Wikipedia - List of Largest Banks

Three small, stable tables rather than one sprawling dump. The primary table ranks the world's 100 largest banks by total assets in US$ billions, reproducing S&P Global Market Intelligence's annual report on the world's 100 largest banks - the current label reads April 2026. A second table counts how many of those top-100 banks are headquartered in each of 24 countries and territories. A third, shorter table lists the top 10 banks by market capitalization, captured at a labeled snapshot date rather than a fixed publication cycle. 134 rows in all, six fields wide.

Two things separate this from a generic community list. First, every figure is attributed to a named third party - S&P Global Market Intelligence for assets, a market-data provider for the market-cap snapshot - and the article states its methodology outright: rankings use assets as reported, with no adjustment for differing accounting treatments, so US GAAP banks net their derivative positions while many non-US peers do not. Second, the shape: because every row carries a bank name, a country and a figure, the tables behave like reference data. They are the fastest legitimate answer to 'which banks dominate globally, and by what margin' - ICBC at 7,645.80, Agricultural Bank of China at 6,974.82, China Construction Bank at 6,524.05 US$ billion, down to Qatar National Bank at rank 100 on 381.61.

Within Datadory's diversified-banks shelf this is the only dataset that hands you a finished worldwide league table. No regulator publishes one, because every regulator stops at its own border. Get a sample of this dataset and the rows arrive typed, decoded and joined-ready instead of copied by hand.

Five real rows spanning the full depth of the table, exactly as they arrive:

# total assets, April 2026 edition - ranks 1-2-3 of 100
rank        : 1
bank name   : Industrial and Commercial Bank of China
country     : China
total assets: 7,645.80   (US$ billion)

rank        : 2
bank name   : Agricultural Bank of China
country     : China
total assets: 6,974.82

rank        : 3
bank name   : China Construction Bank
country     : China
total assets: 6,524.05

# ... rank 5 JPMorgan Chase (United States) 4,424.90
# ... rank 100 Qatar National Bank (Qatar)     381.61

Read what the rows settle immediately. The gap from rank 1 to rank 100 is roughly twentyfold - a cut at rank 25 tells a very different story from a cut at rank 75, so band definitions matter more than the exact boundary. China holds the top three slots outright, but the United States places the largest bank outside Asia at rank 5. And the bottom of the table is not filler: rank 100 means US$381.61 billion in assets, larger than most national banking systems entire.

Six fields carry the entire payload, verified against the tables during the research pass - nothing inferred from headers alone. The smallness is the feature: a schema you can hold in your head is also a schema no analyst misjoins.

Deliveries can extend the core six in useful directions: normalized bank-name variants resolved to a canonical identifier, ISO country codes beside the plain-text country column, rank-band groupings, or the same rows joined to regulator financials keyed on the matched institutions. These fold under additional fields on request - they are specified and populated when you ask for a sample.

Geography - global by construction. The 100 ranked banks sit across 24 countries and territories, and the companion table makes the distribution explicit: China holds 21 of the top 100 slots, the United States 12, Japan 8. Regional rollups are a group-by on the country column, not a merge.

Temporal - the assets table moves annually with S&P Global Market Intelligence's new edition, currently labeled April 2026. The market-cap leaderboard floats on its own snapshot date rather than following that cycle, so the two tables inside one article can disagree about 'now' by design.

Granularity - three tiers that never mix: one row per bank in the assets ranking, one row per country in the count table, one row per bank in the market-cap leaderboard. No aggregates, no sampling, no weighting.

Set against the wider catalog - average quality score 7.81 across all 1,744 datasets - this slice scores 7/10: verified field definitions and five shipped sample rows, marked down honestly for being an annual snapshot of 100 banks rather than a live panel. Depth is not its job; the spine is.

API, files, or your warehouse. Daily, weekly, or hourly.

• Entity spine for global banking joins - every row carries bank name, country and figure, so the table anchors name-to-country resolution before filings, regulator registries or market data attach. Match on the name-plus-country pair and the fuzzy-matching project shrinks to a lookup.
• Peer-set construction - rank bands cut the world's banks into defensible groups - top 10, 11-50, 51-100 - without arguing over inclusion criteria, which is half of what competitive briefs fight about.
• Country concentration analysis - the 24-row table quantifies how concentrated global banking is by headquarters country: China alone holds 21 of the top 100 slots, more than the United States and Japan combined.
• Assets-versus-market-value contrast - the same article carries an asset ranking and a market-cap leaderboard whose orderings disagree, the cleanest demonstration available that scale is not value.
• Deck-ready league tables - small, sortable and widely recognized, so consultants, sales teams and journalists quote it directly instead of assembling a ranking from a hundred filings.

Data scientists and ML engineers seed entity lists and label sets for bank-matching models from rows that resolve cleanly to name-plus-country keys. Investors and quants sanity-check internal league tables against the published global ordering before trusting a home-grown screen. Competitive intel and product teams track which peers sit in which rank band between editions. Market researchers and consultants lift per-country counts straight into sizing decks. Sales and growth teams start enterprise target lists with asset-rank context already attached. Journalists, academics and students cite a ranking everyone recognizes, then audit any single figure against the regulator datasets keyed to the same institutions. Persona workflows live at data scientists x diversified banks, competitive intel product teams x diversified banks and investors & quants x diversified banks.

Against the rest of the shelf, this table competes on speed and scope, not depth. FDIC BankFind Suite API covers roughly 4,700 US insured institutions record-by-record but sees nothing outside the United States; BIS Locational Banking Statistics publish roughly 609,000 series on cross-border positions yet never name an individual bank; SEC EDGAR XBRL resolves listed issuers one concept at a time. None of those hands you a ready-made global top-100 league table - this one does, and asks for depth elsewhere in return.

Its natural partner is the OCC Financial Institution Lists: Wikipedia supplies the global spine of 100 banks across 24 countries, the OCC verifies the US charters among them, and FDIC financials attach the balance sheets. That three-step join - rank, charter, financials - is the standard route from 'list me the biggest banks' to 'show me their numbers'.

Four honest caveats. First, accounting basis: rankings use assets as reported, unadjusted for differing treatments - US GAAP nets derivative positions, so cross-border comparisons near the top flatter and penalize different regimes. Second, dating: the market-cap table carries its own snapshot stamp whose day-month versus month-day ordering is ambiguous, so pin the date before citing those ten rows. Third, stability: the country column should be confirmed to render consistently across revisions before it becomes a production join key - our research pass flags exactly that check. Fourth, cadence: this is an annual-edition snapshot, not a live feed, so a bank's rank is a point-in-time claim while its balance sheet keeps moving. Scheduled captures turn successive editions into the panel a single pull cannot be.

Because the raw artifact is a wiki page built for readers, and most questions want typed rows. Datadory delivers the three tables as clean columns with definitions and examples attached, normalizes bank names toward a canonical identifier so joins hold, and keeps successive editions accumulating on a schedule so rank movement becomes observable history. For coverage beyond the top 100, the article itself points to The Banker's 'World 1000 Largest Banks' ranking, published every July - ask alongside your sample and we will scope it. Browse the rest of the industry on the diversified banks data hub, the best diversified-banks datasets ranking, or the full catalog.