Skip to content
← Return to articles

Drift Protocol Hack: Following the Ethereum Trail

8 min read
Blueprint diagram: four helper wallets feed a 16-address cluster, three trails end at the edge of the Tornado Cash 100 ETH pool with 735 deposits and 73,500 ETH, and one trail ends at a wallet still holding 56,568 ETH on October 27.

CaudexCatena traced the Ethereum funding wallets, counted deposits into Tornado Cash's 100 ETH pool, and checked the remaining holding wallet.

Data snapshot: October 27, 2026. All dates and times are UTC.

One Ethereum wallet linked to the Drift Protocol hack still held 56,568.38 ETH on October 27, with no outgoing native-ETH transactions recorded. Three other holding wallets had already made deposits into Tornado Cash. Together, those three wallets deposited 73,500 ETH into its 100 ETH pool.

CaudexCatena's investigation connects those movements to a wider Ethereum record: 130,262.54 ETH in recorded inflows, including 64,679.50 ETH traced through four helper wallets. The findings extend the address set worth reviewing and show why a new “ETH” entry in a transaction history needs careful inspection.

The Ethereum Snapshot After the Drift Protocol Hack

Drift Protocol suffered an exploit on April 1, 2026. Its May recovery plan put user losses at approximately $295 million and described roughly 130,259 ETH concentrated in four Ethereum wallets.

Our Ethereum analysis uses a broader grouping of 16 associated addresses. At the October 27 check, CaudexCatena MCP returned these aggregate native-ETH figures:

Measure ETH
Recorded inflows to the 16-address cluster 130,262.54
Recorded outflows from the cluster 73,694.05
Remaining cluster balance 56,568.49
Of that balance, held in the remaining main wallet 56,568.38

The recorded inflow minus outflow equals the remaining balance at the precision shown. These are cluster-level accounting figures. Deposit amounts for the 100 ETH pool are counted separately below.

Waterfall chart of the 16-address cluster's native ETH: 130,262.54 ETH recorded inflow, minus 73,694.05 ETH recorded outflow, leaves a 56,568.49 ETH balance, of which 56,568.38 ETH is in the remaining main wallet.

A cluster is an analytical grouping of addresses. Its membership helps organise the investigation; it does not establish the identity or legal ownership of every address. The transactions connecting particular wallets remain the evidence for those links.

Four Helper Wallets Supplied Almost Half the ETH

The Ethereum trail extends beyond the four wallets that held the funds.

CaudexCatena traced 64,679.50 ETH through four helper wallets outside the 16-address cluster. That is approximately 49.7% of the cluster's recorded inflow. The case records show these wallets receiving bridged USDC, converting it to ETH and forwarding the resulting ETH into the cluster.

Helper Address ETH Traced Into the Cluster
0xad38…bb48 30,416.80
0xef43…80c6 24,970.22
0xec31…80fb 5,967.79
0x3b02…36c0 3,324.69
Total 64,679.50

Two helpers received their initial transaction-fee funding on March 31, before the exploit. The receiving wallet funded 0xef43…80c6. Another address in the cluster funded 0xec31…80fb.

Four helper wallets send ETH into the 16-address cluster along trails sized by amount: 0xad38…bb48 30,416.80 ETH, 0xef43…80c6 24,970.22 ETH and 0xec31…80fb 5,967.79 ETH, both with fee funding on March 31, and 0x3b02…36c0 3,324.69 ETH. Together 64,679.50 ETH, about 49.7% of recorded inflow.

The funding and forwarding transactions support the assessment that the helpers participated in the same operation. Common control remains an inference from those flows, with stronger direct funding evidence for the two March 31 helpers.

For an investigator reviewing where the ETH came from, these addresses matter. Looking only at the holding wallets would leave much of the conversion activity outside the review.

Three Deposit Batches Into Tornado Cash

The case records identify deposit batches on July 23, October 20 and October 21. Each involved a different holding wallet.

Tornado Cash's 100 ETH pool accepts fixed-size deposits. Counting its deposit events gives a clear measure for that pool:

Deposit Date Holding Wallet 100 ETH Deposits ETH Deposited
July 23, 2026 Wallet 1: 0xbdda…561b 230 23,000
October 20, 2026 Wallet 2: 0xaa84…57c1 257 25,700
October 21, 2026 Wallet 3: 0xd3fe…f6c7 248 24,800
Total 735 73,500

The investigation also records deposits into smaller pools. The 73,500 ETH figure covers the 100 ETH pool alone.

The October batches have a direct connection. On October 21, wallet 2 transferred 3.97731448 ETH to wallet 3 before wallet 3's deposits. This was an internal transfer between the holding wallets. Adding it to an external-outflow total would count the same funds again.

Three holding wallets send trails into the Tornado Cash 100 ETH pool: wallet 1 on July 23 with 230 deposits, wallet 2 on October 20 with 257 and wallet 3 on October 21 with 248, totalling 735 deposits and 73,500 ETH. A note marks the internal 3.97731448 ETH handoff from wallet 2 to wallet 3 on October 21.

Drift-Linked Wallets Made 31% of the Pool's Deposits

From July 23 to the October 27 cutoff, the 100 ETH pool recorded 2,361 deposits. The three Drift-linked wallets accounted for 735, or approximately 31.1%.

That percentage describes new deposits during the measured period. It is not the share of the pool's entire balance attributable to Drift. The pool also contained deposits made before July 23.

Bar of 2,361 deposits into the 100 ETH pool from July 23 to October 27, with the first 735 lit as Drift-linked (230, 257 and 248 by date) and 1,626 by others. A note says this is a share of new deposits in the period, not of the pool's whole balance.

This distinction matters when interpreting a mixer. A concentration of deposits can describe activity entering the pool without identifying the recipients of later withdrawals.

Withdrawal Counts Leave the Exit Trail Unresolved

During the same period, the 100 ETH pool recorded 1,772 withdrawals to 962 distinct recipient addresses. Those withdrawals represented 177,200 ETH in pool denominations. The pool balance at the cutoff was 283,400 ETH.

The largest withdrawal count for one recipient address was 21, representing 2,100 ETH in pool denominations. That is a per-address observation within this pool and time window. One party can use several recipient addresses, and funds can move again after withdrawal.

The counts do not establish how many of Drift's deposits were withdrawn, how much remains in the pool, or who received any withdrawn funds. Our analysis does not establish a supported link from a Drift deposit to a particular withdrawal recipient.

A lit trail of 735 Drift-linked deposits ends at the 100 ETH pool, which held 283,400 ETH at the cutoff. On the far side only faint dashed lines fan out to unlit recipient nodes, labelled 1,772 withdrawals and 962 recipient addresses, with the largest per address at 21 withdrawals or 2,100 ETH.

Forged Token Transfers Can Resemble New ETH Movement

Some apparent activity around the holding wallets comes from tokens that imitate ETH.

The case review identified 28 ETH-lookalike tokens displaying outgoing entries against the cluster. In July, forged token events copied the amounts of four genuine small transfers. Between October 21 and 27, eleven events imitated the amount of the 3.97731448 ETH handoff between wallets 2 and 3.

The October 27 example names wallet 2 as the token sender and a lookalike of wallet 3 as the recipient. Its token symbol resembles “ETH”, but the event represents an ERC-20 token. It is not another native-ETH handoff by wallet 2.

Wallet 2 sends one glowing native-ETH trail of 3.97731448 ETH to wallet 3, 0xd3feed…1cf6c7, on October 21. Two dashed rose lines go to lookalike addresses 0xd3febe…46f6c7 and 0xd3fe6b…a0f6c7, which share the same start and end characters. They are labelled as 11 token events named “ETH” between October 21 and 27, beside a count of 28 ETH-lookalike tokens with outgoing entries.

This is address poisoning: misleading history entries use familiar-looking addresses to encourage a mistaken interpretation or copied destination. Before treating a new entry as movement of the stolen ETH, check the asset contract, actual transaction sender, full recipient address and native-ETH movement.

The Remaining Wallet Holds 56,568 ETH

The largest remaining balance sits at 0x0fe3b6908318b1f630daa5b31b49a15fc5f6b674.

On October 27, CaudexCatena MCP returned 56,568.38272929 ETH and no outgoing native-ETH transactions for this address. Its reported valuation was approximately $131.3 million at that check.

A single violet trail ends at wallet 0x0fe3…b674, which holds 56,568.38 ETH with 0 outgoing native-ETH transactions, reported at about 131.3 million US dollars at the October 27 check.
Balances and valuation are an October 27, 2026 snapshot. The other holding wallets' deposit history does not establish where this wallet's funds will move next.

Its first outgoing native-ETH transaction would provide a new lead. The destination and timing remain unknown.

How the Analysis Was Built

The investigation combined address and cluster records from CaudexCatena MCP with Ethereum transaction records and public contract event logs. Address histories, counterparties and token activity were used to review the funding and transfer relationships.

Pool counts cover Ethereum blocks 25,591,709 to 26,283,290, from the start of July 23 through October 27, 2026 at 09:56:59 UTC. Deposit and withdrawal events were counted separately. Wallet and cluster balances were checked on October 27.

The scope is the Ethereum trail. It does not reconstruct the full Solana exploit or make an independent claim about the attackers' identity. AI-generated summaries remain drafts for human review against the underlying records.

For related methodology, see our guide to building a source-of-funds evidence trail.

Selected Transaction Evidence

These references anchor the main links described above. Each deposit reference is one example from its batch; the complete counts use the event-log range in the method section.

Date, UTC Evidence Transaction
March 31 Receiving wallet funds helper 0xef43…80c6 Funding transfer
March 31 Cluster address funds helper 0xec31…80fb Funding transfer
July 23 Wallet 1 deposits 100 ETH Deposit
October 20 Wallet 2 deposits 100 ETH Deposit
October 21 Wallet 2 transfers ETH to wallet 3 Internal handoff
October 21 Wallet 3 deposits 100 ETH Deposit
October 27 ETH-lookalike token event imitates the handoff Token-event example

Drift's recovery team and authorised investigators can contact Caudena to discuss the supporting transaction records.

Category: Investigations

Drift ProtocolEthereumTornado CashTransaction TracingAddress Poisoning
Available today

Need Intelligence Now?

CaudexCatena is not open yet. If you already have wallet, transaction, or risk-review work to run today, CaudexCatena MCP gives paid self-serve access to live blockchain intelligence through supported AI workflows.