared, Assets, Deposits, CommonEquityTierOneCapital, EntityCommonStockSharesOutstanding); Goldman Sachs earnings-call and conference transcripts (Q1 2026 Apr-13-2026, Q4 2025 Jan-15-2026, Q3 2025 Oct-15-2025, Bernstein May-28-2026, Apr-29-2026 shareholder call); public market data (2026-06-10).)

APPENDIX A — Standard Diligence Questionnaire

The Goldman Sachs Group, Inc. (NYSE: GS) — Report date 2026-06-10

Supplemental to the note. Fact / Interpretation / Assumption labels applied where material. Where a question does not map to an investment-bank business model, the correct sector analog is given.

General

What thoughtful questions have other investors asked about this company? The central debate is whether Goldman has structurally re-rated to a durable ~20% ROTE business or is simply printing peak cyclical earnings at a peak multiple. Sophisticated investors probe: (1) the true through-cycle ROTE (16% guide vs ~20% implied by 3x TBV); (2) how much of the “durable” $11.4B financing revenue survives a downturn; (3) whether AWM can actually reach $750B fee-paying alternatives against entrenched Blackstone/Apollo/Blue Owl; (4) the credit quality of the fast-growing, untested private-credit book; (5) whether the regulatory-capital tailwind (the engine of the buyback) persists; and (6) Solomon-era execution and succession after the consumer-strategy reversal.

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? High (Interpretation, well-supported). FY2025 ($58.28B revenue, $51.32 EPS, 15% ROE) and Q1 2026 (19.8% ROE / 21.3% ROTE) are records or near-records. Advisory revenue rose 89% YoY in Q1 2026; equities trading is at records; and 2025 booked a net credit-loss benefit of $1.11B rather than a charge. This is a multi-engine cyclical peak.

Driven by the external environment or internal actions? Both. External: the M&A/IPO/trading up-cycle and benign credit. Internal: the consumer-business exit (removing a drag), the durable-revenue build, AWM inflows, and an aggressive buyback. The quality improvement is internal and real; the magnitude of 2025–26 earnings is substantially external/cyclical.

How stable are revenues? The least stable of the megabanks. ~50% of revenue is advisory + trading, which swings 40–60% peak-to-trough (net revenue ran $59.3B → $46.3B → $58.3B across 2021–2025; EPS $30 → $23 → $51). The deliberate growth of “durable” revenue (fees + financing, ~$11.4B financing alone, 37% of FICC+equities) is raising the floor but has not been stress-tested through a downturn. Beta ≈ 1.31.

Outlook for products/services? Management is calling a multi-year M&A up-cycle (backlog at 4-year highs, ~$1T sponsor dry powder, AI-infrastructure financing super-cycle) and high-teens AWM returns. Assumption (management) — treat as hypothesis.

How big will this market be — growing, shrinking, domestic or international? Global. Investment-banking and trading fee pools are large and cyclically growing; the secular tailwind is in alternatives (private credit/equity/infra) and UHNW wealth, where Goldman is positioning AWM. The US bulge bracket has gained share as European banks retreated.

Business Quality & Competitive Moat

Is the industry getting more or less competitive? Mixed. Bulge-bracket IB/trading has consolidated (fewer credible competitors) — favorable. But alternatives and private credit are intensifying as a friendlier regulatory regime attracts capital back into financing/origination (Marathon capital-cycle warning) — a competitive headwind to the returns being extrapolated.

How profitable is the business (ROIC, ROE)? ROE 15.0% / ROTE 16.0% FY2025; 19.8% / 21.3% Q1 2026; management’s through-cycle target is mid-teens ROE. ROIC is not the right metric for a bank; ROE/ROTE are. Fact.

How profitable is the industry — how many competitors, what barriers to entry? At the mega-deal level, a high-barrier oligopoly (Goldman, Morgan Stanley, JPMorgan, BofA + elite boutiques) — barriers are relationships, reputation, league-table credibility, balance sheet, and talent. Trading is a scale/technology oligopoly. Both are genuinely hard to enter; AWM/alternatives is more contestable.

Can the business be easily understood? Partially. The franchise is conceptually simple (advise, trade, manage money) but the balance sheet, trading book, capital ratios, and private-credit exposures are complex and opaque relative to a retail bank. Interpretation.

Can it be undermined by foreign low-cost labor? No — this is a high-end, relationship- and talent-driven business, not a cost-arbitrage one.

Do brands matter? Yes, decisively. The Goldman name signals quality on a deal and is itself a competitive asset — a core intangible moat.

What is the nature of competition? Reputation, talent, balance sheet, league-table position, and increasingly technology/AI and financing capacity (the Capital Solutions cross-sell).

Customers’ switching costs? High in advisory (reputational/relationship), high in prime brokerage/financing (operational, multi-year), lower in commoditized trading and public-market asset management.

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet? The franchise/relationship intangibles and the AWM fee annuity are economically valuable but not on the balance sheet. Tangible book ($335.49/share) excludes ~$22/share of goodwill/intangibles.

Off-balance-sheet liabilities? Standard for a G-SIB: derivatives, guarantees, lending commitments, consolidated/unconsolidated VIEs, and prime-brokerage exposures — disclosed in the 10-K. The most under-appreciated contingent exposure is the private-credit/financing book in a default cycle. Open Question.

How conservative is the accounting? Generally conservative and heavily regulated, but note 2025’s earnings were flattered by a non-repeatable $1.11B credit-loss benefit (Apple Card reserve release) — a normalization headwind. Fair-value/Level-3 marks introduce judgment.

How CapEx-hungry is the business? Not CapEx-hungry in the industrial sense; it is capital-hungry in the regulatory sense (G-SIB CET1, liquidity, stress buffers). Goldman is currently releasing capital (CET1 14.3% → 12.5% via buybacks) as the regime eases. Technology/AI is the main discretionary spend.

Capital Allocation & Management

How much free cash flow does the business generate, and how is it used? FCF is not a meaningful bank metric; the analog is distributable capital. Goldman returned $16.78B to common shareholders in 2025 ($12.36B buybacks + $4.42B dividends) and a record $5B buyback in Q1 2026. Philosophy: reinvest in GBM financing / AWM, then return the rest — accelerated by the regulatory-capital release.

Significant acquisitions recently? No large acquisitions recently; the recent story is divestiture — exiting consumer (GreenSky sold, GM card transitioned Aug-2025, Apple Card held-for-sale Jan-2026). The 2021 GreenSky purchase (~$2.2B) was a capital-allocation error, since reversed.

Buying back shares? Aggressively — share count 315.8M (Jul-2024) → 295.0M (Apr-2026), ~6.6%. Caveat: executed at ~3x tangible book (because capital is available, not because the stock is cheap).

Issuing large amounts of new shares to insiders? No unusual dilution; equity comp is routine and the net share count is falling sharply. Comp ratio ~31.8% of net revenue.

Compensation policy of directors/management? Senior comp is tied to ROE/ROTE, book-value growth, and relative TSR; comp ratio is high in absolute terms (a talent business) but trending down. Fact / Interpretation.

Motivations of management? Mixed record: the 2016–2021 consumer push destroyed capital, but the current team has decisively corrected it and is executing a disciplined, shareholder-friendly capital-return program. Net: improving alignment with an asterisk.

Valuation & Market Data

Is the stock an ADR, MLP, or K-1 issuer? No — a US domestic C-corp common stock (NYSE: GS), standard 1099 treatment.

Dividend policy? Growing: $9.00 (2022) → $18/share forward run-rate ($4.50/quarter, raised Q4-2025), ~22–25% payout, ~1.8% yield.

How profitable is the business? ROE 15% / ROTE 16% (FY2025), peaking ~20%+ in Q1 2026; net margin ~29%. Good, but running above the mid-teens through-cycle target.

Is net income diverging from cash from operations? Operating cash flow is not a clean signal for a trading bank (it swings with balance-sheet/inventory movements). The cleaner quality flag is that 2025 net income was boosted by a non-cash, non-repeatable credit-loss benefit.

Risks & Downside

What factors would cause the stock to decline? A capital-markets cycle rollover (M&A/IPO/trading fade); credit normalization off the 2025 benefit; a private-credit loss event; multiple compression from the 98th-percentile valuation; or a partial reversal of the regulatory-capital easing that throttles the buyback.

Risk of a catastrophic loss? Low probability of total loss (G-SIB capital/liquidity buffers, regulatory backstop), but Goldman carries genuinely higher tail risk than a deposit-funded bank — a large trading book, prime-brokerage counterparty exposure (Archegos-type), and an untested private-credit book. The realistic severe downside is a cyclical earnings halving plus de-rate (as in 2022, ~$420 → ~$290), not insolvency.

Chance of a total loss? Very low, absent a systemic financial crisis combined with a Goldman-specific control failure.

Recent News & Events

Has the business environment changed recently? Yes, favorably on two axes: (1) a friendlier regulatory regime (a “neutral” Basel III endgame, a G-SIB surcharge reproposal, a 320bp stress-capital-buffer improvement) freeing capital; (2) a recovering capital-markets cycle (record advisory/trading). Counter-currents: NIM compression in AWM private banking (“persisting for much of 2026”), rising private-credit scrutiny, and a more cautious macro tone (geopolitics, energy, AI software disruption) in Q1 2026.

Significant acquisitions? No — the action is divestiture (consumer exit) and the AWM alternatives build (organic + small tuck-ins; Innovator ETF added ~$31B AUS in Q1-2026).

Change in accounting policies? None material flagged; note the Apple Card reclassification to held-for-sale drove the 2025 reserve release.

Recent changes — new markets, facilities, management? The Capital Solutions Group (2024) and “One Goldman Sachs 3.0” (firmwide AI operating model) are the notable strategic/operational initiatives; the consumer franchise is in runoff.

APPENDIX B — Source Appendix

The Goldman Sachs Group, Inc. (NYSE: GS) — Report date 2026-06-10

Primary sources prioritized. All quantitative figures reconciled to SEC filings / EDGAR XBRL; management commentary treated as hypothesis and validated against filings and external data where possible.

Primary — SEC filings (EDGAR, CIK 0000886982)

1. Goldman Sachs FY2025 Form 10-K — filed 2026-02-25 (gs-20251231.htm). Source of: net revenue $58.28B; net income $17.18B; diluted EPS $51.32; ROE 15.0% / ROTE 16.0%; book value/common share $357.60; tangible book $335.49; efficiency ratio 64.4%; CET1 14.3% Standardized / 15.1% Advanced; segment net revenue (GBM $41.45B, AWM $16.68B, Platform Solutions ~$2.1B); IB fees $9.34B; FICC market-making $7.24B; equities market-making $10.77B + commissions $5.03B; total AUS $3.606T; AUS net inflows $224B; provision a net benefit of $1.11B; capital returned $16.78B ($12.36B buybacks + $4.42B dividends); total assets $1.81T; deposits $501B.
2. EDGAR XBRL company facts (data.sec.gov) — multi-year series reconciled: RevenuesNetOfInterestExpense (2022 $47.37B; 2023 $46.25B; 2024 $53.51B; 2025 $58.28B); NetIncomeLoss (2022 $11.26B; 2023 $8.52B; 2024 $14.28B; 2025 $17.18B); EarningsPerShareDiluted (2022 $30.06; 2023 $22.87; 2024 $40.54; 2025 $51.32); StockholdersEquity (2023 $116.9B; 2024 $122.0B; 2025 $125.0B); CommonStockDividendsPerShareDeclared (2022 $9.00; 2023 $10.50; 2024 $11.50; 2025 $14.00); Assets; Deposits; CommonEquityTierOneCapital; EntityCommonStockSharesOutstanding (315.8M Jul-2024 → 295.0M Apr-2026).
3. Form 3/4 corpus (2025–2026) — reviewed; insider activity in Apr–May 2026 is the routine annual equity-award settlement cluster (sell-to-cover / grants), no signal-bearing open-market purchases.
4. 8-K / earnings-release corpus — quarterly results, dividend increases, capital actions (mirrored to output/GS/sources/).

Primary — Earnings & event transcripts

5. Q1 2026 Earnings Call — Apr 13, 2026. ROE 19.8% / ROTE 21.3%; record $5B buyback; CET1 12.5%; management/other fees +14% to $3.1B; advisory +89% YoY; #1 global M&A; private-credit defense; $315M provision build.
6. Q4 2025 Earnings Call — Jan 15, 2026. AWM targets raised (30% pretax margin, high-teens return, $750B fee-paying alternatives by 2030, $1B incentive fees); dividend raised to $4.50/qtr; backlog at 4-year high; Apple Card held-for-sale (+$0.46 EPS); Basel III “neutral” framing.
7. Q3 2025 Earnings Call — Oct 15, 2025.
8. Bernstein 42nd Annual Strategic Decisions Conference — May 28, 2026 (Waldron): “winner-take-most”; ~$300B league-table lead; corporate M&A +62% YoY vs PE −4%; Capital Solutions cross-sell ~140% of advisory; AI-infra financing ~$700–800B.
9. Apr 29, 2026 Shareholder/Analyst Call.

Secondary / market data

10. Market-data aggregator (own-history valuation percentiles) (2026-06-09): market cap, P/E 16.9–17.9x, P/B, dividend, analyst distribution (~5 strong-buy / 3 buy / 14 hold / 1 sell; mean target ~$935), short interest (~2.3% of float), and the own-history valuation percentiles (P/E 97th, P/B 99.7th, P/S 99.7th; composite 98.8th). Reconciled to filings; the analyst consensus target is not used as a price target.
11. Public market quotes (2026-06-10): GS ~$1,001, market cap ~$295–299B, 52-week range $609.59–$1,098.36; peer quotes JPM ~$309 / MS ~$206.66 (P/E 18.7x, ROE 16.4%).

Frameworks

12. Analytical frameworks — Greenwald (Competition Demystified) moat taxonomy and Chancellor/Marathon (Capital Returns) capital-cycle lens.

Note on reconciliation: all income-statement, balance-sheet and capital figures are sourced from the FY2025 10-K and EDGAR XBRL. Third-party aggregator data was used only for market prices and own-history valuation percentiles, and reconciled to filings.