Goldman Sachs Group (GS) · Deep analysis

Report previewGS.USReport date Last updated
← Back to reports
Report contents
01

Business model

Business model: flow revenue first, fees second

Goldman Sachs sells advice, execution, financing and asset management to corporations, financial institutions, governments and wealthy individuals. It gets paid three ways: fees for completed transactions, spreads and commissions on the flow it intermediates, and net interest plus management fees on the assets and loans it holds or administers. The paying customer is institutional or high-net-worth, not retail, and the market is global, with revenue reported across the Americas, EMEA and Asia. That makes Goldman a financial intermediary, and a mature one: a cyclical franchise whose revenue follows capital-markets activity more closely than it follows its own expansion plans.

Three segments, two economics

In its FY2025 Form 10-K, filed February 25, 2026, Goldman reports three segments. Global Banking & Markets is the center of gravity, holding investment banking (advisory, equity and debt underwriting), FICC and Equities intermediation, and the lending and financing that support both. Asset & Wealth Management runs public and private market strategies for institutions and individuals, alongside private banking and lending. Platform Solutions carries the consumer and transaction-banking businesses; it is small next to the other two and affects firmwide returns mainly through capital and expenses.

The more useful split is activity revenue versus recurring revenue. GBM earns when a deal closes or a client repositions a hedge, so its top line depends on announcements, issuance windows and volatility. AWM earns a percentage of assets, which compounds with markets and net flows and is far less exposed to a single quarter's deal calendar. AWM is the smaller of the two revenue streams and the more repeatable one, so a sustained shift of revenue toward it would change how the whole firm should be valued. Whether that shift is happening is a question the segment lines can answer; the label on the segment cannot.

Position in the chain, and the limits of pricing power

Goldman stands between issuers who need capital and investors who supply it, and it intermediates risk between counterparties. On advisory it is largely a price taker: fees are negotiated deal by deal against other global banks and independent advisory firms, and a bank that insists on price loses the mandate. In market making, the franchise captures spread on client flow while the trading book absorbs inventory risk, so revenue depends on turnover and volatility more than on any price Goldman sets. What the firm owns is a relationship network, a balance sheet large enough to commit capital, and a brand that gets it into the room. Those win mandates; they do not set fees.

The consequence for growth is that firmwide revenue is largely a share-of-pool number. When industry M&A and issuance volumes are flat, Goldman grows only by taking mandates from rivals; when volumes rise, revenue rises even without share gains. That is why a single strong year says little about the franchise, and why the fee pool, not the firm's own revenue, is the right comparison.

Disclosure is reasonably granular at the segment and line-item level. It stops short of client-level concentration, which is appropriate because no single client is material to a firm this size, and it does not split market-making revenue into spread and inventory gains, so the quality of a strong trading quarter has to be inferred.

Compensation and capital set the return

Compensation and benefits is by far the largest expense, and the ratio of compensation to net revenues is the biggest swing factor in pre-tax margin. The pool is discretionary in the short run, which cushions a weak year, but the discretion has a limit: the people who originate the fees can move. Cutting the ratio to protect margin in a slow year is easy to announce and hard to repeat without losing producers, so a low ratio in a downturn is not by itself evidence of a better business.

Capital is the other constraint. The balance sheet is funded by deposits, secured financing and unsecured debt, and the firm must hold capital against it under Federal Reserve rules, including the stress capital buffer and the surcharge applied to global systemically important banks (G-SIBs, firms whose failure regulators judge would disrupt the financial system). Those rules, not management preference, set how fast the balance sheet can grow and how much capital can be returned, which ties buybacks to the cycle in a way that operating companies do not face.

What would change the view

The judgment here, that flow revenue dominates, fee revenue stabilizes and people costs decide the margin, would be overturned by a slow-deal year in which AWM management fees and GBM financing revenue both grow while the compensation ratio stays flat. That combination would show a more recurring revenue base than "flow business" implies. The opposite result, revenue falling with the compensation ratio falling faster, would indicate the margin rests on discretionary pay cuts, a less durable source of earnings.

Questions worth following

  • Can Goldman grow AWM fees without adding more balance sheet?

    Recurring asset-management fees are the most durable revenue in the firm, but AWM also lends through private banking, which consumes capital. If fee growth requires more loans and deposits, the mix improves less than headline segment growth suggests and returns stay tied to balance-sheet size.

  • Will Goldman's investment banking fees recover with deal volumes?

    Investment banking is a high-margin fee pool, but it depends on completed transactions rather than announced mandates. A durable recovery would lift revenue and absorb compensation; a share loss would leave Goldman with the same cost base and a smaller franchise.

  • Can Goldman hold its compensation ratio without losing senior bankers?

    Compensation is the largest cost, and its ratio to net revenues sets the pre-tax margin. Management can lower the ratio in a weak year, but repeated cuts risk the producers who generate the fees. Durable margins need the ratio to fall for structural reasons.

02

Financial quality

Financial quality: strong returns, more help from the balance sheet

Goldman's earnings are recurring and increasingly profitable; the open question is how much of the latest improvement comes from a larger balance sheet. Net revenues rose from $46.3 billion in 2023 to $58.3 billion in 2025, a 26% gain, while operating expenses moved only from $35.5 billion to $36.4 billion. That operating leverage took net income from $8.5 billion to $17.2 billion and ROE from 7.3% to 13.9%. The swing is cyclical as well as operational: 2021's 21.0% ROE and 36.4% net margin were a boom, 2023's 7.3% a trough, and 2025 sits between them. Financial quality: Strong, with capital adequacy the item that decides whether the return holds. Net revenues here are revenue less interest expense, the usual reading of a bank's top line, and the same calculation reproduces the $18.68 billion Goldman reported for the September 2026 quarter.

USD bn unless statedFY2021FY2022FY2023FY2024FY2025Q3 FY2026Assessment
Net revenues59.347.446.353.558.318.7Strong
Net income21.611.38.514.317.25.3Strong
Net margin (%)36.423.818.426.729.528.3Strong
ROE (%)21.09.97.312.013.94.3 (17.1 ann.)Strong
ROA (%)1.650.780.550.860.990.25 (1.0 ann.)Adequate
Total assets1,4641,4421,6421,6761,8092,182Adequate
Total equity109.9117.2116.9122.0125.0125.0Adequate
Assets/equity (x)13.312.314.013.714.517.5Adequate
Operating cash flow0.98.7-12.6-13.2-45.24.2Weak
Free cash flow-3.85.0-14.9-15.3-47.23.7Weak

The Q3 FY2026 column is a single quarter: flows are not comparable with the full years, and its ROE and ROA are period figures with annualized equivalents shown.

The September quarter extended the recovery. Goldman reported net revenues of $18.68 billion, up 23% year over year, net earnings to common of $5.12 billion, diluted EPS of $17.01 (up 39%), an annualized ROE of 18.7% and an efficiency ratio of 60.5%. The mix was narrow. Equities revenue rose 72%, investment banking fees 21%, FICC was flat, and Platform Solutions fell 72% on markdowns of the Apple Card portfolio, which had been moved to held for sale. A quarter carried by equities financing and advisory is a good quarter, not a diversified one, and equities financing is precisely the business that consumes balance sheet.

USD bn unless statedQ1 FY2025Q2 FY2025Q3 FY2025Q4 FY2025Q1 FY2026Q2 FY2026Q3 FY2026AssessmentTrend
Net revenues15.114.615.213.517.220.318.7StrongRising
Net income4.73.74.14.65.66.65.3StrongRising
Total assets1,7661,7851,8081,8092,0602,1282,182AdequateAccelerating
Total liabilities1,6421,6611,6761,6841,9372,0052,057AdequateRising
Net PP&E9.99.89.89.59.59.49.0AdequateFlat
Operating cash flow-37.25.72.7-16.3-31.96.14.2WeakVolatile
Assets/equity (x)14.214.414.614.516.817.317.5AdequateRising

The March 2026 quarter reports revenue net of interest expense, unlike the other columns, which is why it appears to halve; on a consistent basis revenue rose about 14% year over year in that quarter.

The return now leans on the balance sheet

Assets rose $373 billion, or 21%, in the first nine months of 2026 while total equity stayed at $125 billion, lifting assets to 17.5 times equity from 14.5 times at year-end. Short-term borrowings jumped to $487 billion from $311 billion and long-term debt to $364 billion from $296 billion; deposits reached $568 billion. The standardized CET1 ratio slipped to 12.4% from 12.9% during the quarter and the supplementary leverage ratio held at 4.3%, the constraint that governs a firm adding secured financing and trading assets. Both ROA (about 1.0% annualized) and leverage now feed the 18.7% quarterly ROE; only the first compounds without consuming capital.

Operating cash flow was negative $13.2 billion in 2024 and negative $45.2 billion in 2025, and free cash flow has been negative in four of the last five years. For a broker-dealer these lines mostly track trading assets and secured financing: 2025's outflow accompanied an 8% increase in assets, and the first nine months of 2026 totaled negative $21.6 billion after a $31.9 billion March-quarter outflow. That is a funding fact, not evidence that reported earnings are soft. Distributions are where the tension shows. Goldman returned $4.49 billion in the September quarter, $3 billion of it buybacks, against $5.29 billion of net income, roughly 85% of earnings, while the balance sheet was expanding. Capital spending is not the constraint at about $2.1 billion a year, near 4% of net revenues; net PP&E has fallen to $9.0 billion from $20.4 billion in 2021 because this business invests in people.

Provisions stayed small at $368 million, up from $339 million a year earlier, driven by impairments on wholesale loans as loans grew to $276 billion from $261 billion. Neither credit costs nor the consumer marks are large enough to change the grade, but they mean the 2026 revenue base contains items that will not repeat.

What to check
  • Composition of the $373 billion. Matched-book resale agreements and high-quality liquid assets carry little market risk; equities financing, where revenue grew 72%, carries more. The filings separate these lines, and the split decides how demanding 17.5x really is.
  • Apple Card marks and timing. The held-for-sale portfolio produced the Platform Solutions loss. Remaining carrying value and the sale timetable determine whether another charge lands in 2027.
  • Wholesale credit. A $368 million provision against a $276 billion loan book is small, but the charge concentrated in wholesale loans. Level 3 marks and private-credit exposure are where to test it.
  • Equities revenue durability. A 72% increase needs a peer comparison to separate a cyclical trading quarter from a share gain; the next two quarters' equities revenue will show whether the level holds.

Questions worth following

  • Can Goldman sustain a mid-teens ROE without further balance-sheet leverage?

    Returns have recovered since 2023 largely on revenue growth and flat expenses, but the 2026 leg arrived with assets up 21% and equity flat. If ROE holds while leverage stabilizes, fees and spreads are doing the work and the return is more durable; if not, capital-consuming growth is carrying it.

  • Will Goldman's Apple Card exit close without further write-downs?

    Platform Solutions revenue fell 72% on markdowns of a portfolio already moved to held for sale. Additional marks would hit revenue and capital directly and signal that carrying values still exceed what a buyer will pay, which affects how clean the 2027 earnings base looks.

  • Can Goldman keep buying back roughly $3 billion a quarter as assets grow?

    Buybacks and dividends absorbed about 85% of September-quarter net income while the balance sheet expanded 21% year to date and CET1 fell. The pace of repurchases relative to capital generation determines whether growth is funded from earnings or competes directly with distributions.

Generated by DeepVal AI. Data and models may contain errors or be outdated. For research only; not personalized investment advice.

Go beyond the preview

Continue reading the full report

  1. Key risksWhat could break the investment case
  2. Valuation and margin of safetyWhat the current price assumes, and how much room for error it leaves
  3. SummaryOur overall judgement and what to watch next

Free to sign up · 20 research points · unlocking this report uses 10

20 points cover 2 full reports. No card needed.

Research reflects its generation date. Preview the first two modules.

More on Goldman Sachs Group Inc