More Than Half of the EOS Supply Remained Unmapped Before Its Token Freeze

More than half of the planned EOS token supply remained unregistered for the network transition on May 25, 2018, according to an analysis published by blockchain-security company PeckShield one week before the expected end of the year-long token distribution.

PeckShield classified 51.7% of the one-billion-token supply as not registered. That consisted of 49.43% held in circulating Ethereum addresses without a mapped EOS public key and 2.27% still assigned to the continuing crowdsale. The other 48.3% was registered, but that category included the 100 million tokens—10% of the planned supply—reserved for Block.one. On that basis, PeckShield calculated that registered, externally circulating tokens represented 38.3% of the total supply.

Those percentages measured token amounts, not people. They did not establish that 51.7% of token holders had failed to register, nor that the same proportion would ultimately be excluded from the new ledger.

Why registration mattered

The EOS tokens trading in May 2018 were ERC-20 assets recorded on Ethereum. They were not native assets on a functioning EOS blockchain. Block.one’s distribution software provided a registration mechanism through which an Ethereum address could be associated with an EOS public key. A launch group could then use that mapping when constructing the opening ledger of an EOSIO-based chain.

That architecture made the transition materially different from an automatic software upgrade. Holders controlling their own Ethereum addresses needed to complete the mapping process, while customers whose tokens remained at exchanges depended on those intermediaries to register pooled balances and later allocate native assets correctly.

PeckShield separately examined the ten largest token-holding addresses using a May 23 address snapshot. It classified two as EOS-related addresses, six as exchanges and two as individual addresses. Of those ten, it identified only Bitfinex as having registered an EOS public key. The exchange labels relied partly on external address-attribution work, so they should not be treated as independently proven identities for every wallet.

A warning, not a final ledger

The 51.7% figure was a point-in-time result. It combined tokens that were already circulating but unmapped with tokens that had not yet left the distribution contract. Registration could continue after PeckShield collected its May 25 data, and exchanges could map large pooled balances in a small number of transactions. The analysis therefore documented migration readiness on May 25; it was not a forecast of the final snapshot.

Nor did an unmapped balance necessarily mean permanent loss. The precise treatment of unregistered addresses depended on the snapshot tools and launch procedures ultimately selected by the community. What the measurement demonstrated was operational concentration: a limited number of exchange or large-holder actions could materially change the mapped share shortly before the freeze.

The finding also carried governance implications. EOSIO’s proposed delegated proof-of-stake system depended on native-token voting to select block producers. A large amount of supply lacking mapped keys could initially reduce the portion capable of participating directly. That was a risk assessment, not proof that the network would fail to launch or that every unmapped token would remain unable to vote.

Later context

A launch plan published by EOS New York on May 31 later stated that the EOS Mainnet Launch Group had united on May 25 around delivering one chain called EOS. The plan called for verifying the final Ethereum snapshot, booting and testing the candidate network, opening voting, and requiring 15% of all tokens to stake votes before activation.

A September 2019 SEC order subsequently recorded that Block.one’s distribution closed on June 1, 2018, after 900 million tokens were sold and 100 million were reserved. That later regulatory record confirms the distribution and mapping mechanics, but it does not independently reproduce PeckShield’s May 25 percentages.