Charlie Quant Lab

Institutional map · L0 to L4

Crypto Infrastructure Map

The layer number is the easy part. Shared control is the part that breaks things.

The top three networks hold 84.1% of tracked stablecoin stock, while keys and permissions account for 34.9% of known one-year incident loss.

Stablecoin stock$304.2bnsitting across tracked networks
Top network48.1%Ethereum
Top three networks84.1%share of tracked stock
Next after leader59.6%Tron share after removing Ethereum
General bridges146$51.8bn reported value
Official bridges77$9.9bn reported value
Liquid staking254$52.0bn reported value
Reused-security leader65.3%EigenCloud
One-year incidents3098 have no loss figure
Known loss$2.1bnunknown amounts remain unknown
Keys and permissions34.9%share of known loss
Largest three failures70.3%share of known loss

Five labels. One real question.

Where can one failure stop money, pricing, withdrawals or settlement? The stack below translates the labels without pretending that L3 and L4 have one industry-wide meaning.

L0 · Connection and coordination

Moves messages or security between networks. If the same group of computers approves several networks, one failure can reach all of them.

L1 · Base settlement

The network that records the final state. Stablecoins and apps can cluster here even when users enter elsewhere.

L2 · Cheaper execution

Bundles activity and settles back to a base chain. The official transfer route and the operator that orders transactions can matter more than the layer number.

L3 · Application chains

A network built for one product or group. This label is not used consistently, so compare the actual control path.

L4 · Apps, wallets and access

Where users trade, borrow and sign. It is not a formal blockchain layer, but failures here can still block access.

The contrarian read · Layer labels are not risk scores

Two products on different layers may share the same custodian, bridge, key holder or price source.

The layer number tells you where something sits. It does not tell you who can pause it, upgrade it or restore it.

Capital can gather at different points

Each bar shows the largest provider inside its own group. The groups overlap, so their dollar totals are not added together.

ChainPolygon Bridge · 86.6%
RestakingEigenCloud · 65.3%
CEXBinance CEX · 55.6%
Liquid StakingLido · 46.2%
Canonical BridgeArbitrum Bridge · 32.7%
BridgeWBTC · 17.3%

Polygon Bridge · 86.6% is the highest leader share on this screen. It tells us where a failure could have a large reach. It does not say that a failure is likely.

Where the digital dollars sit

This is the stock of tracked stablecoins on each network. It is not daily trading and it is not money owned by the chain.

Ethereum$146.4bn
Tron$94.1bn
Solana$15.4bn
BSC$13.8bn
Hyperliquid L1$6.8bn
Base$4.9bn
Arbitrum$3.8bn
Polygon$2.9bn
X Layer$1.7bn
Avalanche$1.4bn

Remove Ethereum, and Tron holds 59.6% of what remains. Concentration does not disappear; it changes address.

The failure record points sideways

Known losses do not line up neatly by layer. Keys, permissions, bridges, prices and websites can cut across the full stack.

Known loss · one year

$2.1bn across 301 records with a loss amount.

Keys and permissions

34.9% of known loss came from access-control or private-key failures.

Bridge losses

36.3% of known loss was marked as bridge or cross-chain.

The deepest risk often runs across the diagram, not down one numbered layer.

What actually broke?

These are known loss amounts grouped by the failure named in the incident record. Missing amounts remain missing.

Bridge & Cross-Chain$697.3mn
Token & Share Accounting$389.3mn
Access Control$387.9mn
Key Compromise$343.2mn
Oracle Manipulation$74.0mn
Frontend & Infrastructure$63.8mn
Social Engineering$41.7mn
Governance$33.8mn

The three largest failure types account for 70.3% of known loss. A long list of smaller categories should not hide where most of the money went.

Losses arrive in lumps

Monthly known loss over the last year. A quiet month does not mean the system became safe; one large event can redraw the year.

Oct 25$26.6mn
Nov 25$179.7mn
Dec 25$24.9mn
Jan 26$104.1mn
Feb 26$21.5mn
Mar 26$42.7mn
Apr 26$648.1mn
May 26$83.1mn
Jun 26$88.9mn
Jul 26$254.4mn
Aug 26$255.4mn
Sep 26$326.6mn

The latest quarter contains 40.3% of the one-year known loss. Timing changes the story, so the page keeps the full twelve-month strip visible.

The latest recorded breaks

Recent events are shown with the stated loss. “Not reported” is kept separate from zero.

Swipe the table, or focus it and use the arrow keys.

DateProjectWhat brokeNetworkKnown loss
16 Sep 2026StartaleAccess ControlEthereum$2,876
12 Sep 2026Chainflip AMMBridge & Cross-ChainTron$736,442
11 Sep 2026ether.fi LiquidAccess ControlEthereum$43,260
10 Sep 2026DominionKey CompromiseSolanaNot reported
10 Sep 2026SymbiosisBridge & Cross-ChainBSC, Ethereum$336,000
09 Sep 2026NomicBridge & Cross-ChainNomic$3.1mn
09 Sep 2026Zentra FinanceToken & Share AccountingCitrea$140,030
09 Sep 2026BeatXswapOracle ManipulationBSC$77,512
08 Sep 2026WealthManagementV2Key CompromiseBSC$26,414
07 Sep 2026Cozy V2Protocol LogicOptimism$163,326

Only 5 records in the one-year window state returned funds. The known returned amount is $1.9mn, or 0.1% of known loss. That is incomplete reporting, not a recovery rate for the whole market.

Research by for Charlie Quant Lab · Updated