GS 10-K Analysis: 73% of Growth Came From Interest, Not Deal-Making

Goldman Sachs, the world's preeminent investment bank, reported record FY2025 earnings of $51.32 per share — up 26.6% — on revenue of $58.3 billion. Wall Street credits the M&A super-cycle. But a line-by-line read of the 10-K reveals that 73% of the flagship Global Banking & Markets segment's $6.4 billion revenue increase came from net interest income, not deal-making fees. Core franchise fee revenue grew just 5.7%.

The FY2025 10-K, filed with the SEC on February 25, 2026, tells a story the earnings release cannot. The headline numbers — $58.28 billion in net revenues, 15.0% return on equity, $16.78 billion returned to shareholders — paint a picture of a firing-on-all-cylinders financial institution. Three operating segments delivered: Global Banking & Markets (GBM) earned $41.5 billion in revenue, Asset & Wealth Management (AWM) contributed $16.7 billion, and the winding-down Platform Solutions posted $0.15 billion after the Apple Card exit. Management hit the midpoint of its 14-16% ROE target and accelerated its buyback program, retiring 4.8% of diluted shares.

But beneath those numbers, the filing reveals three dynamics that fundamentally reframe what investors are paying for. First, net interest income — not deal-making fees — was the dominant driver of earnings growth, surging 68.3% to $13.56 billion and now representing 23.3% of total revenue. Second, the wealth management diversification strategy is generating negative operating leverage, with AWM pre-tax earnings falling 15.2% despite revenue growing 2.2%. Third, Goldman returned 103% of net income to common shareholders while simultaneously issuing $43.6 billion in new debt — a capital return posture that works only if the rate-driven earnings tailwind proves durable.

At $879 per share and 2.21x tangible book value, the market is pricing Goldman Sachs as a deal-making franchise earning durable mid-teens returns. The filing describes something different: a rate-sensitive balance sheet play with a compensation problem and a wealth management cost structure that gets worse, not better, at scale.

The $5.5 Billion Mirage — NII Hijacked the Deal-Making Narrative

The market narrative that Goldman Sachs is riding a deal-making super-cycle is mathematically incomplete. When you decompose the flagship GBM segment's $6.39 billion revenue increase, the composition is striking: $4.66 billion — 73% of the total — came from net interest income surging 182%, from $2.56 billion to $7.22 billion. The remaining $1.73 billion came from non-interest revenue, which grew just 5.3%.

This decomposition matters because the market assigns different durability to different revenue streams. NII is balance-sheet-driven and rate-sensitive — it scales with the rate environment, not with deal activity. Fee revenue from investment banking advisory, underwriting, commissions, and market making is franchise-driven — it reflects client relationships, market share, and transaction volumes that persist through rate cycles. When investors buy GS "for the M&A cycle," they are implicitly making a statement about fee growth. The filing shows that core fee revenue — the sum of investment banking ($9.35B, +20.8%), investment management ($11.75B, +10.9%), commissions ($4.04B, -1.1%), and market making ($17.99B, -2.2%) — grew from $40.81 billion to $43.13 billion, an increase of just 5.7%.

Investment banking fees did grow a strong 20.8%, confirming the deal cycle is real. But IB fees are only one component of the revenue base, and their $1.61 billion increase was dwarfed by the $4.66 billion NII surge. Meanwhile, other principal transactions — mark-to-market gains on proprietary investments within GBM, distinct from the Apple Card exit in Platform Solutions — collapsed 65.7%, from $4.65 billion to $1.59 billion. This $3.05 billion decline was almost entirely offset by NII, creating an illusion of steady diversified growth.

"Net revenues were $58.28 billion for 2025, 9% higher than 2024, reflecting higher net revenues in Global Banking & Markets, partially offset by significantly lower net revenues in Platform Solutions."

The investment implication is direct: investors who believe they are buying GS for deal-making exposure are actually buying rate exposure. Goldman Sachs' Global Banking & Markets segment reported $41.5 billion in FY2025 revenue, but 73% of the $6.4 billion year-over-year increase came from net interest income surging 182% — not from the deal-making fees that define the firm's brand. If rates fall and NII compresses, GBM's revenue growth evaporates regardless of the M&A pipeline.

The Compensation Paradox — AWM Costs Devour the Efficiency Promise

Goldman's strategic pivot toward wealth management — the "OneGS 3.0" initiative designed to build a stable, fee-driven earnings base akin to Morgan Stanley's franchise — is running directly into a compensation wall. Asset & Wealth Management revenue grew 2.2% to $16.68 billion in FY2025. Pre-tax earnings fell 15.2% to $4.13 billion. The pre-tax margin compressed 510 basis points, from 29.8% to 24.7%. The culprit is straightforward: compensation grew 13.2% while revenue barely moved, expanding the AWM compensation ratio by 440 basis points to 44.6%.

The internal divergence is dramatic. GBM operates at 56.7% efficiency with a 26.6% compensation ratio — a capital markets machine that prints profit. AWM runs at 75.9% efficiency with a 44.6% comp ratio — a wealth management business that consumes most of the revenue it generates. The firm-wide 64.4% efficiency ratio is a blend of a highly efficient trading operation and a structurally expensive advisory business. AWM's 440-basis-point comp expansion was the primary drag on firm-wide efficiency, which worsened 130 basis points year-over-year despite record revenue. The Q4 efficiency ratio derived from quarterly filing data spiked to approximately 72.3% — well above the FY2025 annual average of 64.4%.

The compensation problem extends beyond AWM. Stock-based compensation surged 30.7% firm-wide to $3,624 million — 6.22% of total revenue, growing 3.5 times faster than the top line. This ratio is 2.3 times Morgan Stanley's 2.73%. On a smaller revenue base ($58.3B vs MS's $70.6B), Goldman pays nearly twice as much in stock comp ($3.62B vs MS's ~$1.93B). The dilutive effect is measurable: approximately 4.9 million RSU shares offset 31% of the 16.0 million shares retired through the buyback program.

"Operating expenses were $37.54 billion for 2025, 11% higher than 2024, primarily reflecting higher compensation and benefits expenses (reflecting improved operating performance) and higher transaction based expenses."

Management attributes the compensation growth to "improved operating performance" — the logic being that higher revenue triggers higher bonus accruals. But this is precisely the operating leverage problem: in a year where revenue grew 8.9%, compensation grew 13.2% and SBC grew 30.7%. Goldman Sachs' Asset & Wealth Management segment expanded its compensation ratio by 440 basis points to 44.6% in FY2025, driving a 15.2% decline in pre-tax earnings despite 2.2% revenue growth — the opposite of the operating leverage that management's OneGS 3.0 initiative was supposed to deliver. AWM's segment ROE fell to 12.5%, below management's own 14% target floor.

"Share-based compensation was $3,624 million in 2025, up from $2,772 million in 2024."

Capital Return at Cycle Peak — The $16.8 Billion Question

Goldman Sachs returned $16.78 billion to common shareholders in FY2025 — $12.36 billion in share repurchases and $4.42 billion in common dividends. That $16.78 billion represents 103% of net income attributable to common shareholders, meaning Goldman distributed more than it earned and funded the gap with new debt. The buyback retired 4.8% of diluted shares, accelerating from a 3.7% pace in 2024 — the most aggressive program among mega-cap financial institutions.

"During 2025, we returned a total of $16.78 billion of capital to common shareholders, including $12.36 billion of common share repurchases and $4.42 billion of common stock dividends."

The capital return confidence is underpinned by a 14.3% CET1 ratio — roughly 430 basis points above regulatory minimums — and the record earnings base. But the return intensity coincides with a balance sheet that grew more leveraged, not less. Total borrowings reached $355.96 billion, up 14.0% from $312.34 billion — growing nearly twice as fast as total assets (+7.9%). Goldman issued $43.6 billion in net new debt to fund balance sheet expansion, with $70.46 billion (19.8% of the total) maturing within twelve months.

The buyback mathematics depend entirely on durability. At 2.21x tangible book, Goldman is repurchasing shares at a significant premium to liquidation value. If ROE sustains at 15% and ROTE at 16%, buying at 2.21x TBV is accretive — the company earns more than the cost of equity on the incremental tangible book acquired by each retired share. But if the NII tailwind reverses and ROE drops to 11% or below, every dollar of buyback at $879 per share destroys value relative to what Goldman could have earned deploying that capital elsewhere. The share count acceleration — diluted shares declining from 345.8 million (2023) to 333.6 million (2024) to 317.6 million (2025) — amplifies per-share results in both directions.

The SBC dilution partially erodes the buyback. Approximately 4.9 million dilutive RSU shares offset 31% of the 16.0 million share buyback, leaving a net share reduction of roughly 11.1 million. This means EPS was boosted approximately $2.50 beyond organic growth — a meaningful contributor to the headline 26.6% increase, but one that requires capital expenditure (the buyback dollars) to sustain. Goldman Sachs returned $16.78 billion to common shareholders in FY2025 — 103% of net income — while simultaneously issuing $43.6 billion in new debt, creating a capital return posture that works at 15% ROE but destroys value if the rate-driven earnings tailwind reverses.

The Rate Reckoning — What Breaks When NII Normalizes

The thread connecting every finding in this analysis is a single number: $13.56 billion. That is Goldman Sachs' FY2025 net interest income — up 68.3% from $8.06 billion in FY2024 — now representing 23.3% of total revenue, up from 15.1% a year earlier. NII is not a supporting player in Goldman's earnings story. It is the earnings story. The $5.50 billion NII improvement was larger than any other single revenue driver, larger than the IB fee increase ($1.61B), and larger than the net income improvement itself ($2.78B to common).

The filing contains a disclosure that makes this dependence especially precarious. In the debt footnotes, Goldman states that while the "vast majority of unsecured long-term borrowings consists of fixed-rate obligations," after hedging, "the vast majority consists of floating-rate obligations." This means Goldman has used interest rate swaps to convert its $356 billion debt book from fixed to floating — creating enormous positive carry when rates are high (the asset side earns more while the swapped-to-floating funding cost adjusts) but symmetric downside when rates fall. The firm is synthetically positioned to benefit from high rates across its entire balance sheet.

"The vast majority of unsecured long-term borrowings consists of fixed-rate obligations. However, after hedging, the vast majority consists of floating-rate obligations."

What does NII normalization mean for valuation? At $879 and 2.21x tangible book, the market is implicitly pricing Goldman to sustain approximately 18.5% ROTE through the cycle — derived from the residual income framework where P/TBV = (ROTE – g) / (CoE – g), assuming 10% cost of equity and 3% terminal growth. The filing shows actual ROTE of 16.0%, already a 250 basis point shortfall from what the multiple implies. The gap is the NII premium embedded in the stock.

The peer context sharpens the concern. Goldman's NII dependency (23.3% of revenue) sits between Morgan Stanley (14.2%) and Bank of America (54.7%). GS is converging toward BAC's NII-dependent profile while the market prices it as an investment banking franchise closer to MS. Morgan Stanley commands a 43% P/TBV premium (3.16x vs 2.21x) on identical ROE (15.0%) — justified by higher ROTCE (19.1% vs 16.0%), lower SBC dilution (2.73% vs 6.22% of revenue), a high-margin wealth management franchise, and critically, lower NII dependency. The market is right to pay more for MS. The question is whether GS deserves its current premium given the NII fragility the filing reveals.

"Our target (through-the-cycle) is to achieve ROE within a range of 14% to 16% and ROTE within a range of 15% to 17%."

Management's "through-the-cycle" qualifier does heavy lifting. In FY2023, Goldman's ROE was 7.5% — well below the 14% floor. The FY2025 15.0% result was achieved with a $5.50 billion NII tailwind that management cannot control. The target range implies durability. The filing reveals dependence on conditions that are, by definition, cyclical. At $879 and 2.21x tangible book, Goldman Sachs' valuation implies an 18.5% sustainable ROTE — but the filing reveals 16.0% actual ROTE, with $5.5 billion of the earnings base depending on a net interest income tailwind that reverses if rates fall.

What to Watch: Five Metrics That Will Resolve the Thesis

The NII-driven earnings thesis is testable. These five metrics will determine whether Goldman's record results represent a structural repositioning or a cyclical peak:

1. Quarterly NII trajectory (Q1-Q2 2026). FY2025 quarterly NII averaged approximately $3.39 billion. If Q1 2026 NII exceeds $3.5 billion despite any rate cuts, the balance sheet has been repositioned to generate NII independent of the rate cycle — thesis weakened. If NII drops below $3.0 billion, the rate sensitivity is confirmed and FY2026 NII could fall to the $10-11 billion range.

2. AWM compensation ratio (next two quarters). The 44.6% comp ratio expansion (+440bps) was the single largest drag on firm-wide efficiency. If the comp ratio stabilizes below 42% in Q1-Q2 2026, retention spending was transient and AWM margin recovery toward 28-30% pre-tax is possible. If it holds above 45%, the cost problem is structural and the wealth management pivot remains a financial drag.

3. Core fee revenue growth rate. The 5.7% FY2025 growth rate strips out NII and one-time items to reveal the underlying franchise trajectory. Acceleration above 10% in FY2026 would prove the deal-making franchise — not NII — is the dominant growth engine. Deceleration below 5% would confirm the franchise is mature and the earnings record was NII-manufactured.

4. Share count direction net of SBC dilution. With buybacks offsetting only 69% of SBC dilution on a net basis, watch whether net share reduction accelerates or decelerates. If SBC dilution increases further (above 5M dilutive shares), the per-share growth story weakens even if revenue grows.

5. Total debt trajectory. The 14.0% debt growth rate — nearly double asset growth — intensified leverage. If borrowings continue growing faster than assets in FY2026, the balance sheet is becoming more fragile, not less. Combined with the floating-rate hedging posture, rising debt load amplifies NII sensitivity in both directions.

At $879, the market prices Goldman Sachs as a deal-making franchise earning 16% ROTE with a clear path to sustained mid-teens returns. The filing reveals an earnings engine that runs primarily on interest carry, a wealth management pivot that costs more than it earns at the margin, and a compensation structure that dilutes a third of the buyback. The bull case requires sustained NII above $12 billion, AWM comp ratio compression below 42%, and core fee growth acceleration — all of which depend on conditions Goldman cannot unilaterally control. The bear case at full NII normalization ($494, 1.24x TBV) represents a 44% decline; even partial normalization ($626, 1.57x TBV) implies 29% downside. The filing tells you which bet you are making. The question is whether you knew it before you read the 10-K.

Frequently Asked Questions

What drove Goldman Sachs' record FY2025 earnings?

Goldman Sachs reported $51.32 diluted EPS in FY2025, up 26.6% from $40.54 in FY2024. The headline narrative credits the M&A super-cycle, but the 10-K reveals the single largest driver was net interest income (NII), which grew 68.3% to $13.56 billion — now 23.3% of total revenue. Within the flagship GBM segment, NII surged 182% and accounted for 73% of the segment's $6.4 billion revenue increase. Core fee revenue (IB, investment management, commissions, market making) grew just 5.7%. The remaining EPS growth came from a 4.8% share count reduction via $12.36 billion in buybacks.

How rate-sensitive is Goldman Sachs' earnings base?

Very. The filing discloses that the vast majority of unsecured long-term borrowings are floating-rate obligations after hedging, despite being issued as fixed-rate bonds. FY2025 NII of $13.56 billion was $5.50 billion higher than FY2024. If NII normalizes to FY2024 levels ($8.06 billion), ROTE would drop from 16.0% to approximately 11.7% — well below management's 14-16% ROE target range.

How does Goldman Sachs' efficiency compare to peers?

Goldman Sachs posted a 64.4% FY2025 efficiency ratio, matching Bank of America (64.4%) but better than Morgan Stanley (68.4%). However, the trend worsened 130 basis points year-over-year, and Q4 2025 spiked to 72.3%. Internally, GBM operates at 56.7% efficiency while AWM runs at 75.9% — the wealth management segment's 440-basis-point comp ratio expansion was the primary drag.

How does Goldman Sachs compare to Morgan Stanley?

GS and MS share identical ROE (15.0%) but MS commands a 43% P/TBV premium (3.16x vs 2.21x). MS achieves higher ROTCE (19.1% vs 16.0%), lower SBC dilution (2.73% vs 6.22% of revenue), a high-margin wealth management franchise (comp ratio ~36% vs GS AWM 44.6%), and lower NII dependency (14.2% vs 23.3% of revenue). MS has successfully diversified into wealth management as a profit center; GS's AWM attempt is currently a cost center.

Is Goldman Sachs' stock overvalued at 2.21x tangible book?

The 2.21x P/TBV multiple implies the market expects Goldman to sustain approximately 18.5% ROTE (assuming 10% cost of equity, 3% growth). Actual FY2025 ROTE was 16.0% — already a 250bps shortfall. If NII normalizes to FY2024 levels, ROTE drops to ~11.7% and justified P/TBV compresses to approximately 1.24x — implying ~$494, a 44% decline. Even partial NII normalization to ~$11B implies 1.57x P/TBV ($626, -29%).

What happened with the Apple Card exit?

Goldman Sachs completed its Apple Card exit in FY2025. The impact: a $2.26 billion revenue markdown offset by a $2.48 billion reserve release. Net pre-tax: +$220 million — largely accounting noise. Platform Solutions swung from -$997 million to +$151 million pre-tax, but the underlying result excluding one-time items was still -$69 million. The segment is 0.3% of revenue and immaterial going forward.

How aggressive is Goldman Sachs' capital return program?

Goldman returned $16.78 billion to common shareholders in FY2025 — 103% of net income to common. Buybacks of $12.36 billion retired 4.8% of diluted shares. However, SBC dilution (4.9 million RSU shares) offset 31% of the buyback. Net share reduction was approximately 11.1 million. The capital return is enabled by a 14.3% CET1 ratio but coincides with $43.6 billion in net new debt issuance.

What is Goldman's debt and leverage situation?

Goldman Sachs operates with $1.81 trillion in total assets on $125 billion in equity — a 14.5x leverage ratio. Total borrowings reached $355.96 billion, up 14.0% — growing nearly 2x faster than assets (+7.9%). Of this, $70.46 billion (19.8%) matures within 12 months. After hedging, the vast majority is synthetically floating-rate, creating NII sensitivity to rate changes. The CET1 ratio of 14.3% provides a buffer above regulatory minimums.

What are the key risks for Goldman Sachs in 2026?

Three primary risks: (1) Rate reversal — if the Fed cuts significantly, NII compresses from $13.56 billion toward $8-10 billion, potentially eliminating the dominant FY2025 earnings driver; (2) M&A deceleration — management cites geopolitical risks, tariffs, and trade policy uncertainty without positive forward catalysts; (3) Compensation cost spiral — AWM comp ratio expansion (+440bps) and SBC growth (+30.7%) at multiples of revenue growth worsen efficiency regardless of revenue trajectory.

What is the OneGS 3.0 efficiency initiative?

OneGS 3.0 is Goldman's internal efficiency improvement program. However, the 10-K provides no quantified savings targets or benchmarks. Every efficiency metric worsened in FY2025: firm-wide efficiency ratio increased 130bps to 64.4%, Q4 hit 72.3%, compensation grew 13.2% vs revenue +8.9%, and SBC grew 30.7%. Management's framework is ROE-based (14-16% target), not efficiency-based.

How much of Goldman's EPS growth came from buybacks vs. organic growth?

Of the $10.78 EPS increase ($40.54 to $51.32), the NII windfall contributed approximately $13.60/share, core fee growth contributed approximately $5/share, and the 4.8% share count reduction added approximately $2.50/share. Without the NII windfall, EPS would have been approximately $38 — a decline from FY2024's $40.54. The headline 26.6% EPS growth substantially overstates the franchise's underlying earnings power improvement.

What is Goldman Sachs' asset and wealth management strategy, and is it working?

Goldman's strategy to build a large-scale wealth management franchise mirrors Morgan Stanley's successful pivot, but is generating negative returns. AWM FY2025 results: revenue +2.2%, pre-tax earnings -15.2%, pre-tax margin compressed 510bps to 24.7%, comp ratio expanded 440bps to 44.6%, and segment ROE fell to 12.5%. MS's wealth management franchise operates at roughly 36% comp ratio with ROTCE above 20%. Unless GS reduces AWM comp toward 40% and grows fee revenue faster than costs, the pivot remains a financial drag.

Methodology

Data Sources

• Primary: Goldman Sachs FY2025 10-K filing (filed 2026-02-25 with the SEC). All figures tagged [FILING] sourced from MD&A, segment footnotes, debt footnotes, and stock compensation footnotes via MetricDuck's filing text extraction pipeline.
• Secondary: MetricDuck metrics pipeline for XBRL-sourced financial data. Pipeline data was cross-verified against filing text; critical corrections applied for NII, provision, and SBC figures where pipeline XBRL concepts diverged from filing presentation.
• Peer data: Morgan Stanley, Bank of America, and Royal Bank of Canada data from MetricDuck pipeline extractions. Peer comparisons use trailing twelve-month or latest fiscal year data available as of March 2026. Peer metrics are directionally correct but have not undergone the same filing-level verification applied to Goldman Sachs.
• Derived calculations: All derived figures computed from filing-sourced inputs with formulas documented inline. The NII normalization scenarios use filing-sourced FY2024/FY2025 NII levels and a simplified residual income framework — not fabricated sensitivity figures.

Limitations

1. NII sensitivity is directional, not precisely quantified. Goldman Sachs' 10-K does not include a quantitative NII sensitivity table showing the per-basis-point impact of rate changes. The normalization scenarios use filing-sourced NII levels ($8.06B FY2024, $13.56B FY2025) as anchors, not modeled rate sensitivities. Actual NII behavior depends on hedging portfolio duration, notional amounts, and repricing schedules that are not fully disclosed.

2. Peer data is pipeline-sourced, not filing-verified. Morgan Stanley, Bank of America, and Royal Bank of Canada metrics are from MetricDuck's XBRL pipeline and may contain the same concept mismatches identified in Goldman's pipeline data. Peer comparisons should be treated as directionally correct.

3. Q4 efficiency ratio is derived, not directly disclosed. The 72.3% Q4 figure is calculated from quarterly revenue and expense data in the filing. Goldman does not disclose a quarterly efficiency ratio as a standalone metric.

4. No earnings call data. Forward guidance, management tone, and IB backlog commentary are not available from the 10-K alone. The filing contains no forward revenue or margin guidance.

5. Valuation framework is simplified. The residual income model (P/TBV = (ROTE – g) / (CoE – g)) assumes stable growth, constant cost of equity, and steady-state returns. Banks' actual valuations reflect franchise optionality, regulatory capital cycles, and market sentiment that this model does not capture. The normalized price scenarios should be interpreted as directional stress tests, not price targets.

6. AWM AUM/AUS figure is approximate. The filing intelligence notes approximately $3 trillion in assets under supervision but no exact figure was extracted from the filing text sections, limiting the precision of AUM-based peer comparisons with Morgan Stanley (~$6T AUM).

Disclaimer:

This analysis is for informational purposes only and does not constitute investment advice. The author does not hold positions in GS, MS, or BAC. Past performance and current metrics do not guarantee future results. All data is derived from public SEC filings and may contain errors or omissions from the automated extraction process.