0x02 is 2% of active validators and 33% of stake

EIP-8148 says the rate of validators moving to 0x02 compounding credentials has been very low. That is true if a validator means a key: on the complete UTC day of August 21, 0x02 was 17,780 of 901,667 active indices, or 1.97%. Those keys carried 14.05 million of Ethereum's 42.33 million active effective ETH, or 33.19%.

That denominator flip matters because this proposal moved twice in the past few days. An August 20 edit added deposit-time thresholds and lowered the minimum to 32 ETH. Then consensus-specs PR #5557 merged on August 22, setting the default threshold when a validator switches to 0x02 and capping effective balance at the chosen threshold. EIP-8148 is still Draft and only Proposed for Inclusion in Hegotá; none of this is live.

The denominator flips

I resolved the latest credential for every validator first. The withdrawal-credential table keeps historical credential rows, so filtering every row beginning with 0x02 would retain validators after another credential became authoritative. argMax makes the cohort current.

SELECT index
FROM (
    SELECT
        index,
        argMax(
            withdrawal_credentials,
            tuple(epoch, updated_date_time)
        ) AS credential
    FROM default.canonical_beacon_validators_withdrawal_credentials FINAL
    WHERE meta_network_name = 'mainnet'
    GROUP BY index
)
WHERE startsWith(credential, '0x02')
ORDER BY index;

That returned 24,971 indices. I passed them to mainnet.fct_validator_balance_daily FINAL in literal batches of at most 3,000, then kept the active_* rows from the complete August 21 snapshot. This is the actual bounded query shape; a large distributed cross-table subquery was slower and less reliable than fetching the small credential cohort first.

-- Repeated with each literal batch of at most 3,000 indices.
SELECT
    validator_index,
    end_balance,
    effective_balance,
    status
FROM mainnet.fct_validator_balance_daily FINAL
WHERE day_start_date = toDate('2026-08-21')
  AND validator_index IN (<0x02_indices>);

For the network denominator, I used the same table and day. It contained 901,667 active indices and 42,333,847 active effective ETH. The 0x02 cohort had only 17,780 keys but 14,050,074 effective ETH. The exact daily cohort packet hashes to d6be137b...bdddb7; rerunning the current-state path at epoch 470,352 produced the same shape with 17,811 active keys and 14,059,229 effective ETH.

The concentration is not subtle. At epoch 470,352, 6,801 active 0x02 validators had at least 1,024 ETH of effective balance. They were 38.18% of the active 0x02 keys and carried 12.14 million ETH, or 86.35% of the cohort's effective stake. MaxEB packed a lot of old validator weight behind a small number of indices, so "low adoption" changes meaning depending on which unit gets counted.

The 128 ETH example is the small end

EIP-8148 uses a 128 ETH validator as its worked example: let rewards compound to 128 ETH, then sweep future excess automatically. The request rule has an important edge, though. A validator may only set a threshold at or above its current balance; lowering the balance first still requires the partial-withdrawal path the proposal is trying to make less annoying for routine rewards.

I froze epoch 470,352, at 2026-08-22 22:53:11 UTC, and queried the 24,971-index cohort from mainnet.fct_validator_balance FINAL in the same 3,000-index batches:

SELECT
    validator_index,
    balance,
    effective_balance,
    status
FROM mainnet.fct_validator_balance FINAL
WHERE epoch = 470352
  AND validator_index IN (<0x02_indices>);

Of the 17,811 active 0x02 validators, 7,459 had a current balance at or below 128 ETH. That is a healthy-looking 41.88% of the keys, but they carried only 317,324 effective ETH, or 2.26% of active 0x02 stake. The median active balance was 291.25 ETH. The other 10,352 active validators were already above 128 ETH and held 13.74 million effective ETH.

This is not a hidden 128 ETH haircut. A larger validator can choose a larger threshold, and existing 0x02 validators receive the 2,048 ETH default at activation. The newly merged effective-balance cap also creates no immediate loss for the validators currently eligible at 128 ETH: all of them already had effective balance at or below that threshold. The point is narrower and uglier. The proposal's clean 128 ETH example describes 41.88% of today's compounding keys but only 2.26% of their stake.

So yes, 0x02 adoption is low by key count. By the unit Ethereum uses for rewards, consensus weight, churn and finality, it is already one-third of active stake. Both numbers are true. Calling the cohort simply "small" is not.